The tests/silimate suite (which aborted the Run tests CI job) exposed two
issues in the generalized passes:
- opt_compact_prefix: the forward dense pack regressions at 64 and 128 bits
no longer rewrote. The ConstEval fingerprint was uint64_t-based (capped at
62 bits) and the per-cone cell cap (max_width*96) was below the O(width^2)
cell count of a wide pack. The fingerprint now drives whole-width Const
bit patterns (no width cap) and the cone cap scales quadratically; total
work stays bounded by the shared walk/eval budgets.
- opt_priority_onehot: the "max-width below lane count" negative test set
max_width=8 on a 16-lane design expecting no rewrite, but the generalized
matcher legitimately (and equivalence-provably) rewrites the 8-lane
sub-region. The test now uses max_width=3 (below min_width 4) to verify
the width gate suppresses all matching.
Co-authored-by: Cursor <cursoragent@cursor.com>
- update CMakeLists.txt to load two new files:
- SilimateConfig.cmake: sets Silimate configuration options and defaults
- SilimateVerific.cmake: compiles Verific library, optionally with Silimate modifications
- include silimate tests in test Makefile
Make explicit the location for the `CMakeUserPresets.json` file.
Update docs tests documentation.
Move `BUILD_DIR` and `PROGRAM_PREFIX` comments into the functional tests todo.
Add cmake to MacPorts/FreeBSD install instructions.
Add ccache to example `CMakeUserPresets.json` (it isn't otherwise documented, and is no longer as obvious without an `ENABLE_CCACHE` makevar).
Addressing comments
Also skip calling `tail` on stubnets log outputs.
Fix auxprog prefixes
i.e. if we've set `YOSYS_PROGRAM_PREFIX=why`, we shouldn't be trying to call `build/yosys` when building docs (because if we've done a clean build it doesn't exist).
I suspect this also affects tests, but I was working on docs when I noticed it.
Pipes the help through sed for the tools that include the path. Also means we can drop the `|| true` from abc, since we're now reading the sed return instead.
The CMake refactor changes relative paths to absolute when calling `source_location::current()`. Use the path to `register.cc` to find the root yosys dir, and use that as the base for other paths.
Includes provisions for when plugins have a valid location; if it's not in the yosys source tree we shouldn't try to make the path relative, and we shouldn't try to auto group by path (since such groups will never be allocated, leaving them as unknown is preferable).
Limited to non wasm (I think that's the only one where we can't (easily) use `<filesystem>`). But where we do use `<filesystem>` we can also simplify the auto group to use `fs::path::parent_path()`.
This commit reimplements the (no longer recommended) setuptools based
build system using a standards-based in-tree PEP517 build backend.
The implementation is partially based on
https://codeberg.org/ziglang/zig-pypi/src/branch/main/make_wheels.py
which is licensed under BSD-0-clause.
It also adds a new option `YOSYS_BUILD_PYTHON_ONLY` that is available
only if the binary or the library aren't going to be installed, which
turns off these targets entirely, as well as some dependent ones
(e.g. tests).
Co-authored-by: Mohamed Gaber <me@donn.website>
