Releases: google/xls
Releases · google/xls
v0.0.0-8401-g8966779b9
Merge pull request #2811 from hzeller:feature-20250808-inline-constexpr PiperOrigin-RevId: 792780388
v0.0.0-8392-g93a62b8bd
Migrate away from ArrayRef(std::nullopt_t) The upstream LLVM has deprecated ArrayRef(std::nullopt_t). This CL migrates away from that. PiperOrigin-RevId: 792491328
v0.0.0-8384-g20c59789b
Ensure pass_profile works correctly even if multiple threads are runn…
v0.0.0-8369-g6bd3a42b2
Factor out SweepPipelineStagesFixture to passes_ng_text_fixtures.h. PiperOrigin-RevId: 791354381
v0.0.0-8362-g86a7a985c
Merge pull request #2769 from hzeller:feature-20250802-use-stdlib-rea…
v0.0.0-8349-g104e2ebde
Rewrite SLessThan using ITE in abstract eval. PiperOrigin-RevId: 789957912
v0.0.0-8321-g6ea1a9c95
Include evaluation results in TIv2 traces, and stop truncating the st…
v0.0.0-8312-g50c3490a2
Switch some errors away from InternalError(). We generally don't care too much about which abseil error code we use, but these really seem more like invalid arguments than internal errors. The motivation is to fuzz codegen for correctly-returned errors. We want InternalErrors to be "bad" fuzz findings and InvalidArgumentErrors (among others) will be "good" fuzz findings. PiperOrigin-RevId: 788681051
v0.0.0-8301-g8aed080e3
xls/examples/matmul_4x4: load weights from channel - add weight state to PE node - add one-of weights load on first activation - send weights from test - improve comments PiperOrigin-RevId: 788272142
v0.0.0-8280-g390405a30
Add FuzzTestPrintSourceCode implementations for a few types In a near future version of gfuzztest this function allows one to override source-code printing for these types. This uses https://github.com/google/fuzztest/pull/1780. PiperOrigin-RevId: 787288665