Merge pull request #69 from LNIS-Projects/dev

Bug fix in yosys-vpr flow using OpenFPGA shell;
2020-07-22 14:39:09 -06:00 · 2020-07-22 14:39:09 -06:00 · 734b8ea1d0
parent ffc087336d 8ade40713a
commit 734b8ea1d0
67 changed files with 2851 additions and 270 deletions
--- a/docs/source/manual/openfpga_shell/openfpga_commands/fpga_verilog_commands.rst
+++ b/docs/source/manual/openfpga_shell/openfpga_commands/fpga_verilog_commands.rst
@ -31,7 +31,7 @@ write_verilog_testbench

  - ``--reference_benchmark_file_path`` Must specify the reference benchmark Verilog file if you want to output any testbenches

-  - ``--fast_configuration`` Enable fast configuration phase for the top-level testbench in order to reduce runtime of simulations. It is applicable to memory bank and frame-based configuration protocols. When enabled, all the zero configuration bits will be skipped. So ensure that your memory cells can be correctly reset to zero with a reset signal. 
+  - ``--fast_configuration`` Enable fast configuration phase for the top-level testbench in order to reduce runtime of simulations. It is applicable to configuration chain, memory bank and frame-based configuration protocols. For configuration chain, when enabled, the zeros at the head of the bitstream will be skipped. For memory bank and frame-based, when enabled, all the zero configuration bits will be skipped. So ensure that your memory cells can be correctly reset to zero with a reset signal. 

  - ``--print_top_testbench`` Enable top-level testbench which is a full verification including programming circuit and core logic of FPGA

--- a/openfpga/src/base/openfpga_spice.cpp
+++ b/openfpga/src/base/openfpga_spice.cpp
@ -39,7 +39,7 @@ int write_fabric_spice(OpenfpgaContext& openfpga_ctx,
  int status = CMD_EXEC_SUCCESS;
  status = fpga_fabric_spice(openfpga_ctx.module_graph(),
                             openfpga_ctx.mutable_spice_netlists(),
-                             openfpga_ctx.arch().tech_lib,
+                             openfpga_ctx.arch(),
                             options);

  return status;
--- a/openfpga/src/fpga_spice/spice_api.cpp
+++ b/openfpga/src/fpga_spice/spice_api.cpp
@ -39,7 +39,7 @@ namespace openfpga {
 ********************************************************************/
 int fpga_fabric_spice(const ModuleManager& module_manager,
                      NetlistManager& netlist_manager,
-                      const TechnologyLibrary& tech_lib,
+                      const Arch& openfpga_arch,
                      const FabricSpiceOption& options) {

  vtr::ScopedStartFinishTimer timer("Write SPICE netlists for FPGA fabric\n");
@ -71,7 +71,8 @@ int fpga_fabric_spice(const ModuleManager& module_manager,
  int status = CMD_EXEC_SUCCESS;

  status = print_spice_submodule(netlist_manager,
-                                 tech_lib,
+                                 module_manager,
+                                 openfpga_arch,
                                 submodule_dir_path);
 
  if (CMD_EXEC_SUCCESS != status) {
--- a/openfpga/src/fpga_spice/spice_api.h
+++ b/openfpga/src/fpga_spice/spice_api.h
@ -9,7 +9,7 @@
 #include <vector>
 #include "netlist_manager.h"
 #include "module_manager.h"
-#include "technology_library.h"
+#include "openfpga_arch.h"
 #include "fabric_spice_options.h"

 /********************************************************************
@ -21,7 +21,7 @@ namespace openfpga {

 int fpga_fabric_spice(const ModuleManager& module_manager,
                      NetlistManager& netlist_manager,
-                      const TechnologyLibrary& tech_lib,
+                      const Arch& openfpga_arch,
                      const FabricSpiceOption& options);

 } /* end namespace openfpga */
--- a/openfpga/src/fpga_spice/spice_essential_gates.cpp
+++ b/openfpga/src/fpga_spice/spice_essential_gates.cpp
@ -18,6 +18,7 @@
 #include "openfpga_digest.h"

 #include "spice_constants.h"
+#include "spice_writer_utils.h"
 #include "spice_essential_gates.h"

 /* begin namespace openfpga */
@ -82,7 +83,7 @@ int print_spice_transistor_wrapper(NetlistManager& netlist_manager,
  check_file_stream(spice_fname.c_str(), fp); 

  /* Create file */
-  VTR_LOG("Generating SPICE netlist '%s' for transistor wrappers...",
+  VTR_LOG("Generating SPICE netlist '%s' for essential gates...",
          spice_fname.c_str()); 

  /* Iterate over the transistor models */
@ -110,4 +111,165 @@ int print_spice_transistor_wrapper(NetlistManager& netlist_manager,
  return CMD_EXEC_SUCCESS;
 }

+/************************************************
+ * Generate the SPICE subckt for an inverter
+ * Schematic
+ *          LVDD
+ *            |
+ *           -
+ *      +-o||
+ *      |    -
+ *      |     |
+ * in-->+     +--> OUT
+ *      |     |
+ *      |    -
+ *      +--||
+ *           -
+ *            |
+ *          LGND
+ *
+ ***********************************************/
+static 
+int print_spice_inverter_subckt(std::fstream& fp,
+                                const ModuleManager& module_manager,
+                                const ModuleId& module_id,
+                                const CircuitLibrary& circuit_lib,
+                                const CircuitModelId& circuit_model,
+                                const TechnologyLibrary& tech_lib,
+                                const TechnologyModelId& tech_model) {
+  if (false == valid_file_stream(fp)) {
+    return CMD_EXEC_FATAL_ERROR;
+  }
+
+  /* Print the inverter subckt definition */
+  print_spice_subckt_definition(fp, module_manager, module_id); 
+
+  /* Find the input and output ports:
+   * we do NOT support global ports here, 
+   * it should be handled in another type of inverter subckt (power-gated)
+   */
+  std::vector<CircuitPortId> input_ports = circuit_lib.model_ports_by_type(circuit_model, CIRCUIT_MODEL_PORT_INPUT, true);
+  std::vector<CircuitPortId> output_ports = circuit_lib.model_ports_by_type(circuit_model, CIRCUIT_MODEL_PORT_OUTPUT, true);
+
+  /* Make sure:
+   * There is only 1 input port and 1 output port, 
+   * each size of which is 1
+   */
+  VTR_ASSERT( (1 == input_ports.size()) && (1 == circuit_lib.port_size(input_ports[0])) );
+  VTR_ASSERT( (1 == output_ports.size()) && (1 == circuit_lib.port_size(output_ports[0])) );
+
+  /* TODO: may consider use size/bin to compact layout etc. */
+  for (size_t i = 0; i < circuit_lib.buffer_size(circuit_model); ++i) { 
+    /* Write transistor pairs using the technology model */
+    fp << "Xpmos_" << i << " ";
+    fp << circuit_lib.port_prefix(output_ports[0]) << " "; 
+    fp << circuit_lib.port_prefix(input_ports[0]) << " "; 
+    fp << "LVDD "; 
+    fp << "LVDD "; 
+    fp << tech_lib.transistor_model_name(tech_model, TECH_LIB_TRANSISTOR_PMOS) << TRANSISTOR_WRAPPER_POSTFIX; 
+
+    fp << "Xnmos_" << i << " ";
+    fp << circuit_lib.port_prefix(output_ports[0]) << " "; 
+    fp << circuit_lib.port_prefix(input_ports[0]) << " "; 
+    fp << "LGND "; 
+    fp << "LGND "; 
+    fp << tech_lib.transistor_model_name(tech_model, TECH_LIB_TRANSISTOR_NMOS) << TRANSISTOR_WRAPPER_POSTFIX; 
+  }
+
+  print_spice_subckt_end(fp, module_manager.module_name(module_id)); 
+
+  return CMD_EXEC_SUCCESS;
+}
+
+/************************************************
+ * Generate the SPICE netlist for essential gates:
+ * - inverters and their templates
+ * - buffers and their templates
+ * - pass-transistor or transmission gates
+ * - logic gates
+ ***********************************************/
+int print_spice_essential_gates(NetlistManager& netlist_manager,
+                                const ModuleManager& module_manager,
+                                const CircuitLibrary& circuit_lib,
+                                const TechnologyLibrary& tech_lib,
+                                const std::map<CircuitModelId, TechnologyModelId>& circuit_tech_binding,
+                                const std::string& submodule_dir) {
+  std::string spice_fname = submodule_dir + std::string(ESSENTIALS_SPICE_FILE_NAME);
+
+  std::fstream fp;
+
+  /* Create the file stream */
+  fp.open(spice_fname, std::fstream::out | std::fstream::trunc);
+  /* Check if the file stream if valid or not */
+  check_file_stream(spice_fname.c_str(), fp); 
+
+  /* Create file */
+  VTR_LOG("Generating SPICE netlist '%s' for transistor wrappers...",
+          spice_fname.c_str()); 
+
+  int status = CMD_EXEC_SUCCESS;
+
+  /* Iterate over the circuit models */
+  for (const CircuitModelId& circuit_model : circuit_lib.models()) {
+    /* Bypass models require extern netlists */
+    if (true == circuit_lib.model_circuit_netlist(circuit_model).empty()) {
+      continue;
+    }
+
+    /* Spot module id */
+    const ModuleId& module_id = module_manager.find_module(circuit_lib.model_name(circuit_model));
+
+    TechnologyModelId tech_model; 
+    /* Focus on inverter/buffer/pass-gate/logic gates only */
+    if ( (CIRCUIT_MODEL_INVBUF == circuit_lib.model_type(circuit_model))
+      || (CIRCUIT_MODEL_PASSGATE == circuit_lib.model_type(circuit_model))
+      || (CIRCUIT_MODEL_GATE == circuit_lib.model_type(circuit_model))) {
+      auto result = circuit_tech_binding.find(circuit_model);
+      if (result == circuit_tech_binding.end()) {
+        VTR_LOGF_ERROR(__FILE__, __LINE__,
+                       "Unable to find technology binding for circuit model '%s'!\n",
+                       circuit_lib.model_name(circuit_model).c_str()); 
+        return CMD_EXEC_FATAL_ERROR;
+      }
+      /* Valid technology binding. Assign techology model */
+      tech_model = result->second;
+      /* Ensure we have a valid technology model */
+      VTR_ASSERT(true == tech_lib.valid_model_id(tech_model));
+      VTR_ASSERT(TECH_LIB_MODEL_TRANSISTOR == tech_lib.model_type(tech_model));
+    }
+
+    /* Now branch on netlist writing */
+    if (CIRCUIT_MODEL_INVBUF == circuit_lib.model_type(circuit_model)) {
+      if (CIRCUIT_MODEL_BUF_INV == circuit_lib.buffer_type(circuit_model)) {
+        VTR_ASSERT(true == module_manager.valid_module_id(module_id));
+        status = print_spice_inverter_subckt(fp,
+                                             module_manager, module_id,
+                                             circuit_lib, circuit_model,
+                                             tech_lib, tech_model);
+      } else {
+        VTR_ASSERT(CIRCUIT_MODEL_BUF_BUF == circuit_lib.buffer_type(circuit_model));
+      }
+
+      if (CMD_EXEC_FATAL_ERROR == status) {
+        break;
+      }
+
+      /* Finish, go to the next */
+      continue;
+    }
+  } 
+
+  /* Close file handler*/
+  fp.close();
+
+  /* Add fname to the netlist name list */
+  NetlistId nlist_id = netlist_manager.add_netlist(spice_fname);
+  VTR_ASSERT(NetlistId::INVALID() != nlist_id);
+  netlist_manager.set_netlist_type(nlist_id, NetlistManager::SUBMODULE_NETLIST);
+
+  VTR_LOG("Done\n");
+
+  return status;
+}
+
 } /* end namespace openfpga */
--- a/openfpga/src/fpga_spice/spice_essential_gates.h
+++ b/openfpga/src/fpga_spice/spice_essential_gates.h
@ -5,7 +5,10 @@
 * Include header files that are required by function declaration
 *******************************************************************/
 #include <string>
+#include <map>
 #include "netlist_manager.h"
+#include "module_manager.h"
+#include "circuit_library.h"
 #include "technology_library.h"

 /********************************************************************
@ -19,6 +22,13 @@ int print_spice_transistor_wrapper(NetlistManager& netlist_manager,
                                   const TechnologyLibrary& tech_lib,
                                   const std::string& submodule_dir);

+int print_spice_essential_gates(NetlistManager& netlist_manager,
+                                const ModuleManager& module_manager,
+                                const CircuitLibrary& circuit_lib,
+                                const TechnologyLibrary& tech_lib,
+                                const std::map<CircuitModelId, TechnologyModelId>& circuit_tech_binding,
+                                const std::string& submodule_dir);
+
 } /* end namespace openfpga */

 #endif
--- a/openfpga/src/fpga_spice/spice_submodule.cpp
+++ b/openfpga/src/fpga_spice/spice_submodule.cpp
@ -28,13 +28,21 @@ namespace openfpga {
 * 6. TODO: Configuration memory blocks
 ********************************************************************/
 int print_spice_submodule(NetlistManager& netlist_manager,
-                          const TechnologyLibrary& tech_lib,
+                          const ModuleManager& module_manager,
+                          const Arch& openfpga_arch,
                          const std::string& submodule_dir) {

  int status = CMD_EXEC_SUCCESS;

  status = print_spice_transistor_wrapper(netlist_manager,
-                                          tech_lib,
+                                          openfpga_arch.tech_lib,
+                                          submodule_dir);
+
+  status = print_spice_essential_gates(netlist_manager,
+                                       module_manager,
+                                       openfpga_arch.circuit_lib,
+                                       openfpga_arch.tech_lib,
+                                       openfpga_arch.circuit_tech_binding,
                                       submodule_dir);

  return status;
--- a/openfpga/src/fpga_spice/spice_submodule.h
+++ b/openfpga/src/fpga_spice/spice_submodule.h
@ -5,7 +5,8 @@
 * Include header files that are required by function declaration
 *******************************************************************/
 #include "netlist_manager.h"
-#include "technology_library.h"
+#include "module_manager.h"
+#include "openfpga_arch.h"

 /********************************************************************
 * Function declaration
@ -15,7 +16,8 @@
 namespace openfpga {

 int print_spice_submodule(NetlistManager& netlist_manager,
-                          const TechnologyLibrary& tech_lib,
+                          const ModuleManager& module_manager,
+                          const Arch& openfpga_arch,
                          const std::string& submodule_dir);

 } /* end namespace openfpga */
--- a/openfpga/src/fpga_verilog/verilog_top_testbench.cpp
+++ b/openfpga/src/fpga_verilog/verilog_top_testbench.cpp
@ -1154,17 +1154,18 @@ void print_verilog_top_testbench_configuration_chain_bitstream(std::fstream& fp,
  /* Attention: when the fast configuration is enabled, we will start from the first bit '1'
   * This requires a reset signal (as we forced in the first clock cycle)
   */
-  bool first_bit_one = false;
+  bool start_config = false;
  for (const FabricBitId& bit_id : fabric_bitstream.bits()) {
-    if (true == bitstream_manager.bit_value(fabric_bitstream.config_bit(bit_id))) {
-      first_bit_one = true;
+    if ( (false == start_config)
+      && (true == bitstream_manager.bit_value(fabric_bitstream.config_bit(bit_id)))) {
+      start_config = true;
    } 

    /* In fast configuration mode, we do not output anything
     * until we have to (the first bit '1' detected)
     */
    if ( (true == fast_configuration)
-      && (false == first_bit_one)) {
+      && (false == start_config)) {
      continue;
    }

--- a/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_hour.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_hour.v
@ -0,0 +1,81 @@
+module FSM_hour(
+	input wire rst,
+	input wire clk,
+	input wire [5:0] hour_in,
+	input wire hour_in_load,
+	input wire [5:0] min_count,
+	input wire [5:0] sec_count,
+	output reg [5:0] hour_out);
+	
+	reg  [2:0] ps, ns;
+	wire [5:0] hour_data_add;
+	reg  [5:0] hour_data;
+	reg  [5:0] hour_ps, hour_ns;
+	reg  [1:0] hour_sel;
+	wire hour_count;
+	
+	always@(posedge clk)
+	begin
+		if(rst) ps <= 3'd0;
+		else ps <= ns;
+	end
+	
+	always@(posedge clk)
+	begin
+		if(rst) hour_ps <= 6'd0;
+		else hour_ps <= hour_ns;
+	end
+	
+	always@(*)
+	begin
+		hour_sel = 2'd0;
+		case(ps)
+			3'd0: begin
+				ns = 3'd1;
+			end
+			3'd1: begin
+				if(hour_in_load) begin
+					hour_sel = 2'd1;
+					hour_out = hour_data;
+					ns = 3'd2;
+					hour_ns = hour_data;
+				end
+				else ns = 3'd1;
+			end
+			3'd2: begin
+				if(hour_count == 1'd1) begin
+					if(hour_data == 6'd59) begin
+						hour_out = hour_data;
+						ns = 3'd2;
+						hour_ns = 6'd0;
+					end
+					else begin
+						hour_out = hour_data;
+						ns = 3'd2;
+						hour_ns = hour_data_add;
+					end
+				end
+				else begin
+					hour_out = hour_data;
+					hour_ns = hour_data;
+					ns = 3'd2;
+				end
+			end
+			default: begin
+				ns = 3'd0;
+			end
+		endcase
+	end
+	
+	assign hour_data_add = hour_data + 1;
+	assign hour_count = ((sec_count == 6'd59)&&(min_count == 6'd59)) ? 1'd1 : 1'd0;
+	
+	always@(*)
+	begin
+		case(hour_sel)
+			2'd0: hour_data = hour_ps;
+			2'd1: hour_data = hour_in;
+		endcase
+	end
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_minute.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_minute.v
@ -0,0 +1,80 @@
+module FSM_minute(
+	input wire rst,
+	input wire clk,
+	input wire [5:0] min_in,
+	input wire min_in_load,
+	input wire [5:0] sec_count,
+	output reg [5:0] min_out);
+	
+	reg  [2:0] ps, ns;
+	wire [5:0] min_data_add;
+	reg  [5:0] min_data;
+	reg  [5:0] min_ps, min_ns;
+	reg  [1:0] min_sel;
+	wire min_count;
+	
+	always@(posedge clk)
+	begin
+		if(rst) ps <= 3'd0;
+		else ps <= ns;
+	end
+	
+	always@(posedge clk)
+	begin
+		if(rst) min_ps <= 6'd0;
+		else min_ps <= min_ns;
+	end
+	
+	always@(*)
+	begin
+		min_sel = 2'd0;
+		case(ps)
+			3'd0: begin
+				ns = 3'd1;
+			end
+			3'd1: begin
+				if(min_in_load) begin
+					min_sel = 2'd1;
+					min_out = min_data;
+					ns = 3'd2;
+					min_ns = min_data;
+				end
+				else ns = 3'd1;
+			end
+			3'd2: begin
+				if(min_count == 1'd1) begin
+					if(min_data == 6'd59) begin
+						min_out = min_data;
+						ns = 3'd2;
+						min_ns = 6'd0;
+					end
+					else begin
+						min_out = min_data;
+						ns = 3'd2;
+						min_ns = min_data_add;
+					end
+				end
+				else begin
+					min_out = min_data;
+					min_ns = min_data;
+					ns = 3'd2;
+				end
+			end
+			default: begin
+				ns = 3'd0;
+			end
+		endcase
+	end
+	
+	assign min_data_add = min_data + 1;
+	assign min_count = (sec_count == 6'd59) ? 1'd1 : 1'd0;
+	
+	always@(*)
+	begin
+		case(min_sel)
+			2'd0: min_data = min_ps;
+			2'd1: min_data = min_in;
+		endcase
+	end
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_second.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_second.v
@ -0,0 +1,70 @@
+module FSM_second(
+	input wire rst,
+	input wire clk,
+	input wire [5:0] sec_in,
+	input wire sec_in_load,
+	output reg [5:0] sec_out);
+	
+	reg  [2:0] ps, ns;
+	wire [5:0] sec_data_add;
+	reg  [5:0] sec_data;
+	reg  [5:0] sec_ps, sec_ns;
+	reg  [1:0] sec_sel;
+	
+	always@(posedge clk)
+	begin
+		if(rst) ps <= 3'd0;
+		else ps <= ns;
+	end
+	
+	always@(posedge clk)
+	begin
+		if(rst) sec_ps <= 6'd0;
+		else sec_ps <= sec_ns;
+	end
+	
+	always@(*)
+	begin
+		sec_sel = 2'd0;
+		case(ps)
+			3'd0: begin
+				ns = 3'd1;
+			end
+			3'd1: begin
+				if(sec_in_load) begin
+					sec_sel = 2'd1;
+					sec_out = sec_data;
+					ns = 3'd2;
+					sec_ns = sec_data_add;
+				end
+				else ns = 3'd1;
+			end
+			3'd2: begin
+				if(sec_data == 6'd59) begin
+					sec_out = sec_data;
+					ns = 3'd2;
+					sec_ns = 6'd0;
+				end
+				else begin
+					sec_out = sec_data;
+					ns = 3'd2;
+					sec_ns = sec_data_add;
+				end
+			end
+			default: begin
+				ns = 3'd0;
+			end
+		endcase
+	end
+	
+	assign sec_data_add = sec_data + 1;
+	
+	always@(*)
+	begin
+		case(sec_sel)
+			2'd0: sec_data = sec_ps;
+			2'd1: sec_data = sec_in;
+		endcase
+	end
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_three_tb.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_three_tb.v
@ -0,0 +1,34 @@
+module FSM_three_tb;
+
+	reg rst;
+	reg clk;
+	reg [5:0] sec_in, min_in, hour_in;
+	reg load_in;
+	wire [5:0] sec_out, min_out, hour_out;
+
+	FSM_top FSM_1(
+		.rst(rst),
+		.clk(clk),
+		.sec_in(sec_in),
+		.load_in(load_in),
+		.sec_out(sec_out),
+		.min_in(min_in),
+		.min_out(min_out),
+		.hour_in(hour_in),
+		.hour_out(hour_out));
+		
+	initial begin
+		#0 rst = 1'd1; clk = 1'd0; load_in = 1'd1; sec_in = 6'd33; min_in = 6'd14; hour_in = 6'd5;
+		#100 rst = 1'd0;
+		#50 load_in = 1'd0;
+	end
+	
+	always begin
+		#10 clk = ~clk;
+	end
+	
+	initial begin
+		#100000 $stop;
+	end
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_top.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/FSM_three_code/FSM_top.v
@ -0,0 +1,37 @@
+module FSM_top(
+	input wire rst,
+	input wire clk,
+	input wire load_in,
+	input wire [5:0] sec_in,
+	input wire [5:0] min_in,
+	input wire [5:0] hour_in,
+	output wire [5:0] sec_out,
+	output wire [5:0] min_out,
+	output wire [5:0] hour_out
+	);
+
+	FSM_second FSM_sec(
+		.rst(rst),
+		.clk(clk),
+		.sec_in(sec_in),
+		.sec_in_load(load_in),
+		.sec_out(sec_out));
+
+	FSM_minute FSM_min(
+		.rst(rst),
+		.clk(clk),
+		.min_in(min_in),
+		.min_in_load(load_in),
+		.sec_count(sec_out),
+		.min_out(min_out));
+		
+	FSM_hour FSM_hr(
+		.rst(rst),
+		.clk(clk),
+		.hour_in(hour_in),
+		.hour_in_load(load_in),
+		.min_count(min_out),
+		.hour_out(hour_out),
+		.sec_count(sec_out));
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/ALU.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/ALU.v
@ -0,0 +1,27 @@
+module ALU(zero_flag_out,alu_out,Reg_Y_in,Bus_1_in,IR_code);
+
+output	zero_flag_out;
+output	reg	[7:0]alu_out;
+input	[7:0]Reg_Y_in,Bus_1_in;
+input	[7:0]IR_code;
+
+wire	[3:0]opcode=IR_code[7:4];
+
+
+
+always@(*)
+	begin
+		case(opcode)
+		1:	alu_out=Reg_Y_in+Bus_1_in;
+		2:	alu_out=Bus_1_in+~(Reg_Y_in)+1;
+		3:	alu_out=Reg_Y_in&(Bus_1_in);
+		4:  alu_out=~(Bus_1_in);
+		default:alu_out=8'b0;
+		endcase
+	end
+
+assign zero_flag_out=~|alu_out;
+
+endmodule
+
+	
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Controller.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Controller.v
@ -0,0 +1,211 @@
+module Controller(L_R0,L_R1,L_R2,L_R3,L_PC,Inc_PC,
+Sel_Bus1,L_IR,L_ADD_R,L_R_Y,L_R_Z,Sel_Bus2,write,
+zero,instruction,nclk,rst);
+
+
+//狀態
+parameter	S_idle=0,S_fet1=1,S_fet2=2,S_dec=3,
+			S_ex1=4,S_rd1=5,S_rd2=6,S_wr1=7,S_wr2=8,
+			S_br1=9,S_br2=10,S_halt=11;
+//指令
+parameter	NOP=0,ADD=1,SUB=2,AND=3,NOT=4,
+			RD=5,WR=6,BR=7,BRZ=8;
+
+
+output reg	L_R0,L_R1,L_R2,L_R3,L_PC,Inc_PC,
+			L_IR,L_ADD_R,L_R_Y,L_R_Z,write;
+output	reg[2:0]Sel_Bus1;
+output	reg [1:0]Sel_Bus2;
+
+input	zero,nclk,rst;
+input	[7:0]instruction;
+
+reg	[15:0]Con_out;
+reg	[3:0]PS,NS;
+reg	err_flag;
+
+wire [1:0]src=instruction[3:2];
+wire [1:0]dest=instruction[1:0];
+wire [3:0]opcode=instruction[7:4];
+
+always@(posedge nclk)
+	begin
+		if(rst==1)PS<=0;
+		else PS<=NS;
+	end
+
+always@(PS,opcode,src,dest,zero)
+	begin
+	L_R0=0;
+	L_R1=0;
+	L_R2=0;
+	L_R3=0;
+	L_PC=0;
+	Inc_PC=0;
+	Sel_Bus1=0;
+	L_IR=0;
+	L_ADD_R=0;
+	L_R_Y=0;
+	L_R_Z=0;
+	Sel_Bus2=0;
+	write=0;
+	err_flag=0;
+		case(PS)
+			S_idle:		NS=S_fet1;
+			
+			S_fet1:		begin
+						 NS=S_fet2;
+						 Sel_Bus1=3'b100;//Sel_PC
+						 Sel_Bus2=2'b01;//Sel_Bus1	
+						 L_ADD_R=1;
+						end
+						
+			S_fet2:		begin
+						 NS=S_dec;
+						 Sel_Bus2=2'b10;//Sel_Mem
+						 L_IR=1;
+						 Inc_PC=1;
+						end
+						
+			S_dec:		begin
+						case(opcode)
+							NOP:NS=S_fet1;							
+							ADD,SUB,AND:begin
+								NS=S_ex1;
+								Sel_Bus2=2'b01;//Sel_Bus1
+								L_R_Y=1;
+								case(src)
+									0:		Sel_Bus1=3'b000;//R0
+									1:		Sel_Bus1=3'b001;//R1
+									2:		Sel_Bus1=3'b010;//R2
+									3:		Sel_Bus1=3'b011;//R3
+									default	err_flag=1;
+								endcase
+								end//ADD,SUB,AND
+								
+							NOT:begin
+								NS=S_fet1;
+								L_R_Z=1;
+								Sel_Bus2=2'b00;//Sel_ALU
+								case(src)
+									0:		Sel_Bus1=3'b000;//R0
+									1:		Sel_Bus1=3'b001;//R1
+									2:		Sel_Bus1=3'b010;//R2
+									3:		Sel_Bus1=3'b011;//R3
+									default	err_flag=1;
+								endcase
+								case(dest)
+									0:		L_R0=1;
+									1:		L_R1=1;
+									2:		L_R2=1;
+									3:		L_R3=1;
+									default	err_flag=1;
+								endcase		
+								end//NOT
+								
+							RD:	begin
+								NS=S_rd1;
+								Sel_Bus1=3'b100;//Sel_PC
+								Sel_Bus2=3'b001;//Sel_Bus1
+								L_ADD_R=1;
+								end//RD
+							
+							WR:	begin
+								NS=S_wr1;
+								Sel_Bus1=3'b100;//Sel_PC
+								Sel_Bus2=3'b001;//Sel_Bus1
+								L_ADD_R=1;
+								end//WR
+
+							BR: begin
+								NS=S_br1;
+								Sel_Bus1=3'b100;//Sel_PC
+								Sel_Bus2=3'b001;//Sel_Bus1
+								L_ADD_R=1;
+								end//BR
+							
+							BRZ:begin
+								 if(zero==1)begin
+									NS=S_br1;
+									Sel_Bus1=3'b100;//Sel_PC
+									Sel_Bus2=3'b001;//Sel_Bus1
+									L_ADD_R=1;
+								 end
+								 else begin
+									NS=S_fet1;
+									Inc_PC=1;
+								 end
+								end//BRZ
+								 
+							default	NS=S_halt;
+				 
+						endcase//opcode
+						end
+				
+			S_ex1:		begin
+						 NS=S_fet1;
+						 L_R_Z=1;
+						 Sel_Bus2=2'b00;//Sel_ALU
+						 case(dest)
+							0:	begin Sel_Bus1=3'b000;L_R0=1;end
+							1:	begin Sel_Bus1=3'b001;L_R1=1;end
+							2:	begin Sel_Bus1=3'b010;L_R2=1;end
+							3:	begin Sel_Bus1=3'b011;L_R3=1;end
+							default	err_flag=1;
+						 endcase		
+						end	
+			
+			S_rd1:		begin
+						 NS=S_rd2;						 						 
+						 Inc_PC=1;
+						 Sel_Bus2=2'b10;//Sel_Mem
+						 L_ADD_R=1;
+						end
+						
+			S_wr1:		begin
+						 NS=S_wr2;
+						 Inc_PC=1;
+						 Sel_Bus2=2'b10;//Sel_Mem
+						 L_ADD_R=1;
+						end
+						
+			S_rd2:		begin
+						 NS=S_fet1;
+						 Sel_Bus2=2'b10;//Sel_Mem
+						 case(dest)
+							0:		L_R0=1;
+							1:		L_R1=1;
+							2:		L_R2=1;
+							3:		L_R3=1;
+							default	err_flag=1;
+						 endcase
+						end
+						
+			S_wr2:		begin
+						 NS=S_fet1;
+						 write=1;
+						 case(src)
+							0:		Sel_Bus1=3'b000;//R0
+							1:		Sel_Bus1=3'b001;//R1
+							2:		Sel_Bus1=3'b010;//R2
+							3:		Sel_Bus1=3'b011;//R3
+							default	err_flag=1;
+						 endcase	
+						end
+			
+			S_br1:		begin
+						 NS=S_br2;
+						 Sel_Bus2=2'b10;//Sel_Mem	
+						 L_ADD_R=1;
+						end
+						
+			S_br2:		begin	
+						 NS=S_fet1;
+						 Sel_Bus2=2'b10;//Sel_Mem 
+						 L_PC=1;
+						end
+			S_halt:		NS=S_halt;
+			default		NS=S_idle;
+		endcase								 
+	end
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/IR.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/IR.v
@ -0,0 +1,13 @@
+module IR(IR_out,IR_in,load,clk,rst);
+
+output reg	[7:0]IR_out;
+input	[7:0]IR_in;
+input	load,clk,rst;
+
+always@(posedge clk)
+	begin
+		if(rst==1)IR_out<=8'b0;
+		else if(load==1)IR_out<=IR_in;		
+	end
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Memory.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Memory.v
@ -0,0 +1,61 @@
+module Memory(Data_out,Address);
+
+	output	[7:0]Data_out;
+	input	[7:0]Address;
+	
+	reg	[7:0]mem[255:0];
+	assign	Data_out=mem[Address];
+	
+	always@(Address)
+	begin
+		case(Address)
+							//opcode_src_dest
+		//NOP
+		0:	mem[Address]=8'b0000_00_00;
+		
+		//rd 00 10		//Read MEM[130] to R2
+		1:	mem[Address]=8'b0101_00_10; //Instruction 
+		2:	mem[Address]=130;	//Address
+		
+		//rd 00 11		//Read MEM[131] to R3
+		3:	mem[Address]=8'b0101_00_11; //Instruction 
+		4:	mem[Address]=131; //Address
+		
+		//rd 00 01		//Read MEM[128] to R1
+		5:	mem[Address]=8'b0101_00_01; //Instruction
+		6:	mem[Address]=128; //Address
+
+		//rd 00 00		//Read MEM[129] to R0		
+		7:	mem[Address]=8'b0101_00_00; //Instruction
+		8:	mem[Address]=129; //Address
+		
+		//Sub 00 01		//Sub R1-R0 to R1
+		9:	mem[Address]=8'b0010_00_01; //Instruction
+		
+		//BRZ 00 00    
+		10:	mem[Address]=8'b1000_00_00; //Instruction
+		11:	mem[Address]=134; //Address
+		
+		//Add 10 11		//Add R2+R3 to R3
+		12:	mem[Address]=8'b00011011;
+		
+		//BR
+		13:	mem[Address]=8'b01110011; //Instruction
+		14:	mem[Address]=140; //Address
+		
+		128:mem[Address]=6;
+		129:mem[Address]=1;
+		130:mem[Address]=2;
+		131:mem[Address]=0;
+		134:mem[Address]=139; //Address
+		135:mem[Address]=0;
+		//HAL
+		139:mem[Address]=8'b1111_00_00; //Instruction
+		140:mem[Address]=9; //Address
+		default mem[Address]=8'bx;
+    endcase
+  end
+	
+endmodule
+	
+	
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Mux_31.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Mux_31.v
@ -0,0 +1,20 @@
+module Mux_31(Y,A0,A1,A2,sel);
+
+output	[7:0]Y;
+input	[7:0]A2,A1,A0;
+input	[1:0]sel;
+
+reg		[7:0]Y;
+
+always@(*)
+	begin
+		case(sel)
+			0:	Y=A0;
+			1:	Y=A1;
+			2:	Y=A2;
+			default:Y=8'bz;
+		endcase
+	end
+
+endmodule
+
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Mux_51.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Mux_51.v
@ -0,0 +1,22 @@
+module Mux_51(Y,A0,A1,A2,A3,A4,sel);
+
+output	[7:0]Y;
+input	[7:0]A4,A3,A2,A1,A0;
+input	[2:0]sel;
+
+reg		[7:0]Y;
+
+always@(*)
+	begin
+		case(sel)
+			0:	Y=A0;
+			1:	Y=A1;
+			2:	Y=A2;
+			3:	Y=A3;
+			4:	Y=A4;
+			default:Y=8'bx;
+		endcase
+	end
+
+endmodule
+
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/PC.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/PC.v
@ -0,0 +1,17 @@
+module PC(PC_out,PC_in,load,inc,clk,rst);
+
+output	[7:0]PC_out;
+input	[7:0]PC_in;
+input	load,inc,clk,rst;
+
+reg	[7:0]PC_out;
+
+always@(posedge clk)
+	begin
+		if(rst==1)PC_out<=8'b0;
+		else if(load==1)PC_out<=PC_in;
+		else if(inc==1)PC_out<=PC_out+8'b00000001;		
+	end
+	
+endmodule
+
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/RISC_core_mem_top.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/RISC_core_mem_top.v
@ -0,0 +1,23 @@
+module RISC_core_mem_top(Reg_R0_out,Reg_R1_out,Reg_R2_out,Reg_R3_out,bus_1_out,clk,rst);
+
+output	[7:0]bus_1_out;
+input	clk,rst;
+output	[7:0]Reg_R0_out;
+output	[7:0]Reg_R1_out;
+output	[7:0]Reg_R2_out;
+output	[7:0]Reg_R3_out;
+
+wire	[7:0]bus_1_out,MEMAddress;
+wire	clk,rst;
+wire	[7:0]MEMdataout;
+wire	[7:0]Reg_R0_out;
+wire	[7:0]Reg_R1_out;
+wire	[7:0]Reg_R2_out;
+wire	[7:0]Reg_R3_out;
+
+
+RISC_core_top core(Reg_R0_out,Reg_R1_out,Reg_R2_out,Reg_R3_out,bus_1_out,clk,rst,MEMdataout,MEMAddress);
+
+Memory MEM(MEMdataout,MEMAddress);
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/RISC_core_top.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/RISC_core_top.v
@ -0,0 +1,48 @@
+module RISC_core_top(Reg_R0_out,Reg_R1_out,Reg_R2_out,Reg_R3_out,bus_1_out,clk,rst,MEMdataout,MEMAddress);
+
+output	[7:0]bus_1_out,MEMAddress;
+input	clk,rst;
+input	[7:0]MEMdataout;
+output	[7:0]Reg_R0_out;
+output	[7:0]Reg_R1_out;
+output	[7:0]Reg_R2_out;
+output	[7:0]Reg_R3_out;
+
+wire	[7:0]BUS_2,BUS_1,MEMAddress;
+wire	[7:0]alu_out;
+wire	[7:0]MEMdataout;
+wire	[7:0]Reg_Y_out,Reg_R0_out,Reg_R1_out,Reg_R2_out,Reg_R3_out,PC_out;
+wire	[7:0]IR_out;
+wire	zero_flag_out;
+wire	[2:0]Sel_Bus1;
+wire	[1:0]Sel_Bus2;
+wire	L_R0,L_R1,L_R2,L_R3,L_PC,Inc_PC,L_IR,L_ADD_R,L_R_Y,L_R_Z,MEMwrite,zero;
+
+assign	bus_1_out=BUS_1;
+assign	bus_2_out=BUS_2;
+
+Controller CON(L_R0,L_R1,L_R2,L_R3,L_PC,Inc_PC,Sel_Bus1,L_IR,L_ADD_R,L_R_Y,L_R_Z,Sel_Bus2,MEMwrite,zero,IR_out,clk,rst);
+//module PC(PC_out,PC_in,load,inc,clk,rst);
+PC Program_Counter(PC_out,BUS_2,L_PC,Inc_PC,clk,rst);
+//module ALU(zero_flag_out,alu_out,Reg_Y_in,Bus_1_in,IR_code);
+ALU	 Arithmetic_Logic_Unit(zero_flag_out,alu_out,Reg_Y_out,BUS_1,IR_out);
+//module Memory(Data_out,Data_in,MEMAddress,clk,MEMwrite);
+//Memory MEM(MEMdataout,BUS_1,MEMAddress,clk,MEMwrite);
+//module Mux_31(Y,A0,A1,A2,sel);
+Mux_31	 Mux31(BUS_2,alu_out,BUS_1,MEMdataout,Sel_Bus2);
+//module Reg_1bit(Q,D,load,clk,rst);
+Reg_1bit Reg_Z(zero,zero_flag_out,L_R_Z,clk,rst);
+//module Reg_8bit(Q,D,load,clk,rst);
+Reg_8bit Reg_Y(Reg_Y_out,BUS_2,L_R_Y,clk,rst);
+Reg_8bit Add_R(MEMAddress,BUS_2,L_ADD_R,clk,rst);
+//R0~R3
+Reg_8bit Reg_R0(Reg_R0_out,BUS_2,L_R0,clk,rst);
+Reg_8bit Reg_R1(Reg_R1_out,BUS_2,L_R1,clk,rst);
+Reg_8bit Reg_R2(Reg_R2_out,BUS_2,L_R2,clk,rst);
+Reg_8bit Reg_R3(Reg_R3_out,BUS_2,L_R3,clk,rst);
+//module Mux_51(Y,A0,A1,A2,A3,A4,sel);
+Mux_51	Mux51(BUS_1,Reg_R0_out,Reg_R1_out,Reg_R2_out,Reg_R3_out,PC_out,Sel_Bus1);
+//module IR(IR_out,IR_in,load,clk,rst);
+IR	Instruction_Register(IR_out,BUS_2,L_IR,clk,rst);
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/RISC_testbench.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/RISC_testbench.v
@ -0,0 +1,53 @@
+`timescale 1ns/1ns
+
+module RISC_testbench;
+   
+wire	[7:0]bus_1_out;
+reg	clk,rst;
+wire	[7:0]Reg_R0_out;
+wire	[7:0]Reg_R1_out;
+wire	[7:0]Reg_R2_out;
+wire	[7:0]Reg_R3_out;
+
+/* wire	[7:0]MEMAddress;
+wire	[7:0]MEMdataout;
+wire  	MEMwrite; */
+
+/* assign MEMAddress = top.MEMAddress;
+assign MEMdataout = top.MEMdataout;
+assign MEMwrite   = top.MEMwrite; */
+
+RISC_core_mem_top top(Reg_R0_out,Reg_R1_out,Reg_R2_out,Reg_R3_out,bus_1_out,clk,rst);
+
+ always#20 clk=~clk;
+   
+ initial
+    begin
+      clk=0;rst=1;
+      #30 rst=0;
+      #6000 $stop;       
+      end
+
+/* //----------
+
+	integer fp;
+	initial
+	begin
+		fp = $fopen("RISC_xa.vec");
+		
+		$fdisplay(fp, "radix		1	  1	44	44	44	44	44	1	44	44");
+		$fdisplay(fp, "vname	  clk	rst		Reg_R0_out[[7:0]]	Reg_R1_out[[7:0]]	Reg_R2_out[[7:0]]	Reg_R3_out[[7:0]]	bus_1_out[[7:0]]	MEMwrite	MEMAddress	MEMdataout");
+		$fdisplay(fp, "   io		i	  i	oo	oo	oo	oo	oo	o	oo	ii");
+		$fdisplay(fp, "slope		0.3");
+		$fdisplay(fp, "  vih		3.3");
+		$fdisplay(fp, "  vil		0");
+		$fdisplay(fp, "tunit		ns");
+	end
+	always@(clk)
+	begin
+		$fdisplay(fp, "%t %b %b %h %h %h %h %h %b %h %h", $time, clk, rst, Reg_R0_out, Reg_R1_out, Reg_R2_out, Reg_R3_out, bus_1_out, MEMwrite, MEMAddress, MEMdataout);
+	end
+
+//---------- */
+   
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Reg_1bit.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Reg_1bit.v
@ -0,0 +1,16 @@
+module Reg_1bit(Q,D,load,clk,rst);
+
+output	Q;
+input	D;
+input	load,clk,rst;
+
+reg 	Q;
+
+always@(posedge clk)
+	begin
+		if(rst==1)Q<=0;
+		else if(load==1)Q<=D;
+	end
+
+endmodule
+		
--- a/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Reg_8bit.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/RISC_posedge_clk/Reg_8bit.v
@ -0,0 +1,16 @@
+module Reg_8bit(Q,D,load,clk,rst);
+
+output	[7:0]Q;
+input	[7:0]D;
+input	load,clk,rst;
+
+reg [7:0]Q;
+
+always@(posedge clk)
+	begin
+		if(rst==1)Q<=8'b0;
+		else if(load==1)Q<=D;
+	end
+
+endmodule
+		
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/ACC.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/ACC.v
@ -0,0 +1,19 @@
+module ACC(
+	output [7:0] acc_out1,
+	output [7:0] acc_out2,
+	input [7:0] acc_in,
+	input la_,
+	input clk,
+	input clr_
+	);
+	
+	reg [7:0] q;
+	
+	always @(posedge clk)
+		if (~clr_) q <= 8'b0;
+		else if(~la_) q <= acc_in;	
+	
+	assign acc_out1 = q;
+	assign acc_out2 = q;
+	
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/ADDSUB.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/ADDSUB.v
@ -0,0 +1,13 @@
+module ADDSUB(
+	output [7:0] ADDSUB_out,
+	input [7:0] ADDSUB_in1,
+	input [7:0] ADDSUB_in2,
+	input su
+	);
+
+	wire [7:0] d;
+	
+	assign d = su ? ADDSUB_in1 - ADDSUB_in2 : ADDSUB_in1 + ADDSUB_in2;
+	assign ADDSUB_out = d;
+	
+endmodule	
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/BRegister.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/BRegister.v
@ -0,0 +1,13 @@
+module BRegister(
+	output reg [7:0] BRegister_out,
+	input [7:0] BRegister_in,
+	input lb_,
+	input clk,
+	input clr_
+	);
+
+	always @(posedge clk)
+		if(~clr_) BRegister_out <= 8'b0;
+		else if(~lb_) BRegister_out <= BRegister_in;	
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/Controller.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/Controller.v
@ -0,0 +1,119 @@
+module Controller(
+	output reg [11:0] control_signals,
+	input [3:0] opcode,
+	input clk,
+	input clr_
+	); 
+	
+	reg [3:0] ps, ns;
+	
+	always @(posedge clk)
+	begin
+		if(~clr_) ps <= 4'd0;
+		else ps <= ns;
+	end
+	
+	always @(*)
+	begin
+		case(ps)
+			0:
+			begin
+				control_signals = 12'h3e3;
+				ns = 4'd1;		
+			end
+			
+			1: //T1
+			begin
+				control_signals = 12'h5e3;
+				ns = 4'd2;	
+			end
+			
+			2: //T2
+			begin
+//				control_signals = 12'hbe3;
+				control_signals = 12'h863;
+				ns = 4'd3;	
+			end
+			
+			3: //T3
+			begin
+//				control_signals = 12'h263;
+				control_signals = 12'h3e3;
+				if(opcode == 4'd0) //LDA
+					ns = 4'd4;
+				else if(opcode == 4'd1) //ADD
+					ns = 4'd6;
+				else if(opcode == 4'd2) //SUB
+					ns = 4'd9;
+				else if(opcode == 4'd14) //OUT
+					ns = 4'd12;
+				else if(opcode == 4'd15) //HLT
+					ns = 4'd13;
+			end
+			
+			4: //LDA
+			begin
+				control_signals = 12'h1a3;
+				ns = 4'd5;	
+			end
+			
+			5: //LDA
+			begin
+				control_signals = 12'h2c3;
+				ns = 4'd1;	
+			end
+			
+			6: //ADD
+			begin
+				control_signals = 12'h1a3;
+				ns = 4'd7;	
+			end
+			
+			7: //ADD
+			begin
+				control_signals = 12'h2e1;
+				ns = 4'd8;	
+			end
+			
+			8: //ADD
+			begin
+				control_signals = 12'h3c7;
+				ns = 4'd1;	
+			end
+			
+			9: //SUB
+			begin
+				control_signals = 12'h1a3;
+				ns = 4'd10;	
+			end
+			
+			10: //SUB
+			begin
+				control_signals = 12'h2e1;
+				ns = 4'd11;	
+			end
+			
+			11: //SUB
+			begin
+				control_signals = 12'h3cf;
+				ns = 4'd1;	
+			end
+			
+			12: //OUT
+			begin
+				control_signals = 12'h3f2;
+				ns = 4'd1;	
+			end
+			
+			13: //HLT
+				ns = 4'd13;	
+			
+			default: 
+			begin
+				ns = 4'd0;
+				control_signals = 12'h3e3;
+			end
+		
+		endcase
+	end	
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/IR.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/IR.v
@ -0,0 +1,21 @@
+module IR(
+	output [7:4] opcode,
+	output [3:0] oprand,
+	input wire [7:0] IR_in,
+	input li_,
+	input clk,
+	input clr_
+	);
+	
+	reg [7:0] q;
+
+	always @(posedge clk)
+	begin
+		if(~clr_) q <=8'b0;
+		else if(~li_) q <= IR_in;
+	end
+	
+	assign opcode = q[7:4];
+	assign oprand = q[3:0];
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/MAR.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/MAR.v
@ -0,0 +1,13 @@
+module MAR(
+	output reg [3:0] mar_out,
+	input wire [3:0] mar_in,
+	input lm_,
+	input clk,
+	input clr_
+	);
+	
+	always @(posedge clk)
+		if(~clr_) mar_out <= 4'b0;
+		else if(~lm_) mar_out <= mar_in;	
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/OutputRegister.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/OutputRegister.v
@ -0,0 +1,13 @@
+module OutputRegister(
+	output reg [7:0] OutputRegister_out,
+	input [7:0] OutputRegister_in,
+	input lo_,
+	input clk,
+	input clr_
+	);
+
+	always @(posedge clk)
+		if(~clr_) OutputRegister_out <= 8'b0;
+		else if(~lo_) OutputRegister_out <= OutputRegister_in;
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/PC.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/PC.v
@ -0,0 +1,15 @@
+ module PC(
+	output reg [3:0] pc_out,
+	input cp,
+	input clk,
+	input clr_
+	);
+	
+	always @(posedge clk)
+	begin
+		if(~clr_) pc_out <= 0;
+		else if (cp) pc_out <= pc_out + 1;
+	end
+			
+endmodule
+
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/ROM.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/ROM.v
@ -0,0 +1,26 @@
+module ROM(
+	output reg [7:0] rom_out,
+	input [3:0] rom_in
+	);
+	
+	always @(*)
+	begin
+		rom_out = 8'bx;
+		case(rom_in)
+			4'b0000: rom_out = 8'b0000_1001; //LDA
+			4'b0001: rom_out = 8'b0001_1010; //ADD
+			4'b0010: rom_out = 8'b0001_1011; //ADD
+			4'b0011: rom_out = 8'b0010_1100; //SUB
+			4'b0100: rom_out = 8'b1110_xxxx; //OUT
+			4'b0101: rom_out = 8'b1111_xxxx; //HLT
+			4'b0110: rom_out = 8'bxxxx_xxxx;
+			4'b0111: rom_out = 8'bxxxx_xxxx;
+			4'b1000: rom_out = 8'bxxxx_xxxx;
+			4'b1001: rom_out = 8'b0001_0000;
+			4'b1010: rom_out = 8'b0001_0100;
+			4'b1011: rom_out = 8'b0001_1000;
+			4'b1100: rom_out = 8'b0010_0000;	
+		endcase
+	end
+	
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/SAPone.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/SAPone.v
@ -0,0 +1,100 @@
+module SAPone(
+	output wire [7:0] SAP_out,
+	output wire [11:0] con,
+	output reg [7:0] bus,
+	input clk,
+	input clr_
+	);
+	
+	wire cp, ep, lm_, ce_, li_, ei_, la_, ea, su, eu, lb_, lo_;
+	wire [7:0] acc_out2, BRegister_out, OutputRegister_out;
+	wire [3:0] IR_out, mar_out;
+	wire [4:0] bus_sel;
+	wire [3:0] pc_out, oprand;
+	wire [7:0] rom_out, acc_out1, ADDSUB_out;
+	
+	assign {cp, ep, lm_, ce_, li_, ei_, la_, ea, su, eu, lb_, lo_} = con;
+	assign bus_sel = {ep, ce_, ei_, ea, eu};
+	
+	always@(*)
+	begin
+		case(bus_sel)
+			5'b11100: bus[3:0] = pc_out;
+			5'b00100: bus[7:0] = rom_out;
+			5'b01000: bus[3:0] = oprand;
+			5'b01110: bus[7:0] = acc_out1;
+			5'b01101: bus[7:0] = ADDSUB_out;
+			default:  bus[7:0] = 8'bx;
+		endcase
+	end
+	
+	PC pc1(
+		.pc_out(pc_out),
+		.cp(cp),
+		.clk(clk),
+		.clr_(clr_)
+		);
+		
+	MAR mar1(
+		.mar_out(mar_out),
+		.mar_in(bus[3:0]),
+		.lm_(lm_),
+		.clk(clk),
+		.clr_(clr_)
+		);
+		
+	ROM roml(
+		.rom_out(rom_out),
+		.rom_in(mar_out)
+		);
+	
+	IR ir1(
+		.opcode(IR_out),
+		.oprand(oprand),
+		.IR_in(bus[7:0]),
+		.li_(li_),
+		.clk(clk),
+		.clr_(clr_)
+		);
+	
+	Controller cont1(
+		.control_signals(con),
+		.opcode(IR_out),
+		.clk(clk),
+		.clr_(clr_)
+		); 
+
+	ACC acc1(
+		.acc_out1(acc_out1),
+		.acc_out2(acc_out2),
+		.acc_in(bus[7:0]),
+		.la_(la_),
+		.clk(clk),
+		.clr_(clr_)
+		);
+			
+	ADDSUB addsub1(
+		.ADDSUB_out(ADDSUB_out),
+		.ADDSUB_in1(acc_out2),
+		.ADDSUB_in2(BRegister_out),
+		.su(su)
+		);
+		
+	BRegister bregister1(
+		.BRegister_out(BRegister_out),
+		.BRegister_in(bus[7:0]),
+		.lb_(lb_),
+		.clk(clk),
+		.clr_(clr_)
+		);
+		
+	OutputRegister outputregister1(
+		.OutputRegister_out(SAP_out),
+		.OutputRegister_in(bus[7:0]),
+		.lo_(lo_),
+		.clk(clk),
+		.clr_(clr_)
+		);			
+
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/SAPone/testSAPone.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/SAPone/testSAPone.v
@ -0,0 +1,34 @@
+module testSAPone;
+
+	wire [7:0] SAP_out;
+	wire [11:0] con;
+	wire [7:0] bus;
+//	wire clk_out, clr_out;
+	reg clk, clr_;
+
+	always #5 clk = ~clk;
+	
+	SAPone sapone1(
+		.SAP_out(SAP_out),
+		.con(con),
+		.bus(bus),
+//		.clk_out(clk_out),
+//		.clr_out(clr_out),
+		.clk(clk),
+		.clr_(clr_)
+		);
+		
+//	PC pc1(bus[3:0], clk, clr_, cp, ep);
+//	MAR mar1(mar, clk, lm_, bus[3:0]);
+
+	initial
+	begin
+		clk = 0; clr_ = 0;
+		#10 clr_ = 1;
+		
+		
+		
+		#990 $stop;	
+	end
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/counter/counter.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/counter/counter.v
@ -0,0 +1,16 @@
+module counter(clk_counter, q_counter, rst_counter);
+
+    input clk_counter;
+    input rst_counter;
+    output [7:0] q_counter;
+    reg [7:0] q_counter;
+
+    always @ (posedge clk_counter)
+    begin
+        if(rst_counter)
+		q_counter <= 8'b00000000;
+	else
+		q_counter <= q_counter + 1;
+    end
+
+endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/counter/counter_output_verilog.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/counter/counter_output_verilog.v
@ -1,84 +0,0 @@
-/* Generated by Yosys 0.9 (git sha1 f110c953, gcc 8.4.0-1ubuntu1~18.04 -fPIC -Os) */
-
-module counter(clk_counter, rst_counter, \q_counter[0] , \q_counter[1] , \q_counter[2] , \q_counter[3] , \q_counter[4] , \q_counter[5] , \q_counter[6] , \q_counter[7] );
-  wire _00_;
-  wire _01_;
-  input clk_counter;
-  wire n22;
-  wire n26;
-  wire n30;
-  wire n34;
-  wire n38;
-  wire n42;
-  wire n46;
-  wire n50;
-  output \q_counter[0] ;
-  reg \q_counter[0] ;
-  output \q_counter[1] ;
-  reg \q_counter[1] ;
-  output \q_counter[2] ;
-  reg \q_counter[2] ;
-  output \q_counter[3] ;
-  reg \q_counter[3] ;
-  output \q_counter[4] ;
-  reg \q_counter[4] ;
-  output \q_counter[5] ;
-  reg \q_counter[5] ;
-  output \q_counter[6] ;
-  reg \q_counter[6] ;
-  output \q_counter[7] ;
-  reg \q_counter[7] ;
-  input rst_counter;
-  
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[0] <= 1'b0;
-    else \q_counter[0]  <= n22;
-  end
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[1] <= 1'b0;
-    else \q_counter[1]  <= n26;
-  end
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[2] <= 1'b0;
-    else \q_counter[2]  <= n30;
-  end
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[3] <= 1'b0;
-    else \q_counter[3]  <= n34;
-  end
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[4] <= 1'b0;
-    else \q_counter[4]  <= n38;
-  end
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[5] <= 1'b0;
-    else \q_counter[5]  <= n42;
-  end
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[6] <= 1'b0;
-    else \q_counter[6]  <= n46;
-  end
-  always @(posedge clk_counter)
-  begin
-	if(rst_counter) \q_counter[7] <= 1'b0;
-    else \q_counter[7]  <= n50;
-  end
-  
-  assign n26 = 8'h14 >> { \q_counter[0] , \q_counter[1] , rst_counter };
-  assign n30 = 16'h0708 >> { \q_counter[2] , rst_counter, \q_counter[0] , \q_counter[1]  };
-  assign n34 = 32'd8323200 >> { \q_counter[3] , rst_counter, \q_counter[0] , \q_counter[1] , \q_counter[2]  };
-  assign n38 = 64'h00007fff00008000 >> { \q_counter[4] , rst_counter, \q_counter[0] , \q_counter[1] , \q_counter[2] , \q_counter[3]  };
-  assign n42 = 8'h14 >> { _00_, \q_counter[5] , rst_counter };
-  assign _00_ = 32'd2147483648 >> { \q_counter[0] , \q_counter[1] , \q_counter[2] , \q_counter[3] , \q_counter[4]  };
-  assign n46 = 8'h14 >> { _01_, \q_counter[6] , rst_counter };
-  assign _01_ = 64'h8000000000000000 >> { \q_counter[0] , \q_counter[1] , \q_counter[2] , \q_counter[3] , \q_counter[4] , \q_counter[5]  };
-  assign n50 = 16'h0708 >> { \q_counter[7] , rst_counter, _01_, \q_counter[6]  };
-  assign n22 = 4'h1 >> { \q_counter[0] , rst_counter };
-endmodule
--- a/openfpga_flow/benchmarks/micro_benchmark/counter/counter_post_yosys.blif
+++ b/openfpga_flow/benchmarks/micro_benchmark/counter/counter_post_yosys.blif
@ -1,69 +0,0 @@
-# Generated by Yosys 0.9 (git sha1 UNKNOWN, clang 7.0.0 -fPIC -Os)
-
-.model counter
-.inputs clk_counter rst_counter
-.outputs q_counter[0] q_counter[1] q_counter[2] q_counter[3] q_counter[4] q_counter[5] q_counter[6] q_counter[7]
-.names $false
-.names $true
-1
-.names $undef
-.names q_counter[7] rst_counter q_counter[6] $abc$3686$new_n20_ $0\q_counter[7][0:0]
-0011 1
-1000 1
-1001 1
-1010 1
-.names q_counter[4] q_counter[5] q_counter[3] q_counter[2] q_counter[1] q_counter[0] $abc$3686$new_n20_
-111111 1
-.names q_counter[6] $abc$3686$new_n20_ rst_counter $0\q_counter[6][0:0]
-010 1
-100 1
-.names q_counter[5] $abc$3686$new_n23_ rst_counter $0\q_counter[5][0:0]
-010 1
-100 1
-.names q_counter[4] q_counter[3] q_counter[2] q_counter[1] q_counter[0] $abc$3686$new_n23_
-11111 1
-.names q_counter[2] rst_counter q_counter[1] q_counter[0] $0\q_counter[2][0:0]
-0011 1
-1000 1
-1001 1
-1010 1
-.names q_counter[4] rst_counter q_counter[3] q_counter[2] q_counter[1] q_counter[0] $0\q_counter[4][0:0]
-001111 1
-100000 1
-100001 1
-100010 1
-100011 1
-100100 1
-100101 1
-100110 1
-100111 1
-101000 1
-101001 1
-101010 1
-101011 1
-101100 1
-101101 1
-101110 1
-.names q_counter[3] rst_counter q_counter[2] q_counter[1] q_counter[0] $0\q_counter[3][0:0]
-00111 1
-10000 1
-10001 1
-10010 1
-10011 1
-10100 1
-10101 1
-10110 1
-.names q_counter[1] q_counter[0] rst_counter $0\q_counter[1][0:0]
-010 1
-100 1
-.names q_counter[0] rst_counter $0\q_counter[0][0:0]
-00 1
-.latch $0\q_counter[7][0:0] q_counter[7] re clk_counter 2
-.latch $0\q_counter[6][0:0] q_counter[6] re clk_counter 2
-.latch $0\q_counter[5][0:0] q_counter[5] re clk_counter 2
-.latch $0\q_counter[4][0:0] q_counter[4] re clk_counter 2
-.latch $0\q_counter[3][0:0] q_counter[3] re clk_counter 2
-.latch $0\q_counter[2][0:0] q_counter[2] re clk_counter 2
-.latch $0\q_counter[1][0:0] q_counter[1] re clk_counter 2
-.latch $0\q_counter[0][0:0] q_counter[0] re clk_counter 2
-.end
--- a/openfpga_flow/benchmarks/micro_benchmark/counter/counter_pre_vpr.act
+++ b/openfpga_flow/benchmarks/micro_benchmark/counter/counter_pre_vpr.act
@ -1,20 +0,0 @@
-clk_counter 0.500000 2.000000
-rst_counter 0.492200 0.201800
-q_counter[0] 0.281800 0.563400
-q_counter[1] 0.248200 0.273600
-q_counter[2] 0.183200 0.125600
-q_counter[3] 0.097400 0.044800
-q_counter[4] 0.022600 0.007200
-q_counter[5] 0.002200 0.000800
-q_counter[6] 0.000000 0.000000
-q_counter[7] 0.000000 0.000000
-$0\q_counter[7][0:0] 0 0
-$0\q_counter[6][0:0] 0 0
-$0\q_counter[5][0:0] 0 0
-$0\q_counter[4][0:0] 0 0
-$0\q_counter[3][0:0] 0 0
-$0\q_counter[2][0:0] 0 0
-$0\q_counter[1][0:0] 0 0
-$0\q_counter[0][0:0] 0 0
-$abc$3686$new_n23_ 0 0
-$abc$3686$new_n20_ 0 0
--- a/openfpga_flow/benchmarks/micro_benchmark/counter/counter_tb.v
+++ b/openfpga_flow/benchmarks/micro_benchmark/counter/counter_tb.v
@ -0,0 +1,24 @@
+module counter_tb;
+
+	reg clk_counter, rst_counter;
+	wire [7:0] q_counter;
+
+	counter_original C_1(
+		clk_counter, 
+		q_counter, 
+		rst_counter);
+	
+	initial begin
+		#0 rst_counter = 1'b1; clk_counter = 1'b0;
+		#100 rst_counter = 1'b0;
+	end
+
+	always begin
+		#10 clk_counter = ~clk_counter;
+	end
+	
+	initial begin
+		#5000 $stop;
+	end
+
+endmodule
--- a/openfpga_flow/openfpga_arch/k6_frac_N10_adder_register_scan_chain_depop50_spypad_40nm_openfpga.xml
+++ b/openfpga_flow/openfpga_arch/k6_frac_N10_adder_register_scan_chain_depop50_spypad_40nm_openfpga.xml
@ -242,7 +242,7 @@
    <!-- physical pb_type binding in complex block CLB -->
    <!-- physical mode will be the default mode if not specified -->
    <pb_type name="clb.fle" physical_mode_name="physical"/>
-    <pb_type name="clb.fle[physical].frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="11"/>
+    <pb_type name="clb.fle[physical].frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="00"/>
    <pb_type name="clb.fle[physical].ff_phy" circuit_model_name="scan_chain_ff"/>
    <pb_type name="clb.fle[physical].frac_logic.adder_phy" circuit_model_name="adder"/>
    <!-- Binding operating pb_type to physical pb_type -->
@ -277,7 +277,7 @@
    <pb_type name="clb_spypad.fle" physical_mode_name="physical"/>

    <!-- Binding regular FLEs -->
-    <pb_type name="clb_spypad.fle[physical].frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="11"/>
+    <pb_type name="clb_spypad.fle[physical].frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="00"/>
    <pb_type name="clb_spypad.fle[physical].ff_phy" circuit_model_name="scan_chain_ff"/>
    <pb_type name="clb_spypad.fle[physical].frac_logic.adder_phy" circuit_model_name="adder"/>
    <!-- Binding operating pb_type to physical pb_type -->
@ -310,7 +310,7 @@
    <!-- physical mode will be the default mode if not specified -->
    <pb_type name="clb_spypad.fle_spypad" physical_mode_name="physical"/>

-    <pb_type name="clb_spypad.fle_spypad[physical].frac_logic.frac_lut6" circuit_model_name="frac_lut6_spypad" mode_bits="11"/>
+    <pb_type name="clb_spypad.fle_spypad[physical].frac_logic.frac_lut6" circuit_model_name="frac_lut6_spypad" mode_bits="00"/>
    <pb_type name="clb_spypad.fle_spypad[physical].ff_phy" circuit_model_name="scan_chain_ff"/>
    <pb_type name="clb_spypad.fle_spypad[physical].frac_logic.adder_phy" circuit_model_name="adder"/>
    <!-- Binding operating pb_type to physical pb_type -->
--- a/openfpga_flow/openfpga_arch/k6_frac_N8_40nm_openfpga.xml
+++ b/openfpga_flow/openfpga_arch/k6_frac_N8_40nm_openfpga.xml
@ -0,0 +1,231 @@
+<!-- Architecture annotation for OpenFPGA framework
+     This annotation supports the k6_N8_40nm.xml 
+     - General purpose logic block
+       - K = 6, N = 8, I = 32
+       - Single mode
+     - Routing architecture
+       - L = 4, fc_in = 0.15, fc_out = 0.1
+  -->
+<openfpga_architecture>
+  <technology_library>
+    <device_library>
+      <device_model name="logic" type="transistor">
+        <lib type="industry" corner="TOP_TT" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="0.9" pn_ratio="2"/>
+        <pmos name="pch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+        <nmos name="nch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+      </device_model>
+      <device_model name="io" type="transistor">
+        <lib type="academia" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="2.5" pn_ratio="3"/>
+        <pmos name="pch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+        <nmos name="nch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+      </device_model>
+    </device_library>
+    <variation_library>
+      <variation name="logic_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+      <variation name="io_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+    </variation_library>
+  </technology_library>
+  <circuit_library>
+    <circuit_model type="inv_buf" name="INVTX1" prefix="INVTX1" is_default="true">
+      <design_technology type="cmos" topology="inverter" size="1"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="buf4" prefix="buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="2" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="tap_buf4" prefix="tap_buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="3" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="gate" name="OR2" prefix="OR2" is_default="true">
+      <design_technology type="cmos" topology="OR"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="a" size="1"/>
+      <port type="input" prefix="b" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="pass_gate" name="TGATE" prefix="TGATE" is_default="true">
+      <design_technology type="cmos" topology="transmission_gate" nmos_size="1" pmos_size="2"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="input" prefix="sel" size="1"/>
+      <port type="input" prefix="selb" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in sel selb" out_port="out">
+        10e-12 5e-12 5e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in sel selb" out_port="out">
+        10e-12 5e-12 5e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="chan_wire" name="chan_segment" prefix="track_seg" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="101" C="22.5e-15" num_level="1"/> <!-- model_type could be T, res_val and cap_val DON'T CARE -->
+    </circuit_model>
+    <circuit_model type="wire" name="direct_interc" prefix="direct_interc" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="0" C="0" num_level="1"/> <!-- model_type could be T, res_val cap_val should be defined -->
+    </circuit_model>
+    <circuit_model type="mux" name="mux_2level" prefix="mux_2level" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="multi_level" num_level="2" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <circuit_model type="mux" name="mux_2level_tapbuf" prefix="mux_2level_tapbuf" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="multi_level" num_level="2" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="tap_buf4"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <circuit_model type="mux" name="mux_1level_tapbuf" prefix="mux_1level_tapbuf" is_default="true" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="one_level" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="tap_buf4"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <!--DFF subckt ports should be defined as <D> <Q> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ff" name="static_dff" prefix="dff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="input" prefix="set" size="1" is_global="true" default_val="0" is_set="true"/>
+       <port type="input" prefix="reset" size="1" is_global="true" default_val="0" is_reset="true"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="clock" prefix="clk" size="1" is_global="true" default_val="0" />
+    </circuit_model>
+    <circuit_model type="lut" name="frac_lut6" prefix="frac_lut6" dump_structural_verilog="true">
+      <design_technology type="cmos" fracturable_lut="true"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_inverter exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_buffer exist="true" circuit_model_name="buf4"/>
+      <lut_intermediate_buffer exist="true" circuit_model_name="buf4" location_map="-1-1-"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="6" tri_state_map="-----1" circuit_model_name="OR2"/>
+      <port type="output" prefix="lut5_out" size="2" lut_frac_level="5" lut_output_mask="0,1"/>
+      <port type="output" prefix="lut6_out" size="1" lut_output_mask="0"/>
+      <port type="sram" prefix="sram" size="64"/>
+      <port type="sram" prefix="mode" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+    </circuit_model>
+    <!--Scan-chain DFF subckt ports should be defined as <D> <Q> <Qb> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ccff" name="sc_dff_compact" prefix="scff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="pReset" lib_name="reset" size="1" is_global="true" default_val="0" is_reset="true" is_prog="true"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="output" prefix="Qb" size="1"/>
+       <port type="clock" prefix="prog_clk" lib_name="clk" size="1" is_global="true" default_val="0" is_prog="true"/>
+    </circuit_model>
+    <circuit_model type="iopad" name="iopad" prefix="iopad" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/io.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/io.v">
+      <design_technology type="cmos"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <port type="inout" prefix="pad" size="1" is_global="true" is_io="true"/>
+      <port type="sram" prefix="en" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+      <port type="input" prefix="outpad" size="1"/>
+      <port type="output" prefix="inpad" size="1"/>
+    </circuit_model>
+  </circuit_library>
+  <configuration_protocol>
+    <organization type="scan_chain" circuit_model_name="sc_dff_compact"/>
+  </configuration_protocol>
+  <connection_block>
+    <switch name="ipin_cblock" circuit_model_name="mux_2level_tapbuf"/>
+  </connection_block>
+  <switch_block>
+    <switch name="0" circuit_model_name="mux_2level_tapbuf"/>
+  </switch_block>
+  <routing_segment>
+    <segment name="L4" circuit_model_name="chan_segment"/>
+  </routing_segment>
+  <pb_type_annotations>
+    <!-- physical pb_type binding in complex block IO -->
+    <pb_type name="io" physical_mode_name="physical" idle_mode_name="inpad"/>
+    <pb_type name="io[physical].iopad" circuit_model_name="iopad" mode_bits="1"/> 
+    <pb_type name="io[inpad].inpad" physical_pb_type_name="io[physical].iopad" mode_bits="1"/> 
+    <pb_type name="io[outpad].outpad" physical_pb_type_name="io[physical].iopad" mode_bits="0"/> 
+    <!-- End physical pb_type binding in complex block IO -->
+
+    <!-- physical pb_type binding in complex block CLB -->
+    <!-- physical mode will be the default mode if not specified -->
+    <pb_type name="clb">
+      <!-- Binding interconnect to circuit models as their physical implementation, if not defined, we use the default model -->
+      <interconnect name="crossbar" circuit_model_name="mux_2level"/>
+    </pb_type>
+    <pb_type name="clb.fle" physical_mode_name="physical"/>
+    <pb_type name="clb.fle[physical].fabric.frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="0"/>
+    <pb_type name="clb.fle[physical].fabric.ff" circuit_model_name="static_dff"/>
+    <!-- Binding operating pb_type to physical pb_type -->
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.lut5" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="1" physical_pb_type_index_factor="0.5">
+      <!-- Binding the lut5 to the first 5 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:4]"/>
+      <port name="out" physical_mode_port="lut5_out[0:0]" physical_mode_pin_rotate_offset="1"/>
+    </pb_type>
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.ff" physical_pb_type_name="clb.fle[physical].fabric.ff"/>
+    <pb_type name="clb.fle[n1_lut6].ble6.lut6" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="0">
+      <!-- Binding the lut6 to the first 6 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:5]"/>
+      <port name="out" physical_mode_port="lut6_out"/>
+    </pb_type>
+    <pb_type name="clb.fle[n1_lut6].ble6.ff" physical_pb_type_name="clb.fle[physical].fabric.ff" physical_pb_type_index_factor="2" physical_pb_type_index_offset="0"/>
+    <!-- End physical pb_type binding in complex block IO -->
+  </pb_type_annotations>
+</openfpga_architecture>
--- a/openfpga_flow/openfpga_arch/k6_frac_N8_local_encoder_40nm_openfpga.xml
+++ b/openfpga_flow/openfpga_arch/k6_frac_N8_local_encoder_40nm_openfpga.xml
@ -0,0 +1,231 @@
+<!-- Architecture annotation for OpenFPGA framework
+     This annotation supports the k6_N10_40nm.xml 
+     - General purpose logic block
+       - K = 6, N = 8, I = 40
+       - Single mode
+     - Routing architecture
+       - L = 4, fc_in = 0.15, fc_out = 0.1
+  -->
+<openfpga_architecture>
+  <technology_library>
+    <device_library>
+      <device_model name="logic" type="transistor">
+        <lib type="industry" corner="TOP_TT" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="0.9" pn_ratio="2"/>
+        <pmos name="pch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+        <nmos name="nch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+      </device_model>
+      <device_model name="io" type="transistor">
+        <lib type="academia" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="2.5" pn_ratio="3"/>
+        <pmos name="pch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+        <nmos name="nch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+      </device_model>
+    </device_library>
+    <variation_library>
+      <variation name="logic_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+      <variation name="io_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+    </variation_library>
+  </technology_library>
+  <circuit_library>
+    <circuit_model type="inv_buf" name="INVTX1" prefix="INVTX1" is_default="true">
+      <design_technology type="cmos" topology="inverter" size="1"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="buf4" prefix="buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="2" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="tap_buf4" prefix="tap_buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="3" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="gate" name="OR2" prefix="OR2" is_default="true">
+      <design_technology type="cmos" topology="OR"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="a" size="1"/>
+      <port type="input" prefix="b" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="pass_gate" name="TGATE" prefix="TGATE" is_default="true">
+      <design_technology type="cmos" topology="transmission_gate" nmos_size="1" pmos_size="2"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="input" prefix="sel" size="1"/>
+      <port type="input" prefix="selb" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in sel selb" out_port="out">
+        10e-12 5e-12 5e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in sel selb" out_port="out">
+        10e-12 5e-12 5e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="chan_wire" name="chan_segment" prefix="track_seg" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="101" C="22.5e-15" num_level="1"/> <!-- model_type could be T, res_val and cap_val DON'T CARE -->
+    </circuit_model>
+    <circuit_model type="wire" name="direct_interc" prefix="direct_interc" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="0" C="0" num_level="1"/> <!-- model_type could be T, res_val cap_val should be defined -->
+    </circuit_model>
+    <circuit_model type="mux" name="mux_2level" prefix="mux_2level" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="multi_level" num_level="2" add_const_input="true" const_input_val="1" local_encoder="true"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <circuit_model type="mux" name="mux_2level_tapbuf" prefix="mux_2level_tapbuf" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="multi_level" num_level="2" add_const_input="true" const_input_val="1" local_encoder="true"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="tap_buf4"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <circuit_model type="mux" name="mux_1level_tapbuf" prefix="mux_1level_tapbuf" is_default="true" dump_structural_verilog="true" local_encoder="true">
+      <design_technology type="cmos" structure="one_level" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="tap_buf4"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <!--DFF subckt ports should be defined as <D> <Q> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ff" name="static_dff" prefix="dff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="input" prefix="set" size="1" is_global="true" default_val="0" is_set="true"/>
+       <port type="input" prefix="reset" size="1" is_global="true" default_val="0" is_reset="true"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="clock" prefix="clk" size="1" is_global="true" default_val="0" />
+    </circuit_model>
+    <circuit_model type="lut" name="frac_lut6" prefix="frac_lut6" dump_structural_verilog="true">
+      <design_technology type="cmos" fracturable_lut="true"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_inverter exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_buffer exist="true" circuit_model_name="buf4"/>
+      <lut_intermediate_buffer exist="true" circuit_model_name="buf4" location_map="-1-1-"/>
+      <pass_gate_logic circuit_model_name="TGATE"/>
+      <port type="input" prefix="in" size="6" tri_state_map="-----1" circuit_model_name="OR2"/>
+      <port type="output" prefix="lut5_out" size="2" lut_frac_level="5" lut_output_mask="0,1"/>
+      <port type="output" prefix="lut6_out" size="1" lut_output_mask="0"/>
+      <port type="sram" prefix="sram" size="64"/>
+      <port type="sram" prefix="mode" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+    </circuit_model>
+    <!--Scan-chain DFF subckt ports should be defined as <D> <Q> <Qb> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ccff" name="sc_dff_compact" prefix="scff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="pReset" lib_name="reset" size="1" is_global="true" default_val="0" is_reset="true" is_prog="true"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="output" prefix="Qb" size="1"/>
+       <port type="clock" prefix="prog_clk" lib_name="clk" size="1" is_global="true" default_val="0" is_prog="true"/>
+    </circuit_model>
+    <circuit_model type="iopad" name="iopad" prefix="iopad" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/io.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/io.v">
+      <design_technology type="cmos"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <port type="inout" prefix="pad" size="1" is_global="true" is_io="true"/>
+      <port type="sram" prefix="en" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+      <port type="input" prefix="outpad" size="1"/>
+      <port type="output" prefix="inpad" size="1"/>
+    </circuit_model>
+  </circuit_library>
+  <configuration_protocol>
+    <organization type="scan_chain" circuit_model_name="sc_dff_compact"/>
+  </configuration_protocol>
+  <connection_block>
+    <switch name="ipin_cblock" circuit_model_name="mux_2level_tapbuf"/>
+  </connection_block>
+  <switch_block>
+    <switch name="0" circuit_model_name="mux_2level_tapbuf"/>
+  </switch_block>
+  <routing_segment>
+    <segment name="L4" circuit_model_name="chan_segment"/>
+  </routing_segment>
+  <pb_type_annotations>
+    <!-- physical pb_type binding in complex block IO -->
+    <pb_type name="io" physical_mode_name="physical" idle_mode_name="inpad"/>
+    <pb_type name="io[physical].iopad" circuit_model_name="iopad" mode_bits="1"/> 
+    <pb_type name="io[inpad].inpad" physical_pb_type_name="io[physical].iopad" mode_bits="1"/> 
+    <pb_type name="io[outpad].outpad" physical_pb_type_name="io[physical].iopad" mode_bits="0"/> 
+    <!-- End physical pb_type binding in complex block IO -->
+
+    <!-- physical pb_type binding in complex block CLB -->
+    <!-- physical mode will be the default mode if not specified -->
+    <pb_type name="clb">
+      <!-- Binding interconnect to circuit models as their physical implementation, if not defined, we use the default model -->
+      <interconnect name="crossbar" circuit_model_name="mux_2level"/>
+    </pb_type>
+    <pb_type name="clb.fle" physical_mode_name="physical"/>
+    <pb_type name="clb.fle[physical].fabric.frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="0"/>
+    <pb_type name="clb.fle[physical].fabric.ff" circuit_model_name="static_dff"/>
+    <!-- Binding operating pb_type to physical pb_type -->
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.lut5" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="1" physical_pb_type_index_factor="0.5">
+      <!-- Binding the lut5 to the first 5 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:4]"/>
+      <port name="out" physical_mode_port="lut5_out[0:0]" physical_mode_pin_rotate_offset="1"/>
+    </pb_type>
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.ff" physical_pb_type_name="clb.fle[physical].fabric.ff"/>
+    <pb_type name="clb.fle[n1_lut6].ble6.lut6" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="0">
+      <!-- Binding the lut6 to the first 6 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:5]"/>
+      <port name="out" physical_mode_port="lut6_out"/>
+    </pb_type>
+    <pb_type name="clb.fle[n1_lut6].ble6.ff" physical_pb_type_name="clb.fle[physical].fabric.ff" physical_pb_type_index_factor="2" physical_pb_type_index_offset="0"/>
+    <!-- End physical pb_type binding in complex block IO -->
+  </pb_type_annotations>
+</openfpga_architecture>
--- a/openfpga_flow/openfpga_arch/k6_frac_N8_stdcell_mux_40nm_openfpga.xml
+++ b/openfpga_flow/openfpga_arch/k6_frac_N8_stdcell_mux_40nm_openfpga.xml
@ -0,0 +1,223 @@
+<!-- Architecture annotation for OpenFPGA framework
+     This annotation supports the k6_N10_40nm.xml 
+     - General purpose logic block
+       - K = 6, N = 8, I = 40
+       - Single mode
+     - Routing architecture
+       - L = 4, fc_in = 0.15, fc_out = 0.1
+  -->
+<openfpga_architecture>
+  <technology_library>
+    <device_library>
+      <device_model name="logic" type="transistor">
+        <lib type="industry" corner="TOP_TT" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="0.9" pn_ratio="2"/>
+        <pmos name="pch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+        <nmos name="nch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+      </device_model>
+      <device_model name="io" type="transistor">
+        <lib type="academia" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="2.5" pn_ratio="3"/>
+        <pmos name="pch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+        <nmos name="nch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+      </device_model>
+    </device_library>
+    <variation_library>
+      <variation name="logic_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+      <variation name="io_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+    </variation_library>
+  </technology_library>
+  <circuit_library>
+    <circuit_model type="inv_buf" name="INVTX1" prefix="INVTX1" is_default="true">
+      <design_technology type="cmos" topology="inverter" size="1"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="buf4" prefix="buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="2" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="tap_buf4" prefix="tap_buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="3" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="gate" name="OR2" prefix="OR2" is_default="true">
+      <design_technology type="cmos" topology="OR"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="a" size="1"/>
+      <port type="input" prefix="b" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+    </circuit_model>
+    <!-- Define a circuit model for the standard cell MUX2 
+         OpenFPGA requires the following truth table for the MUX2
+         When the select signal sel is enabled, the first input, i.e., in0 
+         will be propagated to the output, i.e., out
+         If your standard cell provider does not offer the exact truth table,
+         you can simply swap the inputs as shown in the example below
+      -->
+    <circuit_model type="gate" name="MUX2" prefix="MUX2" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/mux2.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/mux2.v">
+      <design_technology type="cmos" topology="MUX2"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in0" lib_name="B" size="1"/>
+      <port type="input" prefix="in1" lib_name="A" size="1"/>
+      <port type="input" prefix="sel" lib_name="S0" size="1"/>
+      <port type="output" prefix="out" lib_name="Y" size="1"/>
+    </circuit_model>
+    <circuit_model type="chan_wire" name="chan_segment" prefix="track_seg" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="101" C="22.5e-15" num_level="1"/> <!-- model_type could be T, res_val and cap_val DON'T CARE -->
+    </circuit_model>
+    <circuit_model type="wire" name="direct_interc" prefix="direct_interc" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="0" C="0" num_level="1"/> <!-- model_type could be T, res_val cap_val should be defined -->
+    </circuit_model>
+    <circuit_model type="mux" name="mux_tree" prefix="mux_tree" is_default="true" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="tree" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <pass_gate_logic circuit_model_name="MUX2"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <circuit_model type="mux" name="mux_tree_tapbuf" prefix="mux_tree_tapbuf" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="tree" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="tap_buf4"/>
+      <pass_gate_logic circuit_model_name="MUX2"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <!--DFF subckt ports should be defined as <D> <Q> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ff" name="static_dff" prefix="dff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="input" prefix="set" size="1" is_global="true" default_val="0" is_set="true"/>
+       <port type="input" prefix="reset" size="1" is_global="true" default_val="0" is_reset="true"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="clock" prefix="clk" size="1" is_global="true" default_val="0" />
+    </circuit_model>
+    <circuit_model type="lut" name="frac_lut6" prefix="frac_lut6" dump_structural_verilog="true">
+      <design_technology type="cmos" fracturable_lut="true"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_inverter exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_buffer exist="true" circuit_model_name="buf4"/>
+      <lut_intermediate_buffer exist="true" circuit_model_name="buf4" location_map="-1-1-"/>
+      <pass_gate_logic circuit_model_name="MUX2"/>
+      <port type="input" prefix="in" size="6" tri_state_map="-----1" circuit_model_name="OR2"/>
+      <port type="output" prefix="lut5_out" size="2" lut_frac_level="5" lut_output_mask="0,1"/>
+      <port type="output" prefix="lut6_out" size="1" lut_output_mask="0"/>
+      <port type="sram" prefix="sram" size="64"/>
+      <port type="sram" prefix="mode" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+    </circuit_model>
+    <!--Scan-chain DFF subckt ports should be defined as <D> <Q> <Qb> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ccff" name="sc_dff_compact" prefix="scff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="pReset" lib_name="reset" size="1" is_global="true" default_val="0" is_reset="true" is_prog="true"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="output" prefix="Qb" size="1"/>
+       <port type="clock" prefix="prog_clk" lib_name="clk" size="1" is_global="true" default_val="0" is_prog="true"/>
+    </circuit_model>
+    <circuit_model type="iopad" name="iopad" prefix="iopad" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/io.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/io.v">
+      <design_technology type="cmos"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <port type="inout" prefix="pad" size="1" is_global="true" is_io="true"/>
+      <port type="sram" prefix="en" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+      <port type="input" prefix="outpad" size="1"/>
+      <port type="output" prefix="inpad" size="1"/>
+    </circuit_model>
+  </circuit_library>
+  <configuration_protocol>
+    <organization type="scan_chain" circuit_model_name="sc_dff_compact"/>
+  </configuration_protocol>
+  <connection_block>
+    <switch name="ipin_cblock" circuit_model_name="mux_tree_tapbuf"/>
+  </connection_block>
+  <switch_block>
+    <switch name="0" circuit_model_name="mux_tree_tapbuf"/>
+  </switch_block>
+  <routing_segment>
+    <segment name="L4" circuit_model_name="chan_segment"/>
+  </routing_segment>
+  <pb_type_annotations>
+    <!-- physical pb_type binding in complex block IO -->
+    <pb_type name="io" physical_mode_name="physical" idle_mode_name="inpad"/>
+    <pb_type name="io[physical].iopad" circuit_model_name="iopad" mode_bits="1"/> 
+    <pb_type name="io[inpad].inpad" physical_pb_type_name="io[physical].iopad" mode_bits="1"/> 
+    <pb_type name="io[outpad].outpad" physical_pb_type_name="io[physical].iopad" mode_bits="0"/> 
+    <!-- End physical pb_type binding in complex block IO -->
+
+    <!-- physical pb_type binding in complex block CLB -->
+    <!-- physical mode will be the default mode if not specified -->
+    <pb_type name="clb">
+      <!-- Binding interconnect to circuit models as their physical implementation, if not defined, we use the default model -->
+      <interconnect name="crossbar" circuit_model_name="mux_tree"/>
+    </pb_type>
+    <pb_type name="clb.fle" physical_mode_name="physical"/>
+    <pb_type name="clb.fle[physical].fabric.frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="0"/>
+    <pb_type name="clb.fle[physical].fabric.ff" circuit_model_name="static_dff"/>
+    <!-- Binding operating pb_type to physical pb_type -->
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.lut5" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="1" physical_pb_type_index_factor="0.5">
+      <!-- Binding the lut5 to the first 5 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:4]"/>
+      <port name="out" physical_mode_port="lut5_out[0:0]" physical_mode_pin_rotate_offset="1"/>
+    </pb_type>
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.ff" physical_pb_type_name="clb.fle[physical].fabric.ff"/>
+    <pb_type name="clb.fle[n1_lut6].ble6.lut6" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="0">
+      <!-- Binding the lut6 to the first 6 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:5]"/>
+      <port name="out" physical_mode_port="lut6_out"/>
+    </pb_type>
+    <pb_type name="clb.fle[n1_lut6].ble6.ff" physical_pb_type_name="clb.fle[physical].fabric.ff" physical_pb_type_index_factor="2" physical_pb_type_index_offset="0"/>
+    <!-- End physical pb_type binding in complex block IO -->
+  </pb_type_annotations>
+</openfpga_architecture>
--- a/openfpga_flow/openfpga_arch/k6_frac_N8_tree_mux_40nm_openfpga.xml
+++ b/openfpga_flow/openfpga_arch/k6_frac_N8_tree_mux_40nm_openfpga.xml
@ -0,0 +1,222 @@
+<!-- Architecture annotation for OpenFPGA framework
+     This annotation supports the k6_N10_40nm.xml 
+     - General purpose logic block
+       - K = 6, N = 8, I = 40
+       - Single mode
+     - Routing architecture
+       - L = 4, fc_in = 0.15, fc_out = 0.1
+  -->
+<openfpga_architecture>
+  <technology_library>
+    <device_library>
+      <device_model name="logic" type="transistor">
+        <lib type="industry" corner="TOP_TT" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="0.9" pn_ratio="2"/>
+        <pmos name="pch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+        <nmos name="nch" chan_length="40e-9" min_width="140e-9" variation="logic_transistor_var"/>
+      </device_model>
+      <device_model name="io" type="transistor">
+        <lib type="academia" ref="M" path="${OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.pm"/>
+        <design vdd="2.5" pn_ratio="3"/>
+        <pmos name="pch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+        <nmos name="nch_25" chan_length="270e-9" min_width="320e-9" variation="io_transistor_var"/>
+      </device_model>
+    </device_library>
+    <variation_library>
+      <variation name="logic_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+      <variation name="io_transistor_var" abs_deviation="0.1" num_sigma="3"/>
+    </variation_library>
+  </technology_library>
+  <circuit_library>
+    <circuit_model type="inv_buf" name="INVTX1" prefix="INVTX1" is_default="true">
+      <design_technology type="cmos" topology="inverter" size="1"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="buf4" prefix="buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="2" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="inv_buf" name="tap_buf4" prefix="tap_buf4" is_default="false">
+      <design_technology type="cmos" topology="buffer" size="1" num_level="3" f_per_stage="4"/>
+      <device_technology device_model_name="logic"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in" out_port="out">
+        10e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="gate" name="OR2" prefix="OR2" is_default="true">
+      <design_technology type="cmos" topology="OR"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="a" size="1"/>
+      <port type="input" prefix="b" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="a b" out_port="out">
+        10e-12 5e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="pass_gate" name="MUX2" prefix="MUX2" is_default="true">
+      <design_technology type="cmos" topology="transmission_gate" nmos_size="1" pmos_size="2"/>
+      <device_technology device_model_name="logic"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="input" prefix="sel" size="1"/>
+      <port type="input" prefix="selb" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <delay_matrix type="rise" in_port="in sel selb" out_port="out">
+        10e-12 5e-12 5e-12
+      </delay_matrix>
+      <delay_matrix type="fall" in_port="in sel selb" out_port="out">
+        10e-12 5e-12 5e-12
+      </delay_matrix>
+    </circuit_model>
+    <circuit_model type="chan_wire" name="chan_segment" prefix="track_seg" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="101" C="22.5e-15" num_level="1"/> <!-- model_type could be T, res_val and cap_val DON'T CARE -->
+    </circuit_model>
+    <circuit_model type="wire" name="direct_interc" prefix="direct_interc" is_default="true">
+      <design_technology type="cmos"/>
+      <input_buffer exist="false"/>
+      <output_buffer exist="false"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <wire_param model_type="pi" R="0" C="0" num_level="1"/> <!-- model_type could be T, res_val cap_val should be defined -->
+    </circuit_model>
+    <circuit_model type="mux" name="mux_tree" prefix="mux_tree" is_default="true" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="tree" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <pass_gate_logic circuit_model_name="MUX2"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <circuit_model type="mux" name="mux_tree_tapbuf" prefix="mux_tree_tapbuf" dump_structural_verilog="true">
+      <design_technology type="cmos" structure="tree" add_const_input="true" const_input_val="1"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="tap_buf4"/>
+      <pass_gate_logic circuit_model_name="MUX2"/>
+      <port type="input" prefix="in" size="1"/>
+      <port type="output" prefix="out" size="1"/>
+      <port type="sram" prefix="sram" size="1"/>
+    </circuit_model>
+    <!--DFF subckt ports should be defined as <D> <Q> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ff" name="static_dff" prefix="dff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="input" prefix="set" size="1" is_global="true" default_val="0" is_set="true"/>
+       <port type="input" prefix="reset" size="1" is_global="true" default_val="0" is_reset="true"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="clock" prefix="clk" size="1" is_global="true" default_val="0" />
+    </circuit_model>
+    <circuit_model type="lut" name="frac_lut6" prefix="frac_lut6" dump_structural_verilog="true">
+      <design_technology type="cmos" fracturable_lut="true"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_inverter exist="true" circuit_model_name="INVTX1"/>
+      <lut_input_buffer exist="true" circuit_model_name="buf4"/>
+      <lut_intermediate_buffer exist="true" circuit_model_name="buf4" location_map="-1-1-"/>
+      <pass_gate_logic circuit_model_name="MUX2"/>
+      <port type="input" prefix="in" size="6" tri_state_map="-----1" circuit_model_name="OR2"/>
+      <port type="output" prefix="lut5_out" size="2" lut_frac_level="5" lut_output_mask="0,1"/>
+      <port type="output" prefix="lut6_out" size="1" lut_output_mask="0"/>
+      <port type="sram" prefix="sram" size="64"/>
+      <port type="sram" prefix="mode" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+    </circuit_model>
+    <!--Scan-chain DFF subckt ports should be defined as <D> <Q> <Qb> <CLK> <RESET> <SET>  -->
+    <circuit_model type="ccff" name="sc_dff_compact" prefix="scff" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/ff.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/ff.v">
+       <design_technology type="cmos"/>
+       <input_buffer exist="true" circuit_model_name="INVTX1"/>
+       <output_buffer exist="true" circuit_model_name="INVTX1"/>
+       <port type="input" prefix="pReset" lib_name="reset" size="1" is_global="true" default_val="0" is_reset="true" is_prog="true"/>
+       <port type="input" prefix="D" size="1"/>
+       <port type="output" prefix="Q" size="1"/>
+       <port type="output" prefix="Qb" size="1"/>
+       <port type="clock" prefix="prog_clk" lib_name="clk" size="1" is_global="true" default_val="0" is_prog="true"/>
+    </circuit_model>
+    <circuit_model type="iopad" name="iopad" prefix="iopad" spice_netlist="${OPENFPGA_PATH}/openfpga_flow/SpiceNetlists/io.sp" verilog_netlist="${OPENFPGA_PATH}/openfpga_flow/VerilogNetlists/io.v">
+      <design_technology type="cmos"/>
+      <input_buffer exist="true" circuit_model_name="INVTX1"/>
+      <output_buffer exist="true" circuit_model_name="INVTX1"/>
+      <port type="inout" prefix="pad" size="1" is_global="true" is_io="true"/>
+      <port type="sram" prefix="en" size="1" mode_select="true" circuit_model_name="sc_dff_compact" default_val="1"/>
+      <port type="input" prefix="outpad" size="1"/>
+      <port type="output" prefix="inpad" size="1"/>
+    </circuit_model>
+  </circuit_library>
+  <configuration_protocol>
+    <organization type="scan_chain" circuit_model_name="sc_dff_compact"/>
+  </configuration_protocol>
+  <connection_block>
+    <switch name="ipin_cblock" circuit_model_name="mux_tree_tapbuf"/>
+  </connection_block>
+  <switch_block>
+    <switch name="0" circuit_model_name="mux_tree_tapbuf"/>
+  </switch_block>
+  <routing_segment>
+    <segment name="L4" circuit_model_name="chan_segment"/>
+  </routing_segment>
+  <pb_type_annotations>
+    <!-- physical pb_type binding in complex block IO -->
+    <pb_type name="io" physical_mode_name="physical" idle_mode_name="inpad"/>
+    <pb_type name="io[physical].iopad" circuit_model_name="iopad" mode_bits="1"/> 
+    <pb_type name="io[inpad].inpad" physical_pb_type_name="io[physical].iopad" mode_bits="1"/> 
+    <pb_type name="io[outpad].outpad" physical_pb_type_name="io[physical].iopad" mode_bits="0"/> 
+    <!-- End physical pb_type binding in complex block IO -->
+
+    <!-- physical pb_type binding in complex block CLB -->
+    <!-- physical mode will be the default mode if not specified -->
+    <pb_type name="clb">
+      <!-- Binding interconnect to circuit models as their physical implementation, if not defined, we use the default model -->
+      <interconnect name="crossbar" circuit_model_name="mux_tree"/>
+    </pb_type>
+    <pb_type name="clb.fle" physical_mode_name="physical"/>
+    <pb_type name="clb.fle[physical].fabric.frac_logic.frac_lut6" circuit_model_name="frac_lut6" mode_bits="0"/>
+    <pb_type name="clb.fle[physical].fabric.ff" circuit_model_name="static_dff"/>
+    <!-- Binding operating pb_type to physical pb_type -->
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.lut5" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="1" physical_pb_type_index_factor="0.5">
+      <!-- Binding the lut5 to the first 5 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:4]"/>
+      <port name="out" physical_mode_port="lut5_out[0:0]" physical_mode_pin_rotate_offset="1"/>
+    </pb_type>
+    <pb_type name="clb.fle[n2_lut5].lut5inter.ble5.ff" physical_pb_type_name="clb.fle[physical].fabric.ff"/>
+    <pb_type name="clb.fle[n1_lut6].ble6.lut6" physical_pb_type_name="clb.fle[physical].fabric.frac_logic.frac_lut6" mode_bits="0">
+      <!-- Binding the lut6 to the first 6 inputs of fracturable lut6 -->
+      <port name="in" physical_mode_port="in[0:5]"/>
+      <port name="out" physical_mode_port="lut6_out"/>
+    </pb_type>
+    <pb_type name="clb.fle[n1_lut6].ble6.ff" physical_pb_type_name="clb.fle[physical].fabric.ff" physical_pb_type_index_factor="2" physical_pb_type_index_offset="0"/>
+    <!-- End physical pb_type binding in complex block IO -->
+  </pb_type_annotations>
+</openfpga_architecture>
--- a/openfpga_flow/scripts/run_fpga_flow.py
+++ b/openfpga_flow/scripts/run_fpga_flow.py
@ -419,6 +419,17 @@ def prepare_run_directory(run_dir):
        with open(args.openfpga_arch_file, 'w', encoding='utf-8') as archfile:
            archfile.write(tmpl.substitute(script_env_vars["PATH"]))

+    # Sanitize provided openshell template, if provided
+    if (args.openfpga_shell_template):
+        if not os.path.isfile(args.openfpga_shell_template or ""):
+            logger.error("Openfpga shell file - %s" %
+                         args.openfpga_shell_template)
+            clean_up_and_exit("Provided openfpga_shell_template" +
+                              f" {args.openfpga_shell_template} file not found")
+        else:
+            shutil.copy(args.openfpga_shell_template,
+                        args.top_module+"_template.openfpga")
+
    # Create benchmark dir in run_dir and copy flattern architecture file
    os.mkdir("benchmark")
    try:
@ -600,17 +611,6 @@ def collect_files_for_vpr():
        clean_up_and_exit("Provided base_verilog file not found")
    shutil.copy(args.base_verilog, args.top_module+"_output_verilog.v")

-    # Sanitize provided openshell template, if provided
-    if (args.openfpga_shell_template):
-        if not os.path.isfile(args.openfpga_shell_template or ""):
-            logger.error("Openfpga shell file - %s" %
-                         args.openfpga_shell_template)
-            clean_up_and_exit("Provided openfpga_shell_template" +
-                              f" {args.openfpga_shell_template} file not found")
-        else:
-            shutil.copy(args.openfpga_shell_template,
-                        args.top_module+"_template.openfpga")
-

 def run_vpr():
    ExecTime["VPRStart"] = time.time()
--- a/openfpga_flow/tasks/full_testbench/configuration_chain/config/task.conf
+++ b/openfpga_flow/tasks/full_testbench/configuration_chain/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_cc_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,14 +23,11 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
 end_flow_with_test=
-#vpr_fpga_verilog_formal_verification_top_netlist=
--- a/openfpga_flow/tasks/full_testbench/configuration_frame/config/task.conf
+++ b/openfpga_flow/tasks/full_testbench/configuration_frame/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_frame_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,14 +23,11 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
 end_flow_with_test=
-#vpr_fpga_verilog_formal_verification_top_netlist=
--- a/openfpga_flow/tasks/full_testbench/fast_configuration_chain/config/task.conf
+++ b/openfpga_flow/tasks/full_testbench/fast_configuration_chain/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_cc_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,14 +23,11 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
 end_flow_with_test=
-#vpr_fpga_verilog_formal_verification_top_netlist=
--- a/openfpga_flow/tasks/full_testbench/fast_configuration_frame/config/task.conf
+++ b/openfpga_flow/tasks/full_testbench/fast_configuration_frame/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_frame_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,14 +23,11 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
 end_flow_with_test=
-#vpr_fpga_verilog_formal_verification_top_netlist=
--- a/openfpga_flow/tasks/full_testbench/fast_memory_bank/config/task.conf
+++ b/openfpga_flow/tasks/full_testbench/fast_memory_bank/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_bank_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,14 +23,11 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
 end_flow_with_test=
-#vpr_fpga_verilog_formal_verification_top_netlist=
--- a/openfpga_flow/tasks/full_testbench/flatten_memory/config/task.conf
+++ b/openfpga_flow/tasks/full_testbench/flatten_memory/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_standalone_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,14 +23,11 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
 end_flow_with_test=
-#vpr_fpga_verilog_formal_verification_top_netlist=
--- a/openfpga_flow/tasks/full_testbench/memory_bank/config/task.conf
+++ b/openfpga_flow/tasks/full_testbench/memory_bank/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_bank_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,14 +23,11 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
 end_flow_with_test=
-#vpr_fpga_verilog_formal_verification_top_netlist=
--- a/openfpga_flow/tasks/generate_bitstream/config/task.conf
+++ b/openfpga_flow/tasks/generate_bitstream/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N10_40nm_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,11 +23,9 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N10_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
--- a/openfpga_flow/tasks/implicit_verilog/config/task.conf
+++ b/openfpga_flow/tasks/implicit_verilog/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N10_40nm_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,12 +23,10 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N10_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/counter/counter_post_yosys.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/counter/counter.v

 [SYNTHESIS_PARAM]
 bench0_top = counter
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/counter/counter_pre_vpr.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/counter/counter_output_verilog.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
--- a/openfpga_flow/tasks/lut_design/frac_lut/config/task.conf
+++ b/openfpga_flow/tasks/lut_design/frac_lut/config/task.conf
@ -15,12 +15,12 @@ spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
 fpga_flow=vpr_blif
-openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N10_40nm_openfpga.xml
+openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N8_40nm_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=

 [ARCHITECTURES]
-arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N10_tileable_40nm.xml
+arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N8_tileable_40nm.xml

 [BENCHMARKS]
 #
--- a/openfpga_flow/tasks/mux_design/local_encoder/config/task.conf
+++ b/openfpga_flow/tasks/mux_design/local_encoder/config/task.conf
@ -15,12 +15,12 @@ spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
 fpga_flow=vpr_blif
-openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N10_local_encoder_40nm_openfpga.xml
+openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N8_local_encoder_40nm_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=

 [ARCHITECTURES]
-arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N10_tileable_40nm.xml
+arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N8_tileable_40nm.xml

 [BENCHMARKS]
 bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
--- a/openfpga_flow/tasks/mux_design/stdcell_mux2/config/task.conf
+++ b/openfpga_flow/tasks/mux_design/stdcell_mux2/config/task.conf
@ -15,12 +15,12 @@ spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
 fpga_flow=vpr_blif
-openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N10_stdcell_mux_40nm_openfpga.xml
+openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N8_stdcell_mux_40nm_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=

 [ARCHITECTURES]
-arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N10_tileable_40nm.xml
+arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N8_tileable_40nm.xml

 [BENCHMARKS]
 bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
--- a/openfpga_flow/tasks/mux_design/tree_structure/config/task.conf
+++ b/openfpga_flow/tasks/mux_design/tree_structure/config/task.conf
@ -15,12 +15,12 @@ spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
 fpga_flow=vpr_blif
-openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N10_tree_mux_40nm_openfpga.xml
+openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k6_frac_N8_tree_mux_40nm_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=

 [ARCHITECTURES]
-arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N10_tileable_40nm.xml
+arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k6_frac_N8_tileable_40nm.xml

 [BENCHMARKS]
 bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
--- a/openfpga_flow/tasks/preconfig_testbench/configuration_chain/config/task.conf
+++ b/openfpga_flow/tasks/preconfig_testbench/configuration_chain/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_cc_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,12 +23,10 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
--- a/openfpga_flow/tasks/preconfig_testbench/configuration_frame/config/task.conf
+++ b/openfpga_flow/tasks/preconfig_testbench/configuration_frame/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_frame_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,12 +23,10 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
--- a/openfpga_flow/tasks/preconfig_testbench/flatten_memory/config/task.conf
+++ b/openfpga_flow/tasks/preconfig_testbench/flatten_memory/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_standalone_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,12 +23,10 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
--- a/openfpga_flow/tasks/preconfig_testbench/memory_bank/config/task.conf
+++ b/openfpga_flow/tasks/preconfig_testbench/memory_bank/config/task.conf
@ -14,7 +14,7 @@ power_analysis = true
 spice_output=false
 verilog_output=true
 timeout_each_job = 20*60
-fpga_flow=vpr_blif
+fpga_flow=yosys_vpr
 openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_N4_40nm_bank_openfpga.xml
 openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/auto_sim_openfpga.xml
 external_fabric_key_file=
@ -23,12 +23,10 @@ external_fabric_key_file=
 arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_N4_tileable_40nm.xml

 [BENCHMARKS]
-bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.blif
+bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v

 [SYNTHESIS_PARAM]
 bench0_top = and2
-bench0_act = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.act
-bench0_verilog = ${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/and2/and2.v
 bench0_chan_width = 300

 [SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
--- a/openfpga_flow/tasks/spypad/config/task.conf
+++ b/openfpga_flow/tasks/spypad/config/task.conf
@ -8,7 +8,7 @@

 [GENERAL]
 run_engine=openfpga_shell
-openfpga_shell_template=${PATH:OPENFPGA_PATH}/openfpga_flow/OpenFPGAShellScripts/example_script.openfpga
+openfpga_shell_template=${PATH:OPENFPGA_PATH}/openfpga_flow/OpenFPGAShellScripts/fast_configuration_example_script.openfpga
 power_tech_file = ${PATH:OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.xml
 power_analysis = true
 spice_output=false
--- a/openfpga_flow/vpr_arch/k6_frac_N8_tileable_40nm.xml
+++ b/openfpga_flow/vpr_arch/k6_frac_N8_tileable_40nm.xml
@ -0,0 +1,441 @@
+<!-- 
+  Flagship Heterogeneous Architecture (No Carry Chains) for VTR 7.0.
+
+  - 40 nm technology
+  - General purpose logic block: 
+    K = 6, N = 8, fracturable 6 LUTs (can operate as one 6-LUT or two 5-LUTs with all 5 inputs shared) 
+    with optionally registered outputs
+  - Routing architecture: L = 4, fc_in = 0.15, Fc_out = 0.1
+
+  Details on Modelling:
+
+  Based on flagship k6_frac_N10_mem32K_40nm.xml architecture.
+
+  Authors: Jason Luu, Jeff Goeders, Vaughn Betz
+-->
+<architecture>
+  <!-- 
+       ODIN II specific config begins 
+       Describes the types of user-specified netlist blocks (in blif, this corresponds to 
+       ".model [type_of_block]") that this architecture supports.
+
+       Note: Basic LUTs, I/Os, and flip-flops are not included here as there are 
+       already special structures in blif (.names, .input, .output, and .latch) 
+       that describe them.
+  -->
+  <models>
+    <!-- A virtual model for I/O to be used in the physical mode of io block -->
+    <model name="io">
+      <input_ports>
+        <port name="outpad"/>
+      </input_ports>
+      <output_ports>
+        <port name="inpad"/>
+      </output_ports>
+    </model>
+    <!-- A virtual model for I/O to be used in the physical mode of io block -->
+    <model name="frac_lut6">
+      <input_ports>
+        <port name="in"/>
+      </input_ports>
+      <output_ports>
+        <port name="lut5_out"/>
+        <port name="lut6_out"/>
+      </output_ports>
+    </model>
+  </models>
+  <tiles>
+    <!-- Do NOT add clock pins to I/O here!!! VPR does not build clock network in the way that OpenFPGA can support
+         If you need to register the I/O, define clocks in the circuit models
+         These clocks can be handled in back-end
+     -->
+    <tile name="io" capacity="8" area="0">
+      <equivalent_sites>
+        <site pb_type="io"/>
+      </equivalent_sites>
+      <input name="outpad" num_pins="1"/>
+      <output name="inpad" num_pins="1"/>
+      <fc in_type="frac" in_val="0.15" out_type="frac" out_val="0.10"/>
+      <pinlocations pattern="custom">
+        <loc side="left">io.outpad io.inpad</loc>
+        <loc side="top">io.outpad io.inpad</loc>
+        <loc side="right">io.outpad io.inpad</loc>
+        <loc side="bottom">io.outpad io.inpad</loc>
+      </pinlocations>
+    </tile>
+    <tile name="clb" area="53894">
+      <equivalent_sites>
+        <site pb_type="clb"/>
+      </equivalent_sites>
+      <input name="I" num_pins="32" equivalent="full"/>
+      <output name="O" num_pins="16" equivalent="none"/>
+      <clock name="clk" num_pins="1"/>
+      <fc in_type="frac" in_val="0.15" out_type="frac" out_val="0.10"/>
+      <pinlocations pattern="spread"/>
+    </tile>
+  </tiles>
+  <!-- ODIN II specific config ends -->
+  <!-- Physical descriptions begin -->
+  <layout tileable="true">
+    <!--auto_layout aspect_ratio="1.0"-->
+    <fixed_layout name="2x2" width="4" height="4">
+      <!--Perimeter of 'io' blocks with 'EMPTY' blocks at corners-->
+      <perimeter type="io" priority="100"/>
+      <corners type="EMPTY" priority="101"/>
+      <!--Fill with 'clb'-->
+      <fill type="clb" priority="10"/>
+    </fixed_layout>
+    <!-- /auto_layout -->
+  </layout>
+  <device>
+    <!-- VB & JL: Using Ian Kuon's transistor sizing and drive strength data for routing, at 40 nm. Ian used BPTM 
+			     models. We are modifying the delay values however, to include metal C and R, which allows more architecture
+			     experimentation. We are also modifying the relative resistance of PMOS to be 1.8x that of NMOS
+			     (vs. Ian's 3x) as 1.8x lines up with Jeff G's data from a 45 nm process (and is more typical of 
+			     45 nm in general). I'm upping the Rmin_nmos from Ian's just over 6k to nearly 9k, and dropping 
+			     RminW_pmos from 18k to 16k to hit this 1.8x ratio, while keeping the delays of buffers approximately
+			     lined up with Stratix IV. 
+			     We are using Jeff G.'s capacitance data for 45 nm (in tech/ptm_45nm).
+			     Jeff's tables list C in for transistors with widths in multiples of the minimum feature size (45 nm).
+			     The minimum contactable transistor is 2.5 * 45 nm, so I need to multiply drive strength sizes in this file
+	                     by 2.5x when looking up in Jeff's tables.
+			     The delay values are lined up with Stratix IV, which has an architecture similar to this
+			     proposed FPGA, and which is also 40 nm 
+			     C_ipin_cblock: input capacitance of a track buffer, which VPR assumes is a single-stage
+			     4x minimum drive strength buffer. -->
+    <sizing R_minW_nmos="8926" R_minW_pmos="16067"/>
+    <!-- The grid_logic_tile_area below will be used for all blocks that do not explicitly set their own (non-routing)
+     	  area; set to 0 since we explicitly set the area of all blocks currently in this architecture file.
+	  -->
+    <area grid_logic_tile_area="0"/>
+    <chan_width_distr>
+      <x distr="uniform" peak="1.000000"/>
+      <y distr="uniform" peak="1.000000"/>
+    </chan_width_distr>
+    <switch_block type="wilton" fs="3" sub_type="subset" sub_fs="3"/>
+    <connection_block input_switch_name="ipin_cblock"/>
+  </device>
+  <switchlist>
+    <!-- VB: the mux_trans_size and buf_size data below is in minimum width transistor *areas*, assuming the purple
+	       book area formula. This means the mux transistors are about 5x minimum drive strength.
+	       We assume the first stage of the buffer is 3x min drive strength to be reasonable given the large 
+	       mux transistors, and this gives a reasonable stage ratio of a bit over 5x to the second stage. We assume
+	       the n and p transistors in the first stage are equal-sized to lower the buffer trip point, since it's fed
+	       by a pass transistor mux. We can then reverse engineer the buffer second stage to hit the specified 
+	       buf_size (really buffer area) - 16.2x minimum drive nmos and 1.8*16.2 = 29.2x minimum drive.
+	       I then took the data from Jeff G.'s PTM modeling of 45 nm to get the Cin (gate of first stage) and Cout 
+	       (diff of second stage) listed below.  Jeff's models are in tech/ptm_45nm, and are in min feature multiples.
+	       The minimum contactable transistor is 2.5 * 45 nm, so I need to multiply the drive strength sizes above by 
+	       2.5x when looking up in Jeff's tables.
+	       Finally, we choose a switch delay (58 ps) that leads to length 4 wires having a delay equal to that of SIV of 126 ps.
+	       This also leads to the switch being 46% of the total wire delay, which is reasonable. -->
+    <switch type="mux" name="0" R="551" Cin=".77e-15" Cout="4e-15" Tdel="58e-12" mux_trans_size="2.630740" buf_size="27.645901"/>
+    <!--switch ipin_cblock resistance set to yeild for 4x minimum drive strength buffer-->
+    <switch type="mux" name="ipin_cblock" R="2231.5" Cout="0." Cin="1.47e-15" Tdel="7.247000e-11" mux_trans_size="1.222260" buf_size="auto"/>
+  </switchlist>
+  <segmentlist>
+    <!--- VB & JL: using ITRS metal stack data, 96 nm half pitch wires, which are intermediate metal width/space.  
+			     With the 96 nm half pitch, such wires would take 60 um of height, vs. a 90 nm high (approximated as square) Stratix IV tile so this seems
+			     reasonable. Using a tile length of 90 nm, corresponding to the length of a Stratix IV tile if it were square. -->
+    <!-- GIVE a specific name for the segment! OpenFPGA appreciate that! -->
+    <segment name="L4" freq="1.000000" length="4" type="unidir" Rmetal="101" Cmetal="22.5e-15">
+      <mux name="0"/>
+      <sb type="pattern">1 1 1 1 1</sb>
+      <cb type="pattern">1 1 1 1</cb>
+    </segment>
+  </segmentlist>
+  <complexblocklist>
+    <!-- Define I/O pads begin -->
+    <!-- Capacity is a unique property of I/Os, it is the maximum number of I/Os that can be placed at the same (X,Y) location on the FPGA -->
+    <!-- Not sure of the area of an I/O (varies widely), and it's not relevant to the design of the FPGA core, so we're setting it to 0. -->
+    <pb_type name="io">
+      <input name="outpad" num_pins="1"/>
+      <output name="inpad" num_pins="1"/>
+      <!-- Do NOT add clock pins to I/O here!!! VPR does not build clock network in the way that OpenFPGA can support
+           If you need to register the I/O, define clocks in the circuit models
+           These clocks can be handled in back-end
+       -->
+      <!-- A mode denotes the physical implementation of an I/O 
+           This mode will be not packable but is mainly used for fabric verilog generation   
+        -->
+      <mode name="physical" disabled_in_pack="true">
+        <pb_type name="iopad" blif_model=".subckt io" num_pb="1">
+          <input name="outpad" num_pins="1"/>
+          <output name="inpad" num_pins="1"/>
+        </pb_type>
+        <interconnect>
+          <direct name="outpad" input="io.outpad" output="iopad.outpad">
+            <delay_constant max="1.394e-11" in_port="io.outpad" out_port="iopad.outpad"/>
+          </direct>
+          <direct name="inpad" input="iopad.inpad" output="io.inpad">
+            <delay_constant max="4.243e-11" in_port="iopad.inpad" out_port="io.inpad"/>
+          </direct>
+        </interconnect>
+      </mode>
+
+      <!-- IOs can operate as either inputs or outputs.
+	     Delays below come from Ian Kuon. They are small, so they should be interpreted as
+	     the delays to and from registers in the I/O (and generally I/Os are registered 
+	     today and that is when you timing analyze them.
+	     -->
+      <mode name="inpad">
+        <pb_type name="inpad" blif_model=".input" num_pb="1">
+          <output name="inpad" num_pins="1"/>
+        </pb_type>
+        <interconnect>
+          <direct name="inpad" input="inpad.inpad" output="io.inpad">
+            <delay_constant max="4.243e-11" in_port="inpad.inpad" out_port="io.inpad"/>
+          </direct>
+        </interconnect>
+      </mode>
+      <mode name="outpad">
+        <pb_type name="outpad" blif_model=".output" num_pb="1">
+          <input name="outpad" num_pins="1"/>
+        </pb_type>
+        <interconnect>
+          <direct name="outpad" input="io.outpad" output="outpad.outpad">
+            <delay_constant max="1.394e-11" in_port="io.outpad" out_port="outpad.outpad"/>
+          </direct>
+        </interconnect>
+      </mode>
+      <!-- Every input pin is driven by 15% of the tracks in a channel, every output pin is driven by 10% of the tracks in a channel -->
+      <!-- IOs go on the periphery of the FPGA, for consistency, 
+          make it physically equivalent on all sides so that only one definition of I/Os is needed.
+          If I do not make a physically equivalent definition, then I need to define 4 different I/Os, one for each side of the FPGA
+        -->
+      <!-- Place I/Os on the sides of the FPGA -->
+      <power method="ignore"/>
+    </pb_type>
+    <!-- Define I/O pads ends -->
+    <!-- Define general purpose logic block (CLB) begin -->
+    <!--- Area calculation: Total Stratix IV tile area is about 8100 um^2, and a minimum width transistor 
+	   area is 60 L^2 yields a tile area of 84375 MWTAs.
+	   Routing at W=300 is 30481 MWTAs, leaving us with a total of 53000 MWTAs for logic block area 
+	   This means that only 37% of our area is in the general routing, and 63% is inside the logic
+	   block. Note that the crossbar / local interconnect is considered part of the logic block
+	   area in this analysis. That is a lower proportion of of routing area than most academics
+	   assume, but note that the total routing area really includes the crossbar, which would push
+	   routing area up significantly, we estimate into the ~70% range. 
+	   -->
+    <pb_type name="clb">
+      <input name="I" num_pins="32" equivalent="full"/>
+      <output name="O" num_pins="16" equivalent="none"/>
+      <clock name="clk" num_pins="1"/>
+      <!-- Describe fracturable logic element.  
+             Each fracturable logic element has a 6-LUT that can alternatively operate as two 5-LUTs with shared inputs. 
+             The outputs of the fracturable logic element can be optionally registered
+        -->
+      <pb_type name="fle" num_pb="8">
+        <input name="in" num_pins="6"/>
+        <output name="out" num_pins="2"/>
+        <clock name="clk" num_pins="1"/>
+        <!-- Physical mode definition begin (physical implementation of the fle) -->
+        <mode name="physical" disabled_in_pack="true">
+          <pb_type name="fabric" num_pb="1">
+            <input name="in" num_pins="6"/>
+            <output name="out" num_pins="2"/>
+            <clock name="clk" num_pins="1"/>
+            <pb_type name="frac_logic" num_pb="1">
+              <input name="in" num_pins="6"/>
+              <output name="out" num_pins="2"/>
+              <!-- Define LUT -->
+              <pb_type name="frac_lut6" blif_model=".subckt frac_lut6" num_pb="1">
+                <input name="in" num_pins="6"/>
+                <output name="lut5_out" num_pins="2"/>
+                <output name="lut6_out" num_pins="1"/>
+              </pb_type>
+              <interconnect>
+                <direct name="direct1" input="frac_logic.in" output="frac_lut6.in"/>
+                <direct name="direct2" input="frac_lut6.lut5_out[1]" output="frac_logic.out[1]"/>
+                <!-- Xifan Tang: I use out[0] because the output of lut6 in lut6 mode is wired to the out[0] -->
+                <mux name="mux1" input="frac_lut6.lut6_out frac_lut6.lut5_out[0]" output="frac_logic.out[0]"/>
+              </interconnect>
+            </pb_type>
+            <!-- Define flip-flop -->
+            <pb_type name="ff" blif_model=".latch" num_pb="2" class="flipflop">
+              <input name="D" num_pins="1" port_class="D"/>
+              <output name="Q" num_pins="1" port_class="Q"/>
+              <clock name="clk" num_pins="1" port_class="clock"/>
+              <T_setup value="66e-12" port="ff.D" clock="clk"/>
+              <T_clock_to_Q max="124e-12" port="ff.Q" clock="clk"/>
+            </pb_type>
+            <interconnect>
+              <direct name="direct1" input="fabric.in" output="frac_logic.in"/>
+              <direct name="direct2" input="frac_logic.out[1:0]" output="ff[1:0].D"/>
+              <complete name="direct3" input="fabric.clk" output="ff[1:0].clk"/>
+              <mux name="mux1" input="ff[0].Q frac_logic.out[0]" output="fabric.out[0]">
+                <!-- LUT to output is faster than FF to output on a Stratix IV -->
+                <delay_constant max="25e-12" in_port="frac_logic.out[0]" out_port="fabric.out[0]"/>
+                <delay_constant max="45e-12" in_port="ff[0].Q" out_port="fabric.out[0]"/>
+              </mux>
+              <mux name="mux2" input="ff[1].Q frac_logic.out[1]" output="fabric.out[1]">
+                <!-- LUT to output is faster than FF to output on a Stratix IV -->
+                <delay_constant max="25e-12" in_port="frac_logic.out[1]" out_port="fabric.out[1]"/>
+                <delay_constant max="45e-12" in_port="ff[1].Q" out_port="fabric.out[1]"/>
+              </mux>
+            </interconnect>
+          </pb_type>
+          <interconnect>
+            <direct name="direct1" input="fle.in" output="fabric.in"/>
+            <direct name="direct2" input="fabric.out" output="fle.out"/>
+            <direct name="direct3" input="fle.clk" output="fabric.clk"/>
+          </interconnect>
+        </mode>
+        <!-- Physical mode definition end (physical implementation of the fle) -->
+        <!-- Dual 5-LUT mode definition begin -->
+        <mode name="n2_lut5">
+          <pb_type name="lut5inter" num_pb="1">
+            <input name="in" num_pins="5"/>
+            <output name="out" num_pins="2"/>
+            <clock name="clk" num_pins="1"/>
+            <pb_type name="ble5" num_pb="2">
+              <input name="in" num_pins="5"/>
+              <output name="out" num_pins="1"/>
+              <clock name="clk" num_pins="1"/>
+              <!-- Define the LUT -->
+              <pb_type name="lut5" blif_model=".names" num_pb="1" class="lut">
+                <input name="in" num_pins="5" port_class="lut_in"/>
+                <output name="out" num_pins="1" port_class="lut_out"/>
+                <!-- LUT timing using delay matrix -->
+                <!-- These are the physical delay inputs on a Stratix IV LUT but because VPR cannot do LUT rebalancing,
+                           we instead take the average of these numbers to get more stable results
+                      82e-12
+                      173e-12
+                      261e-12
+                      263e-12
+                      398e-12
+                      -->
+                <delay_matrix type="max" in_port="lut5.in" out_port="lut5.out">
+                  235e-12
+                  235e-12
+                  235e-12
+                  235e-12
+                  235e-12
+                </delay_matrix>
+              </pb_type>
+              <!-- Define the flip-flop -->
+              <pb_type name="ff" blif_model=".latch" num_pb="1" class="flipflop">
+                <input name="D" num_pins="1" port_class="D"/>
+                <output name="Q" num_pins="1" port_class="Q"/>
+                <clock name="clk" num_pins="1" port_class="clock"/>
+                <T_setup value="66e-12" port="ff.D" clock="clk"/>
+                <T_clock_to_Q max="124e-12" port="ff.Q" clock="clk"/>
+              </pb_type>
+              <interconnect>
+                <direct name="direct1" input="ble5.in[4:0]" output="lut5[0:0].in[4:0]"/>
+                <direct name="direct2" input="lut5[0:0].out" output="ff[0:0].D">
+                  <!-- Advanced user option that tells CAD tool to find LUT+FF pairs in netlist -->
+                  <pack_pattern name="ble5" in_port="lut5[0:0].out" out_port="ff[0:0].D"/>
+                </direct>
+                <direct name="direct3" input="ble5.clk" output="ff[0:0].clk"/>
+                <mux name="mux1" input="ff[0:0].Q lut5.out[0:0]" output="ble5.out[0:0]">
+                  <!-- LUT to output is faster than FF to output on a Stratix IV -->
+                  <delay_constant max="25e-12" in_port="lut5.out[0:0]" out_port="ble5.out[0:0]"/>
+                  <delay_constant max="45e-12" in_port="ff[0:0].Q" out_port="ble5.out[0:0]"/>
+                </mux>
+              </interconnect>
+            </pb_type>
+            <interconnect>
+              <direct name="direct1" input="lut5inter.in" output="ble5[0:0].in"/>
+              <direct name="direct2" input="lut5inter.in" output="ble5[1:1].in"/>
+              <direct name="direct3" input="ble5[1:0].out" output="lut5inter.out"/>
+              <complete name="complete1" input="lut5inter.clk" output="ble5[1:0].clk"/>
+            </interconnect>
+          </pb_type>
+          <interconnect>
+            <direct name="direct1" input="fle.in[4:0]" output="lut5inter.in"/>
+            <direct name="direct2" input="lut5inter.out" output="fle.out"/>
+            <direct name="direct3" input="fle.clk" output="lut5inter.clk"/>
+          </interconnect>
+        </mode>
+        <!-- Dual 5-LUT mode definition end -->
+        <!-- 6-LUT mode definition begin -->
+        <mode name="n1_lut6">
+          <!-- Define 6-LUT mode -->
+          <pb_type name="ble6" num_pb="1">
+            <input name="in" num_pins="6"/>
+            <output name="out" num_pins="1"/>
+            <clock name="clk" num_pins="1"/>
+            <!-- Define LUT -->
+            <pb_type name="lut6" blif_model=".names" num_pb="1" class="lut">
+              <input name="in" num_pins="6" port_class="lut_in"/>
+              <output name="out" num_pins="1" port_class="lut_out"/>
+              <!-- LUT timing using delay matrix -->
+              <!-- These are the physical delay inputs on a Stratix IV LUT but because VPR cannot do LUT rebalancing,
+                       we instead take the average of these numbers to get more stable results
+                  82e-12
+                  173e-12
+                  261e-12
+                  263e-12
+                  398e-12
+                  397e-12
+                  -->
+              <delay_matrix type="max" in_port="lut6.in" out_port="lut6.out">
+                261e-12
+                261e-12
+                261e-12
+                261e-12
+                261e-12
+                261e-12
+              </delay_matrix>
+            </pb_type>
+            <!-- Define flip-flop -->
+            <pb_type name="ff" blif_model=".latch" num_pb="1" class="flipflop">
+              <input name="D" num_pins="1" port_class="D"/>
+              <output name="Q" num_pins="1" port_class="Q"/>
+              <clock name="clk" num_pins="1" port_class="clock"/>
+              <T_setup value="66e-12" port="ff.D" clock="clk"/>
+              <T_clock_to_Q max="124e-12" port="ff.Q" clock="clk"/>
+            </pb_type>
+            <interconnect>
+              <direct name="direct1" input="ble6.in" output="lut6[0:0].in"/>
+              <direct name="direct2" input="lut6.out" output="ff.D">
+                <!-- Advanced user option that tells CAD tool to find LUT+FF pairs in netlist -->
+                <pack_pattern name="ble6" in_port="lut6.out" out_port="ff.D"/>
+              </direct>
+              <direct name="direct3" input="ble6.clk" output="ff.clk"/>
+              <mux name="mux1" input="ff.Q lut6.out" output="ble6.out">
+                <!-- LUT to output is faster than FF to output on a Stratix IV -->
+                <delay_constant max="25e-12" in_port="lut6.out" out_port="ble6.out"/>
+                <delay_constant max="45e-12" in_port="ff.Q" out_port="ble6.out"/>
+              </mux>
+            </interconnect>
+          </pb_type>
+          <interconnect>
+            <direct name="direct1" input="fle.in" output="ble6.in"/>
+            <direct name="direct2" input="ble6.out" output="fle.out[0:0]"/>
+            <direct name="direct3" input="fle.clk" output="ble6.clk"/>
+          </interconnect>
+        </mode>
+        <!-- 6-LUT mode definition end -->
+      </pb_type>
+      <interconnect>
+        <!-- We use a full crossbar to get logical equivalence at inputs of CLB 
+		     The delays below come from Stratix IV. the delay through a connection block
+		     input mux + the crossbar in Stratix IV is 167 ps. We already have a 72 ps 
+		     delay on the connection block input mux (modeled by Ian Kuon), so the remaining
+		     delay within the crossbar is 95 ps. 
+		     The delays of cluster feedbacks in Stratix IV is 100 ps, when driven by a LUT.
+		     Since all our outputs LUT outputs go to a BLE output, and have a delay of 
+		     25 ps to do so, we subtract 25 ps from the 100 ps delay of a feedback
+		     to get the part that should be marked on the crossbar.	 -->
+        <complete name="crossbar" input="clb.I fle[7:0].out" output="fle[7:0].in">
+          <delay_constant max="95e-12" in_port="clb.I" out_port="fle[7:0].in"/>
+          <delay_constant max="75e-12" in_port="fle[7:0].out" out_port="fle[7:0].in"/>
+        </complete>
+        <complete name="clks" input="clb.clk" output="fle[7:0].clk">
+        </complete>
+        <!-- This way of specifying direct connection to clb outputs is important because this architecture uses automatic spreading of opins.  
+               By grouping to output pins in this fashion, if a logic block is completely filled by 6-LUTs, 
+               then the outputs those 6-LUTs take get evenly distributed across all four sides of the CLB instead of clumped on two sides (which is what happens with a more
+               naive specification).
+          -->
+        <direct name="clbouts1" input="fle[7:0].out[0:0]" output="clb.O[7:0]"/>
+        <direct name="clbouts2" input="fle[7:0].out[1:1]" output="clb.O[15:8]"/>
+      </interconnect>
+      <!-- Every input pin is driven by 15% of the tracks in a channel, every output pin is driven by 10% of the tracks in a channel -->
+      <!-- Place this general purpose logic block in any unspecified column -->
+    </pb_type>
+    <!-- Define general purpose logic block (CLB) ends -->
+  </complexblocklist>
+</architecture>