riscv
/
OpenFPGA
mirror of https://github.com/lnis-uofu/OpenFPGA.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Merge pull request #1006 from lnis-uofu/local_rst 

Local Reset Support
This commit is contained in:
tangxifan 2023-01-19 16:01:42 -08:00 committed by GitHub
parent b69be407ca 693404d1ac
commit 3667dd5e65
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
14 changed files with 834 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

75
openfpga/src/repack/repack.cpp
View File

@ -304,6 +304,70 @@ static std::vector<int> find_pb_route_by_atom_net(
return pb_route_indices;
}
/***************************************************************************************
* This function will find the actual routing traces of the demanded net
* There is a specific search space applied when searching the routing traces:
* - ONLY applicable to the pb_pin of top-level pb_graph_node
* - First-tier candidates are in the same port of the source pin
* - If nothing is found in first-tier, we find expand the range by considering
*all the pins in the same type that are available at the top-level
*pb_graph_node
* - Exclude all the pb_route that is from a net on a pin which should be
*ignored
***************************************************************************************/
static std::vector<int> find_pb_route_by_atom_net_exclude_blacklist(
const t_pb* pb, const t_pb_graph_pin* source_pb_pin,
const AtomNetId& atom_net_id,
const std::map<AtomNetId, bool>& ignored_atom_nets,
const RepackOption& options) {
VTR_ASSERT(true == source_pb_pin->parent_node->is_root());
std::vector<int> pb_route_indices;
auto result = ignored_atom_nets.find(atom_net_id);
std::vector<int> candidate_pool;
for (int pin = 0; pin < pb->pb_graph_node->total_pb_pins; ++pin) {
/* Bypass unused pins */
if ((0 == pb->pb_route.count(pin)) ||
(AtomNetId::INVALID() == pb->pb_route.at(pin).atom_net_id)) {
continue;
}
/* Get the driver pb pin id, it must be valid */
if (atom_net_id != pb->pb_route.at(pin).atom_net_id) {
continue;
}
BasicPort curr_pin(
std::string(pb->pb_route.at(pin).pb_graph_pin->port->name),
pb->pb_route.at(pin).pb_graph_pin->pin_number,
pb->pb_route.at(pin).pb_graph_pin->pin_number);
if (result != ignored_atom_nets.end() && result->second &&
options.is_pin_ignore_global_nets(
std::string(pb->pb_graph_node->pb_type->name), curr_pin)) {
continue;
}
candidate_pool.push_back(pin);
}
for (int pin : candidate_pool) {
if (source_pb_pin->port == pb->pb_route.at(pin).pb_graph_pin->port) {
pb_route_indices.push_back(pin);
}
}
if (pb_route_indices.empty()) {
for (int pin : candidate_pool) {
if (pb->pb_route.at(pin).pb_graph_pin->parent_node->is_root() &&
is_pb_graph_pins_share_interc(source_pb_pin,
pb->pb_route.at(pin).pb_graph_pin)) {
pb_route_indices.push_back(pin);
}
}
}
return pb_route_indices;
}
/***************************************************************************************
* This function will find the actual source_pb_pin that is mapped by packed in
*the pb route As the inputs of clustered block may be renamed during routing,
@ -681,13 +745,16 @@ static void add_lb_router_nets(
*/
std::vector<int> pb_route_indices;
if (design_constraints.unconstrained_net(constrained_net_name)) {
VTR_LOGV(verbose,
"Search remapped routing traces for the unconstrained net\n");
pb_route_indices = find_pb_route_remapped_source_pb_pin(
pb, source_pb_pin, atom_net_id_to_route);
} else {
VTR_ASSERT_SAFE(
!design_constraints.unconstrained_net(constrained_net_name));
pb_route_indices =
find_pb_route_by_atom_net(pb, source_pb_pin, atom_net_id_to_route);
/* If this is a constrained net but the source pin is not the pin that the
* net is constrained to, w*/
VTR_LOGV(verbose, "Search routing traces for the constrained net\n");
pb_route_indices = find_pb_route_by_atom_net_exclude_blacklist(
pb, source_pb_pin, atom_net_id_to_route, ignored_atom_nets, options);
}
/* It could happen that the constrained net is NOT used in this clb, we just
* skip it for routing For example, a clkB net is never mapped to any ports

11
openfpga_flow/benchmarks/micro_benchmark/two_dff_inv_rst/two_dff_inv_rst.blif Normal file
View File

@ -0,0 +1,11 @@
.model two_dff_inv_rst
.inputs clk_i rst_i d_i
.outputs d0_o d1_o
.names rst_i int_rst
0 1
.subckt dffr D=d_i Q=d0_o C=clk_i R=rst_i
.subckt dffr D=d_i Q=d1_o C=clk_i R=int_rst
.end

41
openfpga_flow/benchmarks/micro_benchmark/two_dff_inv_rst/two_dff_inv_rst.v Normal file
View File

@ -0,0 +1,41 @@
/////////////////////////////////////////
// Functionality: Two FFs share a differential pair of reset. One is controlled by an inversion of the reset, while the other is controlled by the original reset. This is useful to validate
// - reset signal to drive LUT from global network
// - reset signal generated by internal programmable resources
// Author: Xifan Tang
////////////////////////////////////////
`timescale 1ns / 1ps
module two_dff_inv_rst(
clk_i,
rst_i,
d_i,
d0_o, d1_o);
input wire clk_i;
input wire rst_i;
input wire d_i;
output reg d0_o;
output reg d1_o;
wire int_rst;
assign int_rst = ~rst_i;
always @(posedge clk_i or posedge rst_i) begin
if (rst_i) begin
d0_o <= 0;
end else begin
d0_o <= d_i;
end
end
always @(posedge clk_i or posedge int_rst) begin
if (int_rst) begin
d1_o <= 0;
end else begin
d1_o <= d_i;
end
end
endmodule

2
openfpga_flow/openfpga_shell_scripts/ignore_global_nets_on_pins_example_script.openfpga
View File

@ -1,6 +1,6 @@
# Run VPR for the 'and' design
#--write_rr_graph example_rr_graph.xml
vpr ${VPR_ARCH_FILE} ${VPR_TESTBENCH_BLIF} --clock_modeling ideal
vpr ${VPR_ARCH_FILE} ${VPR_TESTBENCH_BLIF} --clock_modeling ideal --device ${OPENFPGA_VPR_DEVICE_LAYOUT}
# Read OpenFPGA architecture definition
read_openfpga_arch -f ${OPENFPGA_ARCH_FILE}

2
openfpga_flow/regression_test_scripts/basic_reg_test.sh
View File

@ -144,6 +144,8 @@ run-task basic_tests/k4_series/k4n4_rstOnLut $@
run-task basic_tests/k4_series/k4n4_rstOnLut_strong $@
echo -e "Testing K4N4 support clock generation by internal resources";
run-task basic_tests/k4_series/k4n4_clk_gen $@
echo -e "Testing K4N4 support reset generation by internal resources";
run-task basic_tests/k4_series/k4n4_rst_gen $@
echo -e "Testing different tile organizations";
echo -e "Testing tiles with pins only on top and left sides";

1
openfpga_flow/tasks/basic_tests/k4_series/k4n4_rstOnLut/config/task.conf
View File

@ -20,6 +20,7 @@ openfpga_shell_template=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_shell_scrip
openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_frac_N4_fracff_40nm_cc_openfpga.xml
openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/fixed_sim_openfpga.xml
openfpga_repack_design_constraint_file=${PATH:TASK_DIR}/config/repack_design_constraints.xml
openfpga_vpr_device_layout=2x2
[ARCHITECTURES]
arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_frac_N4_tileable_fracff_rstOnLut_40nm.xml

1
openfpga_flow/tasks/basic_tests/k4_series/k4n4_rstOnLut_strong/config/task.conf
View File

@ -20,6 +20,7 @@ openfpga_shell_template=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_shell_scrip
openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_frac_N4_fracff_40nm_registerable_io_cc_openfpga.xml
openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/fixed_sim_openfpga.xml
openfpga_repack_design_constraint_file=${PATH:TASK_DIR}/config/repack_design_constraints.xml
openfpga_vpr_device_layout=2x2
[ARCHITECTURES]
arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_frac_N4_tileable_fracff_rstOnLut_registerable_io_40nm.xml

5
openfpga_flow/tasks/basic_tests/k4_series/k4n4_rst_gen/config/pin_constraints.xml Normal file
View File

@ -0,0 +1,5 @@
<pin_constraints>
<set_io pin="op_clk[0]" net="clk_i"/>
<set_io pin="op_reset[0]" net="rst_i"/>
</pin_constraints>

6
openfpga_flow/tasks/basic_tests/k4_series/k4n4_rst_gen/config/repack_design_constraints.xml Normal file
View File

@ -0,0 +1,6 @@
<repack_design_constraints>
<pin_constraint pb_type="clb" pin="clk[0]" net="clk_i"/>
<pin_constraint pb_type="clb" pin="reset[0]" net="rst_i"/>
<!-- Leave lreset unconstrained as it may be mapped to any internal reset signals -->
</repack_design_constraints>

5
openfpga_flow/tasks/basic_tests/k4_series/k4n4_rst_gen/config/rst_on_lut_pc.xml Normal file
View File

@ -0,0 +1,5 @@
<pin_constraints>
<set_io pin="op_clk[0]" net="clk"/>
<set_io pin="op_reset[0]" net="rst"/>
</pin_constraints>

6
openfpga_flow/tasks/basic_tests/k4_series/k4n4_rst_gen/config/rst_on_lut_repack_dc.xml Normal file
View File

@ -0,0 +1,6 @@
<repack_design_constraints>
<pin_constraint pb_type="clb" pin="clk[0]" net="clk"/>
<pin_constraint pb_type="clb" pin="reset[0]" net="rst"/>
<!-- Leave lreset unconstrained as it may be mapped to any internal reset signals -->
</repack_design_constraints>

50
openfpga_flow/tasks/basic_tests/k4_series/k4n4_rst_gen/config/task.conf Normal file
View File

@ -0,0 +1,50 @@
# = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
# Configuration file for running experiments
# = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
# timeout_each_job : FPGA Task script splits fpga flow into multiple jobs
# Each job execute fpga_flow script on combination of architecture & benchmark
# timeout_each_job is timeout for each job
# = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =
[GENERAL]
run_engine=openfpga_shell
power_tech_file = ${PATH:OPENFPGA_PATH}/openfpga_flow/tech/PTM_45nm/45nm.xml
power_analysis = false
spice_output=false
verilog_output=true
timeout_each_job = 3*60
fpga_flow=yosys_vpr
[OpenFPGA_SHELL]
openfpga_shell_template=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_shell_scripts/ignore_global_nets_on_pins_example_script.openfpga
openfpga_arch_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_arch/k4_frac_N4_fracff_40nm_cc_openfpga.xml
openfpga_sim_setting_file=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_simulation_settings/fixed_sim_openfpga.xml
openfpga_repack_design_constraint_file=${PATH:TASK_DIR}/config/repack_design_constraints.xml
openfpga_vpr_device_layout=2x2
[ARCHITECTURES]
arch0=${PATH:OPENFPGA_PATH}/openfpga_flow/vpr_arch/k4_frac_N4_tileable_fracff_localRstGen_40nm.xml
[BENCHMARKS]
bench0=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/two_dff_inv_rst/two_dff_inv_rst.v
bench1=${PATH:OPENFPGA_PATH}/openfpga_flow/benchmarks/micro_benchmark/rst_on_lut/rst_on_lut.v
[SYNTHESIS_PARAM]
# Yosys script parameters
bench_yosys_cell_sim_verilog_common=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_yosys_techlib/openfpga_dff_sim.v
bench_yosys_dff_map_verilog_common=${PATH:OPENFPGA_PATH}/openfpga_flow/openfpga_yosys_techlib/openfpga_dff_map.v
bench_read_verilog_options_common = -nolatches
bench_yosys_common=${PATH:OPENFPGA_PATH}/openfpga_flow/misc/ys_tmpl_yosys_vpr_dff_flow.ys
bench_yosys_rewrite_common=${PATH:OPENFPGA_PATH}/openfpga_flow/misc/ys_tmpl_yosys_vpr_flow_with_rewrite.ys;${PATH:OPENFPGA_PATH}/openfpga_flow/misc/ys_tmpl_rewrite_flow.ys
bench0_top = two_dff_inv_rst
bench0_openfpga_pin_constraints_file = ${PATH:TASK_DIR}/config/pin_constraints.xml
bench0_openfpga_repack_design_constraint_file=${PATH:TASK_DIR}/config/repack_design_constraints.xml
bench1_top = rst_on_lut
bench1_openfpga_pin_constraints_file = ${PATH:TASK_DIR}/config/rst_on_lut_pc.xml
bench1_openfpga_repack_design_constraint_file=${PATH:TASK_DIR}/config/rst_on_lut_repack_dc.xml
[SCRIPT_PARAM_MIN_ROUTE_CHAN_WIDTH]
end_flow_with_test=
vpr_fpga_verilog_formal_verification_top_netlist=

1
openfpga_flow/vpr_arch/README.md
View File

@ -24,6 +24,7 @@ Please reveal the following architecture features in the names to help quickly s
- CustomIoLoc: Use OpenFPGA's extended custom I/O location syntax
- rstOnLut: The reset signal of CLB can feed LUT inputs through a local routing architecture
- localClkGen: The clock signal of CLB can be generated by internal programmable resources
- localRstGen: The reset signal of CLB can be generated by internal programmable resources
- <feature\_size>: The technology node which the delay numbers are extracted from.
- TileOrgz<Type>: How tile is organized.
* Top-left (Tl): the pins of a tile are placed on the top side and left side only

633
openfpga_flow/vpr_arch/k4_frac_N4_tileable_fracff_localRstGen_40nm.xml Normal file
View File

@ -0,0 +1,633 @@
<?xml version="1.0"?>
<!--
Flagship Heterogeneous Architecture (No Carry Chains) for VTR 7.0.
- 40 nm technology
- General purpose logic block:
K = 4, N = 4, fracturable 4 LUTs (can operate as one 4-LUT or two 3-LUTs with all 3 inputs shared)
with optionally registered outputs
- Routing architecture: L = 4, fc_in = 0.15, Fc_out = 0.1
Authors: Xifan Tang
-->
<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_lut4">
<input_ports>
<port name="in"/>
</input_ports>
<output_ports>
<port name="lut3_out"/>
<port name="lut4_out"/>
</output_ports>
</model>
<!-- A virtual model for scan-chain flip-flop to be used in the physical mode of FF -->
<model name="dff">
<input_ports>
<port name="D" clock="C"/>
<port name="C" is_clock="1"/>
</input_ports>
<output_ports>
<port name="Q" clock="C"/>
</output_ports>
</model>
<!-- A virtual model for scan-chain flip-flop to be used in the physical mode of FF -->
<model name="dffr">
<input_ports>
<port name="D" clock="C"/>
<port name="R" clock="C"/>
<port name="C" is_clock="1"/>
</input_ports>
<output_ports>
<port name="Q" clock="C"/>
</output_ports>
</model>
<!-- A virtual model for scan-chain flip-flop to be used in the physical mode of FF -->
<model name="dffrn">
<input_ports>
<port name="D" clock="C"/>
<port name="RN" clock="C"/>
<port name="C" is_clock="1"/>
</input_ports>
<output_ports>
<port name="Q" clock="C"/>
</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" area="0">
<sub_tile name="io" capacity="8">
<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>
</sub_tile>
</tile>
<tile name="clb" area="53894">
<sub_tile name="clb">
<equivalent_sites>
<site pb_type="clb"/>
</equivalent_sites>
<input name="I" num_pins="12" equivalent="full"/>
<input name="reset" num_pins="1" is_non_clock_global="true"/>
<input name="lreset" num_pins="1"/>
<output name="O" num_pins="8" equivalent="none"/>
<clock name="clk" num_pins="1"/>
<fc in_type="frac" in_val="0.15" out_type="frac" out_val="0.10">
<fc_override port_name="clk" fc_type="frac" fc_val="0"/>
<fc_override port_name="reset" fc_type="frac" fc_val="0"/>
</fc>
<pinlocations pattern="spread"/>
</sub_tile>
</tile>
</tiles>
<!-- ODIN II specific config ends -->
<!-- Physical descriptions begin -->
<layout tileable="true">
<auto_layout aspect_ratio="1.0">
<!--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"/>
</auto_layout>
<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>
<fixed_layout name="4x4" width="6" height="6">
<!--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>
<fixed_layout name="48x48" width="50" height="50">
<!--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>
</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" disable_packing="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="12" equivalent="full"/>
<input name="reset" num_pins="1"/>
<input name="lreset" num_pins="1"/>
<output name="O" num_pins="8" 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="4">
<input name="in" num_pins="4"/>
<input name="reset" num_pins="1"/>
<output name="out" num_pins="2"/>
<clock name="clk" num_pins="1"/>
<!-- Physical mode definition begin (physical implementation of the fle) -->
<mode name="physical" disable_packing="true">
<pb_type name="fabric" num_pb="1">
<input name="in" num_pins="4"/>
<input name="reset" num_pins="1"/>
<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="4"/>
<output name="out" num_pins="2"/>
<!-- Define LUT -->
<pb_type name="frac_lut4" blif_model=".subckt frac_lut4" num_pb="1">
<input name="in" num_pins="4"/>
<output name="lut3_out" num_pins="2"/>
<output name="lut4_out" num_pins="1"/>
</pb_type>
<interconnect>
<direct name="direct1" input="frac_logic.in" output="frac_lut4.in"/>
<direct name="direct2" input="frac_lut4.lut3_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_lut4.lut4_out frac_lut4.lut3_out[0]" output="frac_logic.out[0]"/>
</interconnect>
</pb_type>
<!-- Define flip-flop -->
<pb_type name="ff" blif_model=".subckt dffr" num_pb="2">
<input name="D" num_pins="1" port_class="D"/>
<input name="R" num_pins="1"/>
<output name="Q" num_pins="1" port_class="Q"/>
<clock name="C" num_pins="1" port_class="clock"/>
<T_setup value="66e-12" port="ff.D" clock="C"/>
<T_setup value="66e-12" port="ff.R" clock="C"/>
<T_clock_to_Q max="124e-12" port="ff.Q" clock="C"/>
</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].C"/>
<complete name="direct4" input="fabric.reset" output="ff[1:0].R"/>
<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"/>
<direct name="direct4" input="fle.reset" output="fabric.reset"/>
</interconnect>
</mode>
<!-- Physical mode definition end (physical implementation of the fle) -->
<!-- Dual 3-LUT mode definition begin -->
<mode name="n2_lut3">
<pb_type name="lut3inter" num_pb="1">
<input name="in" num_pins="3"/>
<input name="reset" num_pins="1"/>
<output name="out" num_pins="2"/>
<clock name="clk" num_pins="1"/>
<pb_type name="ble3" num_pb="2">
<input name="in" num_pins="3"/>
<input name="reset" num_pins="1"/>
<output name="out" num_pins="1"/>
<clock name="clk" num_pins="1"/>
<!-- Define the LUT -->
<pb_type name="lut3" blif_model=".names" num_pb="1" class="lut">
<input name="in" num_pins="3" 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="lut3.in" out_port="lut3.out">
235e-12
235e-12
235e-12
</delay_matrix>
</pb_type>
<!-- Define the flip-flop -->
<pb_type name="ff" num_pb="1">
<input name="D" num_pins="1"/>
<input name="R" num_pins="1"/>
<output name="Q" num_pins="1"/>
<clock name="C" num_pins="1"/>
<mode name="latch">
<pb_type name="latch" blif_model=".latch" num_pb="1">
<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="latch.D" clock="clk"/>
<T_clock_to_Q max="124e-12" port="latch.Q" clock="clk"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="latch.D"/>
<direct name="direct2" input="ff.C" output="latch.clk"/>
<direct name="direct3" input="latch.Q" output="ff.Q"/>
</interconnect>
</mode>
<mode name="dff">
<pb_type name="dff" blif_model=".subckt dff" num_pb="1">
<input name="D" num_pins="1" port_class="D"/>
<output name="Q" num_pins="1" port_class="Q"/>
<clock name="C" num_pins="1" port_class="clock"/>
<T_setup value="66e-12" port="dff.D" clock="C"/>
<T_clock_to_Q max="124e-12" port="dff.Q" clock="C"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="dff.D"/>
<direct name="direct2" input="ff.C" output="dff.C"/>
<direct name="direct3" input="dff.Q" output="ff.Q"/>
</interconnect>
</mode>
<mode name="dffr">
<pb_type name="dffr" blif_model=".subckt dffr" num_pb="1">
<input name="D" num_pins="1" port_class="D"/>
<input name="R" num_pins="1"/>
<output name="Q" num_pins="1" port_class="Q"/>
<clock name="C" num_pins="1" port_class="clock"/>
<T_setup value="66e-12" port="dffr.D" clock="C"/>
<T_setup value="66e-12" port="dffr.R" clock="C"/>
<T_clock_to_Q max="124e-12" port="dffr.Q" clock="C"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="dffr.D"/>
<direct name="direct2" input="ff.C" output="dffr.C"/>
<direct name="direct3" input="ff.R" output="dffr.R"/>
<direct name="direct4" input="dffr.Q" output="ff.Q"/>
</interconnect>
</mode>
<mode name="dffrn">
<pb_type name="dffrn" blif_model=".subckt dffrn" num_pb="1">
<input name="D" num_pins="1" port_class="D"/>
<input name="RN" num_pins="1"/>
<output name="Q" num_pins="1" port_class="Q"/>
<clock name="C" num_pins="1" port_class="clock"/>
<T_setup value="66e-12" port="dffrn.D" clock="C"/>
<T_setup value="66e-12" port="dffrn.RN" clock="C"/>
<T_clock_to_Q max="124e-12" port="dffrn.Q" clock="C"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="dffrn.D"/>
<direct name="direct2" input="ff.C" output="dffrn.C"/>
<direct name="direct3" input="ff.R" output="dffrn.RN"/>
<direct name="direct4" input="dffrn.Q" output="ff.Q"/>
</interconnect>
</mode>
</pb_type>
<interconnect>
<direct name="direct1" input="ble3.in[2:0]" output="lut3[0:0].in[2:0]"/>
<direct name="direct2" input="lut3[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="ble3" in_port="lut3[0:0].out" out_port="ff[0:0].D"/>
</direct>
<direct name="direct3" input="ble3.clk" output="ff[0:0].C"/>
<direct name="direct4" input="ble3.reset" output="ff[0:0].R"/>
<mux name="mux1" input="ff[0:0].Q lut3.out[0:0]" output="ble3.out[0:0]">
<!-- LUT to output is faster than FF to output on a Stratix IV -->
<delay_constant max="25e-12" in_port="lut3.out[0:0]" out_port="ble3.out[0:0]"/>
<delay_constant max="45e-12" in_port="ff[0:0].Q" out_port="ble3.out[0:0]"/>
</mux>
</interconnect>
</pb_type>
<interconnect>
<direct name="direct1" input="lut3inter.in" output="ble3[0:0].in"/>
<direct name="direct2" input="lut3inter.in" output="ble3[1:1].in"/>
<direct name="direct3" input="ble3[1:0].out" output="lut3inter.out"/>
<complete name="complete1" input="lut3inter.clk" output="ble3[1:0].clk"/>
<complete name="complete2" input="lut3inter.reset" output="ble3[1:0].reset"/>
</interconnect>
</pb_type>
<interconnect>
<direct name="direct1" input="fle.in[2:0]" output="lut3inter.in"/>
<direct name="direct2" input="lut3inter.out" output="fle.out"/>
<direct name="direct3" input="fle.clk" output="lut3inter.clk"/>
<direct name="direct4" input="fle.reset" output="lut3inter.reset"/>
</interconnect>
</mode>
<!-- Dual 3-LUT mode definition end -->
<!-- 4-LUT mode definition begin -->
<mode name="n1_lut4">
<!-- Define 4-LUT mode -->
<pb_type name="ble4" num_pb="1">
<input name="in" num_pins="4"/>
<input name="reset" num_pins="1"/>
<output name="out" num_pins="1"/>
<clock name="clk" num_pins="1"/>
<!-- Define LUT -->
<pb_type name="lut4" blif_model=".names" num_pb="1" class="lut">
<input name="in" num_pins="4" 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="lut4.in" out_port="lut4.out">
261e-12
261e-12
261e-12
261e-12
</delay_matrix>
</pb_type>
<!-- Define the flip-flop -->
<pb_type name="ff" num_pb="1">
<input name="D" num_pins="1"/>
<input name="R" num_pins="1"/>
<output name="Q" num_pins="1"/>
<clock name="C" num_pins="1"/>
<mode name="latch">
<pb_type name="latch" blif_model=".latch" num_pb="1">
<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="latch.D" clock="clk"/>
<T_clock_to_Q max="124e-12" port="latch.Q" clock="clk"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="latch.D"/>
<direct name="direct2" input="ff.C" output="latch.clk"/>
<direct name="direct3" input="latch.Q" output="ff.Q"/>
</interconnect>
</mode>
<mode name="dff">
<pb_type name="dff" blif_model=".subckt dff" num_pb="1">
<input name="D" num_pins="1" port_class="D"/>
<output name="Q" num_pins="1" port_class="Q"/>
<clock name="C" num_pins="1" port_class="clock"/>
<T_setup value="66e-12" port="dff.D" clock="C"/>
<T_clock_to_Q max="124e-12" port="dff.Q" clock="C"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="dff.D"/>
<direct name="direct2" input="ff.C" output="dff.C"/>
<direct name="direct3" input="dff.Q" output="ff.Q"/>
</interconnect>
</mode>
<mode name="dffr">
<pb_type name="dffr" blif_model=".subckt dffr" num_pb="1">
<input name="D" num_pins="1" port_class="D"/>
<input name="R" num_pins="1"/>
<output name="Q" num_pins="1" port_class="Q"/>
<clock name="C" num_pins="1" port_class="clock"/>
<T_setup value="66e-12" port="dffr.D" clock="C"/>
<T_setup value="66e-12" port="dffr.R" clock="C"/>
<T_clock_to_Q max="124e-12" port="dffr.Q" clock="C"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="dffr.D"/>
<direct name="direct2" input="ff.C" output="dffr.C"/>
<direct name="direct3" input="ff.R" output="dffr.R"/>
<direct name="direct4" input="dffr.Q" output="ff.Q"/>
</interconnect>
</mode>
<mode name="dffrn">
<pb_type name="dffrn" blif_model=".subckt dffrn" num_pb="1">
<input name="D" num_pins="1" port_class="D"/>
<input name="RN" num_pins="1"/>
<output name="Q" num_pins="1" port_class="Q"/>
<clock name="C" num_pins="1" port_class="clock"/>
<T_setup value="66e-12" port="dffrn.D" clock="C"/>
<T_setup value="66e-12" port="dffrn.RN" clock="C"/>
<T_clock_to_Q max="124e-12" port="dffrn.Q" clock="C"/>
</pb_type>
<interconnect>
<direct name="direct1" input="ff.D" output="dffrn.D"/>
<direct name="direct2" input="ff.C" output="dffrn.C"/>
<direct name="direct3" input="ff.R" output="dffrn.RN"/>
<direct name="direct4" input="dffrn.Q" output="ff.Q"/>
</interconnect>
</mode>
</pb_type>
<interconnect>
<direct name="direct1" input="ble4.in" output="lut4[0:0].in"/>
<direct name="direct2" input="lut4.out" output="ff.D">
<!-- Advanced user option that tells CAD tool to find LUT+FF pairs in netlist -->
<pack_pattern name="ble4" in_port="lut4.out" out_port="ff.D"/>
</direct>
<direct name="direct3" input="ble4.clk" output="ff.C"/>
<direct name="direct4" input="ble4.reset" output="ff.R"/>
<mux name="mux1" input="ff.Q lut4.out" output="ble4.out">
<!-- LUT to output is faster than FF to output on a Stratix IV -->
<delay_constant max="25e-12" in_port="lut4.out" out_port="ble4.out"/>
<delay_constant max="45e-12" in_port="ff.Q" out_port="ble4.out"/>
</mux>
</interconnect>
</pb_type>
<interconnect>
<direct name="direct1" input="fle.in" output="ble4.in"/>
<direct name="direct2" input="ble4.out" output="fle.out[0:0]"/>
<direct name="direct3" input="fle.clk" output="ble4.clk"/>
<direct name="direct4" input="fle.reset" output="ble4.reset"/>
</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[3:0].out clb.reset clb.lreset" output="fle[3:0].in">
<delay_constant max="95e-12" in_port="clb.I" out_port="fle[3:0].in"/>
<delay_constant max="75e-12" in_port="fle[3:0].out" out_port="fle[3:0].in"/>
</complete>
<complete name="clks" input="clb.clk" output="fle[3:0].clk">
</complete>
<complete name="resets" input="clb.lreset clb.reset" output="fle[3:0].reset">
</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[3:0].out[0:0]" output="clb.O[3:0]"/>
<direct name="clbouts2" input="fle[3:0].out[1:1]" output="clb.O[7:4]"/>
</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>