Commit Graph

590 Commits

Author SHA1 Message Date
Jannis Harder bfd9cf63db Ignore $scopeinfo in opt_merge 2024-02-06 17:51:29 +01:00
Jannis Harder 8902fc94b6 Suport $scopeinfo in flatten and opt_clean 2024-02-06 17:51:29 +01:00
N. Engelhardt f96e27ac14
Merge pull request #4123 from povik/clean-opt_clean
opt_clean: Add commentary, remove dead code
2024-02-05 15:08:34 +01:00
Catherine c7bf0e3b8f Add new `$check` cell to represent assertions with a message. 2024-02-01 20:10:39 +01:00
Martin Povišer 7afc0696e2 opt_clean: Assert an impossible path isn't taken 2024-01-29 11:26:44 +01:00
Martin Povišer ec065186d3 opt_clean: Add commentary around wire cleaning, NFC 2024-01-29 11:26:44 +01:00
Martin Povišer 568418b50b opt_lut: Replace `-dlogic` with `-tech ice40` 2024-01-15 12:35:21 +01:00
Lofty d21c464ae4
Merge pull request #3946 from rmlarsen/toposort
Speed up TopoSort by 2.7-3.3x.
2023-10-17 13:00:18 +01:00
Rasmus Munk Larsen 0a37c2a301 Fix translation bug: The old code really checks for the presense of a node, not an edge in glift and flatten.
Add back statement that inserts nodes in order in opt_expr.cc.
2023-10-05 17:01:42 -07:00
Rasmus Munk Larsen 57a2b4b0cd Explicitly use uint64_t as the type of fingerprint to avoid type mismatch with some compilers. 2023-10-03 15:02:02 -07:00
Rasmus Munk Larsen 8e0308b5e7 Revert changes to celltypes.h. Use dict instead of std::unordered_map and most hash function for uint64_t to hashlib.h to support this. 2023-10-03 14:25:59 -07:00
Rasmus Munk Larsen bce984fa60 Speed up OptMergePass by 1.7x.
The main speedup comes from swithing from using a SHA1 hash to std::hash<std::string>. There is no need to use an expensive cryptographic hash for fingerprinting in this context.
2023-10-02 15:57:18 -07:00
Rasmus Munk Larsen e0042bdff7 Speed up TopoSort. The main sorting algorithm implementation in TopoSort::sort_worker is 11-12x faster. Overall, the complete sequence of building the graph and sorting is about 2.5-3x faster. The overall impact in e.g. the replace_const_cells optimization pass is a ~25% speedup. End-to-end impact on our synthesis flow is about 3%. 2023-09-20 15:49:05 -07:00
Jannis Harder 62b4df4989 dft_tag: Implement `$overwrite_tag` and `$original_tag`
This does not correctly handle an `$overwrite_tag` on a module output,
but since we currently require the user to flatten the design for
cross-module dft, this cannot be observed from within the design, only
by manually inspecting the signals in the design.
2023-09-13 11:32:36 +02:00
Miodrag Milanovic a8809989c4 ecp5_gsr -> lattice_gsr, change opt_lut_ins to accept lattice as tech 2023-08-22 10:50:11 +02:00
whitequark d51ecde8c2 clean: keep $print cells, since they have unmodelled side effects. 2023-08-11 04:46:52 +02:00
Martin Povišer ff3c7873f5 wreduce: Group reconnections
Group the reconnections, so that instead of producing

  connect $auto$wreduce.cc:455:run$24 [0] 1'0
  connect $auto$wreduce.cc:455:run$23 [31] 1'0
  connect $auto$wreduce.cc:455:run$23 [30] 1'0
  ... (40 more lines)

we produce

  connect $auto$wreduce.cc:461:run$23 [31:11] 21'000000000000000000000
  connect $auto$wreduce.cc:461:run$24 [31:10] 22'0000000000000000000000

.
2023-08-04 14:43:59 +01:00
Martin Povišer f8325f66b7 opt_expr: Fix 'signed X>=0' replacement for wide output ports
If the `$ge` cell we are replacing has wide output port, the upper bits
on the port should be driven to zero. That's not what a `$not` cell with
a single-bit input does. Instead opt for a `$logic_not` cell, which does
zero-pad its output.

Fixes #3867.
2023-08-01 13:50:12 +01:00
Martin Povišer f0ae046c5a opt_share: Fix input confusion with ANDNOT, ORNOT gates
Distinguish between the A, B input ports of `$_ANDNOT_`, `$_ORNOT_`
gates when considering those for sharing. Unlike the input ports of the
other supported single-bit gates, those are not interchangeable.

Fixes #3848.
2023-07-20 20:58:52 +01:00
Jannis Harder e36c71b5b7 Use clk2fflogic attr on cells to track original FF names in witnesses
This makes clk2fflogic add an attr to $ff cells that carry the state of
the emulated async FF. The $ff output doesn't have any async updates
that happened in the current cycle, but the $ff input does, so the $ff
input corresponds to the async FF's output in the original design.

Hence this patch also makes the following changes to passes besides
clk2fflogic (but only for FFs with the clk2fflogic attr set):

  * opt_clean treats the input as a register name (instead of the
    output)

  * rename -witness ensures that the input has a public name

  * the formal backends (smt2, btor, aiger) will use the input's
    name for the initial state of the FF in witness files

  * when sim reads a yw witness that assigns an initial value to the
    input signal, the state update is redirected to the output

This ensures that yosys witness files for clk2fflogic designs have
useful and stable public signal names. It also makes it possible to
simulate a clk2fflogic witness on the original design (with some
limitations when the original design is already using $ff cells).

It might seem like setting the output of a clk2fflogic FF to update the
input's initial value might not work in general, but it works fine for
these reasons:

  * Witnesses for FFs are only present in the initial cycle, so we do
    not care about any later cycles.

  * The logic that clk2fflogic generates loops the output of the
    genreated FF back to the input, with muxes in between to apply any
    edge or level sensitive updates. So when there are no active updates
    in the current gclk cycle, there is a combinational path from the
    output back to the input.

  * The logic clk2fflogic generates makes sure that an edge sensitive
    update cannot be active in the first cycle (i.e. the past initial
    value is assumed to be whatever it needs to be to avoid an edge).

  * When a level sensitive update is active in the first gclk cycle, it
    is actively driving the output for the whole gclk cycle, so ignoring
    any witness initialization is the correct behavior.
2023-05-25 12:48:02 +02:00
Miodrag Milanovic 6574553189 Fixes for some of clang scan-build detected issues 2023-01-17 12:58:08 +01:00
Jannis Harder 3ecf85e32c opt_expr: Optimizations for `$bweqx` and `$bwmux` 2022-11-30 18:50:53 +01:00
Jannis Harder 1e67c3a3c2 opt_expr: Fix shift/shiftx optimizations 2022-11-30 18:24:25 +01:00
Jannis Harder fd56d1f79e opt_expr: Constant fold mux, pmux, bmux, demux, eqx, nex cells 2022-11-29 19:06:45 +01:00
Jannis Harder c08242ba41 opt_expr: Optimize bitwise logic ops with one fully const input 2022-11-29 19:06:45 +01:00
Jannis Harder c77b7343d0 Consistent $mux undef handling
* Change simlib's $mux cell to use the ternary operator as $_MUX_
  already does
* Stop opt_expr -keepdc from changing S=x to S=0
* Change const eval of $mux and $pmux to match the updated simlib
  (fixes sim)
* The sat behavior of $mux already matches the updated simlib

The verilog frontend uses $mux for the ternary operators and this
changes all interpreations of the $mux cell (that I found) to match the
verilog simulation behavior for the ternary operator. For 'if' and
'case' expressions the frontend may also use $mux but uses $eqx if the
verilog simulation behavior is requested with the '-ifx' option.

For $pmux there is a remaining mismatch between the sat behavior and the
simlib behavior. Resolving this requires more discussion, as the $pmux
cell does not directly correspond to a specific verilog construct.
2022-10-24 12:03:01 +02:00
KrystalDelusion 9465b2af95 Fitting help messages to 80 character width
Uses the regex below to search (using vscode):
	^\t\tlog\("(.{10,}(?<!\\n)|.{81,}\\n)"\);

Finds any log messages double indented (which help messages are)
and checks if *either* there are is no newline character at the end,
*or* the number of characters before the newline is more than 80.
2022-08-24 10:40:57 +12:00
Jannis Harder 428ad5b9fd wreduce: Keep more x-bits with -keepdc 2022-08-16 13:37:30 +02:00
Jannis Harder c0063288d6 Add the $anyinit cell and the formalff pass
These can be used to protect undefined flip-flop initialization values
from optimizations that are not sound for formal verification and can
help mapping all solver-provided values in witness traces for flows that
use different backends simultaneously.
2022-08-16 13:37:30 +02:00
Marcelina Kościelnicka f679b756d8 opt_reduce: Fix use-after-free.
Fixes #3418.
2022-07-23 17:27:26 +02:00
Jannis Harder a47254bd10 opt_dff: With -keepdc, never turn undef init vals into const drivers 2022-06-29 15:42:39 +02:00
Marcelina Kościelnicka 1ff0e1a58a opt_ffinv: Fix use after free. 2022-06-13 14:04:04 +02:00
Marcelina Kościelnicka 47efc04a7d wreduce: Introduce -mux_undef option (aligned with opt_expr). 2022-06-08 21:28:58 +02:00
Marcelina Kościelnicka d07828b409 opt_ffinv: Harden against simple ff/inv loop. 2022-06-07 08:20:06 +02:00
Marcelina Kościelnicka 98c7804b89 opt_ffinv: Use ModIndex instead of ModWalker.
This avoids using out-of-data index information.
2022-05-17 02:52:21 +02:00
Marcelina Kościelnicka 2858bb03cd Add opt_ffinv pass. 2022-05-13 23:02:30 +02:00
Marcelina Kościelnicka 77b1dfd8c3 opt_mem: Remove constant-value bit lanes. 2022-05-07 23:13:16 +02:00
Marcelina Kościelnicka 48eea3efcf opt_dff: Fix behavior on $ff with D == Q. 2022-04-15 22:00:32 +02:00
Jannis Harder ca5b910296 opt_merge: Add `-keepdc` option required for formal verification
The `-keepdc` option prevents merging flipflops with dont-care bits in
their initial value, as, in general, this is not a valid transform for
formal verification.

The keepdc option of `opt` is passed along to `opt_merge` now.
2022-04-01 21:03:20 +02:00
Marcelina Kościelnicka 07a657fb0c opt_reduce: Add $bmux and $demux optimization patterns. 2022-01-30 03:37:52 +01:00
Marcelina Kościelnicka db33b1e535 opt_dff: Don't mutate muxes while ModWalker is active. 2022-01-28 08:55:56 +01:00
Austin Seipp b022fe61a7 opt_dff: fix sequence point copy paste bug
Newer GCCs emit the following warning for opt_dff:

    passes/opt/opt_dff.cc:560:17: warning: operation on ‘ff.Yosys::FfData::has_clk’ may be undefined [-Wsequence-point]
      560 |      ff.has_clk = ff.has_ce = ff.has_clk = false;
          |      ~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Which is correct: the order of whether the read or write of has_clk
occurs first is undefined since there is no sequence point between them.

This is almost certainly just a typo/copy paste error and objectively
wrong, so just fix it.

Signed-off-by: Austin Seipp <aseipp@pobox.com>
2022-01-04 18:18:08 +01:00
Marcelina Kościelnicka 1184a7f3b4 opt_mem_priority: Fix non-ascii char in help message.
This is a fixed version of #3072.
2021-12-09 00:56:14 +01:00
Pepijn de Vos 0c7461fe5e
gowin: widelut support (#3042) 2021-11-06 16:09:30 +01:00
Marcelina Kościelnicka 4e70c30775 FfData: some refactoring.
- FfData now keeps track of the module and underlying cell, if any (so
  calling emit on FfData created from a cell will replace the existing cell)
- FfData implementation is split off to its own .cc file for faster
  compilation
- the "flip FF data sense by inserting inverters in front and after"
  functionality that zinit uses is moved onto FfData class and beefed up
  to have dffsr support, to support more use cases
2021-10-07 04:24:06 +02:00
Marcelina Kościelnicka e7d89e653c Hook up $aldff support in various passes. 2021-10-02 21:01:21 +02:00
Marcelina Kościelnicka 63b9df8693 kernel/ff: Refactor FfData to enable FFs with async load.
- *_en is split into *_ce (clock enable) and *_aload (async load aka
  latch gate enable), so both can be present at once
- has_d is removed
- has_gclk is added (to have a clear marker for $ff)
- d_is_const and val_d leftovers are removed
- async2sync, clk2fflogic, opt_dff are updated to operate correctly on
  FFs with async load
2021-10-02 20:19:48 +02:00
Marcelina Kościelnicka 9cbff3a4a9 opt_merge: Remove and reinsert init when connecting nets.
Mutating the SigMap by adding a new connection will throw off FfInitVals
index.  Work around this by removing the relevant init values from index
whenever we connect nets, then re-add the new init value.

Should fix #2920.
2021-08-22 18:34:11 +02:00
Marcelina Kościelnicka 62d41d4639 opt_clean: Make the init attribute follow the FF's Q.
Previously, opt_clean would reconnect all ports (including FF Q ports)
to a "canonical" SigBit chosen by complex rules, but would leave the
init attribute on the old wire.  This change applies the same
canonicalization rules to the init attributes, ensuring that init moves
to wherever the Q port moved.

Part of another jab at #2920.
2021-08-22 15:38:29 +02:00
Marcelina Kościelnicka f791328506 Add opt_mem_widen pass.
If all of us are wide, then none of us are!
2021-08-14 01:06:23 +02:00