[AIMIGRAPHX-578] Use Eigen 3rd party library for ref GEMMs#4631
[AIMIGRAPHX-578] Use Eigen 3rd party library for ref GEMMs#4631
Conversation
src/CMakeLists.txt
Outdated
| target_link_libraries(migraphx PUBLIC Threads::Threads) | ||
|
|
||
| if(MIGRAPHX_USE_EIGEN) | ||
| find_path(EIGEN_INCLUDE Eigen/Core PATH_SUFFIXES eigen3) |
There was a problem hiding this comment.
This should use pkgconfig.
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## develop #4631 +/- ##
========================================
Coverage 92.27% 92.27%
========================================
Files 578 578
Lines 28541 28533 -8
========================================
- Hits 26335 26328 -7
+ Misses 2206 2205 -1
🚀 New features to boost your workflow:
|
src/gemm.cpp
Outdated
| template <class Visitor> | ||
| void gemm_ref(const argument& c_arg, const argument& a_arg, const argument& b_arg, Visitor v) | ||
| { | ||
| if(c_arg.get_shape().type() == a_arg.get_shape().type()) |
There was a problem hiding this comment.
I dont think this if statement is needed as the visit below will handle the case when they are the same.
src/gemm.cpp
Outdated
| if(c_arg.get_shape().type() == a_arg.get_shape().type()) | ||
| { | ||
| visit_all(c_arg, a_arg, b_arg)( | ||
| [&](auto cmat, auto amat, auto bmat) { v(cmat, amat, bmat); }); | ||
| } | ||
| else | ||
| { | ||
| c_arg.visit([&](auto cmat) { | ||
| visit_all(a_arg, b_arg)([&](auto amat, auto bmat) { v(cmat, amat, bmat); }); | ||
| }); | ||
| } |
There was a problem hiding this comment.
| if(c_arg.get_shape().type() == a_arg.get_shape().type()) | |
| { | |
| visit_all(c_arg, a_arg, b_arg)( | |
| [&](auto cmat, auto amat, auto bmat) { v(cmat, amat, bmat); }); | |
| } | |
| else | |
| { | |
| c_arg.visit([&](auto cmat) { | |
| visit_all(a_arg, b_arg)([&](auto amat, auto bmat) { v(cmat, amat, bmat); }); | |
| }); | |
| } | |
| c_arg.visit([&](auto cmat) { | |
| visit_all(a_arg, b_arg)([&](auto amat, auto bmat) { v(cmat, amat, bmat); }); | |
| }); |
src/gemm.cpp
Outdated
| batch_slicer a_slicer(amat.get_shape()); | ||
| batch_slicer b_slicer(bmat.get_shape()); | ||
|
|
||
| for(std::size_t batch = 0; batch < slicer.num_batches(); batch++) |
There was a problem hiding this comment.
Will using par_for help as well?
src/gemm.cpp
Outdated
| #endif | ||
|
|
||
| template <class Visitor> | ||
| void gemm_ref(const argument& c_arg, const argument& a_arg, const argument& b_arg, Visitor v) |
There was a problem hiding this comment.
I would name this gemm_ref_visit.
| void gemm_ref(const argument& c_arg, const argument& a_arg, const argument& b_arg, Visitor v) | |
| void gemm_ref_visit(const argument& c_arg, const argument& a_arg, const argument& b_arg, Visitor v) |
src/CMakeLists.txt
Outdated
| find_package(PkgConfig) | ||
| if(PkgConfig_FOUND) | ||
| pkg_check_modules(EIGEN3 eigen3) | ||
| endif() | ||
| if(EIGEN3_FOUND) | ||
| target_compile_definitions(migraphx PUBLIC MIGRAPHX_USE_EIGEN=1) | ||
| target_compile_definitions(migraphx PRIVATE EIGEN_MPL2_ONLY) | ||
| target_include_directories(migraphx SYSTEM PUBLIC $<BUILD_INTERFACE:${EIGEN3_INCLUDE_DIRS}>) |
There was a problem hiding this comment.
I looked and there is eigen3 cmake package config so we dont need to use pkgconfig.
| find_package(PkgConfig) | |
| if(PkgConfig_FOUND) | |
| pkg_check_modules(EIGEN3 eigen3) | |
| endif() | |
| if(EIGEN3_FOUND) | |
| target_compile_definitions(migraphx PUBLIC MIGRAPHX_USE_EIGEN=1) | |
| target_compile_definitions(migraphx PRIVATE EIGEN_MPL2_ONLY) | |
| target_include_directories(migraphx SYSTEM PUBLIC $<BUILD_INTERFACE:${EIGEN3_INCLUDE_DIRS}>) | |
| find_package(Eigen3) | |
| if(EIGEN3_FOUND) | |
| target_compile_definitions(migraphx PUBLIC MIGRAPHX_USE_EIGEN=1) | |
| target_compile_definitions(migraphx PRIVATE EIGEN_MPL2_ONLY) | |
| target_link_libraries(migraphx PRIVATE Eigen3::Eigen) |
This build is not recommended to merge 🔴 |
❌bert_base_cased_fp16: ERROR - check error output�[1;31m2026-03-14 03:57:26.909563346 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181153, index: 63, mask: {64, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m�[1;31m2026-03-14 03:57:26.917903850 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181154, index: 64, mask: {65, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.917938675 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181162, index: 72, mask: {73, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.921893116 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181159, index: 69, mask: {70, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.921903124 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181160, index: 70, mask: {71, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.917929328 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181161, index: 71, mask: {72, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.921895731 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181157, index: 67, mask: {68, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.921899578 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181156, index: 66, mask: {67, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.917903900 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181155, index: 65, mask: {66, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.917903880 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181158, index: 68, mask: {69, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.935949608 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181177, index: 87, mask: {88, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.935963534 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181184, index: 94, mask: {95, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.935979354 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181191, index: 101, mask: {102, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.935999532 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181198, index: 108, mask: {109, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936169812 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181190, index: 100, mask: {101, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936200450 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181204, index: 114, mask: {115, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936114077 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181170, index: 80, mask: {81, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936227621 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181211, index: 121, mask: {122, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936136259 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181183, index: 93, mask: {94, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936027114 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181205, index: 115, mask: {116, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936190682 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181197, index: 107, mask: {108, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936047883 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181212, index: 122, mask: {123, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936114097 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181176, index: 86, mask: {87, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936293846 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181168, index: 78, mask: {79, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936345083 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181189, index: 99, mask: {100, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936309976 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181182, index: 92, mask: {93, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936368427 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181196, index: 106, mask: {107, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936295209 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181175, index: 85, mask: {86, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936378606 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181203, index: 113, mask: {114, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936401479 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181210, index: 120, mask: {121, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936877415 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181167, index: 77, mask: {78, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936877556 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181174, index: 84, mask: {85, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936903334 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181181, index: 91, mask: {92, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936924354 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181188, index: 98, mask: {99, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936938701 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181195, index: 105, mask: {106, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.936991550 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181202, index: 112, mask: {113, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937017539 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181209, index: 119, mask: {120, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937035363 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181166, index: 76, mask: {77, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937068726 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181173, index: 83, mask: {84, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937083874 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181201, index: 111, mask: {112, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937098482 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181180, index: 90, mask: {91, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937133197 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181187, index: 97, mask: {98, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937152614 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181194, index: 104, mask: {105, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937219911 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181208, index: 118, mask: {119, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937219319 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181165, index: 75, mask: {76, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937248434 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181215, index: 125, mask: {126, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937308578 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181172, index: 82, mask: {83, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937440196 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181214, index: 124, mask: {125, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937541476 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181179, index: 89, mask: {90, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937563638 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181186, index: 96, mask: {97, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937583876 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181193, index: 103, mask: {104, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937617730 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181200, index: 110, mask: {111, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937656503 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181207, index: 117, mask: {118, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937753927 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181213, index: 123, mask: {124, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937771490 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181178, index: 88, mask: {89, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937760830 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181164, index: 74, mask: {75, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937753896 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181171, index: 81, mask: {82, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937802969 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181185, index: 95, mask: {96, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937838285 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181192, index: 102, mask: {103, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937859756 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181199, index: 109, mask: {110, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937898629 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181169, index: 79, mask: {80, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937906774 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181206, index: 116, mask: {117, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.937941089 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181216, index: 126, mask: {127, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:26.929889000 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 181163, index: 73, mask: {74, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:57:28.403193143 [E:onnxruntime:, inference_session.cc:2544 operator()] Exception during initialization: /onnxruntime_src/onnxruntime/core/graph/graph_utils.cc:29 int onnxruntime::graph_utils::GetIndexFromName(const onnxruntime::Node&, const std::string&, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_onnx::Pow_1623for node: Mul_53/SimplifiedLayerNormFusion/ �[m Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 359, in main() File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 278, in main sess = ort.InferenceSession(model_name, File "/usr/local/lib/python3.10/dist-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 485, in init self._create_inference_session(providers, provider_options, disabled_optimizers) File "/usr/local/lib/python3.10/dist-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 584, in _create_inference_session sess.initialize_session(providers, provider_options, disabled_optimizers) onnxruntime.capi.onnxruntime_pybind11_state.RuntimeException: [ONNXRuntimeError] : 6 : RUNTIME_EXCEPTION : Exception during initialization: /onnxruntime_src/onnxruntime/core/graph/graph_utils.cc:29 int onnxruntime::graph_utils::GetIndexFromName(const onnxruntime::Node&, const std::string&, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_onnx::Pow_1623for node: Mul_53/SimplifiedLayerNormFusion/ ❌bert_large_uncased_fp16: ERROR - check error output�[1;31m2026-03-14 03:58:08.937228759 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188135, index: 63, mask: {64, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m�[1;31m2026-03-14 03:58:08.945897901 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188136, index: 64, mask: {65, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.949892327 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188141, index: 69, mask: {70, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.949904740 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188154, index: 82, mask: {83, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945903111 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188137, index: 65, mask: {66, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945900105 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188138, index: 66, mask: {67, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945921516 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188145, index: 73, mask: {74, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.953898034 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188155, index: 83, mask: {84, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945924381 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188146, index: 74, mask: {75, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.957890065 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188143, index: 71, mask: {72, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945923740 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188147, index: 75, mask: {76, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.949892768 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188139, index: 67, mask: {68, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945898292 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188140, index: 68, mask: {69, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.949909970 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188148, index: 76, mask: {77, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.949911473 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188152, index: 80, mask: {81, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.949911663 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188153, index: 81, mask: {82, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.953904426 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188150, index: 78, mask: {79, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.953908624 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188151, index: 79, mask: {80, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.953910297 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188149, index: 77, mask: {78, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945937235 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188144, index: 72, mask: {73, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963604620 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188163, index: 91, mask: {92, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963615371 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188170, index: 98, mask: {99, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963634687 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188176, index: 104, mask: {105, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963668972 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188183, index: 111, mask: {112, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963708757 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188190, index: 118, mask: {119, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963736499 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188161, index: 89, mask: {90, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963749273 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188168, index: 96, mask: {97, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963776895 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188175, index: 103, mask: {104, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963800620 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188182, index: 110, mask: {111, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963830726 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188189, index: 117, mask: {118, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963848800 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188196, index: 124, mask: {125, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963901860 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188160, index: 88, mask: {89, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963923010 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188167, index: 95, mask: {96, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963960441 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188174, index: 102, mask: {103, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963971942 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188181, index: 109, mask: {110, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.963992291 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188188, index: 116, mask: {117, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964009122 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188195, index: 123, mask: {124, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964273541 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188159, index: 87, mask: {88, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964295782 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188166, index: 94, mask: {95, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964328534 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188173, index: 101, mask: {102, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964349143 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188180, index: 108, mask: {109, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964367968 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188187, index: 115, mask: {116, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964417742 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188194, index: 122, mask: {123, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964453880 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188165, index: 93, mask: {94, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964477966 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188172, index: 100, mask: {101, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964519995 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188179, index: 107, mask: {108, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964547757 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188158, index: 86, mask: {87, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964582993 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188186, index: 114, mask: {115, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964601719 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188193, index: 121, mask: {122, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964865235 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188164, index: 92, mask: {93, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964889431 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188171, index: 99, mask: {100, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964905551 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188178, index: 106, mask: {107, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964923715 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188185, index: 113, mask: {114, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964959593 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188192, index: 120, mask: {121, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.964977667 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188157, index: 85, mask: {86, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965256652 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188177, index: 105, mask: {106, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965399081 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188156, index: 84, mask: {85, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965421533 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188162, index: 90, mask: {91, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965443024 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188169, index: 97, mask: {98, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965538503 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188184, index: 112, mask: {113, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965553081 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188198, index: 126, mask: {127, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965588928 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188191, index: 119, mask: {120, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.965788755 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188197, index: 125, mask: {126, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:08.945900206 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 188142, index: 70, mask: {71, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 03:58:12.975895784 [E:onnxruntime:, inference_session.cc:2544 operator()] Exception during initialization: /onnxruntime_src/onnxruntime/core/graph/graph_utils.cc:29 int onnxruntime::graph_utils::GetIndexFromName(const onnxruntime::Node&, const std::string&, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_onnx::Pow_3183for node: Mul_53/SimplifiedLayerNormFusion/ �[m Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 359, in main() File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 278, in main sess = ort.InferenceSession(model_name, File "/usr/local/lib/python3.10/dist-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 485, in init self._create_inference_session(providers, provider_options, disabled_optimizers) File "/usr/local/lib/python3.10/dist-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 584, in _create_inference_session sess.initialize_session(providers, provider_options, disabled_optimizers) onnxruntime.capi.onnxruntime_pybind11_state.RuntimeException: [ONNXRuntimeError] : 6 : RUNTIME_EXCEPTION : Exception during initialization: /onnxruntime_src/onnxruntime/core/graph/graph_utils.cc:29 int onnxruntime::graph_utils::GetIndexFromName(const onnxruntime::Node&, const std::string&, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_onnx::Pow_3183for node: Mul_53/SimplifiedLayerNormFusion/ ❌distilgpt2_fp16: ERROR - check error output�[1;31m2026-03-14 04:01:29.989840061 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204120, index: 63, mask: {64, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m�[1;31m2026-03-14 04:01:30.001906814 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204121, index: 64, mask: {65, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.015747300 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204136, index: 79, mask: {80, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016021627 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204143, index: 86, mask: {87, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016032076 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204150, index: 93, mask: {94, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016052044 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204157, index: 100, mask: {101, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016092039 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204164, index: 107, mask: {108, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016123478 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204171, index: 114, mask: {115, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016140741 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204178, index: 121, mask: {122, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016424596 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204142, index: 85, mask: {86, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016438772 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204149, index: 92, mask: {93, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016468989 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204156, index: 99, mask: {100, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016483697 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204163, index: 106, mask: {107, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016515186 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204170, index: 113, mask: {114, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016540845 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204177, index: 120, mask: {121, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016632838 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204141, index: 84, mask: {85, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016632868 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204135, index: 78, mask: {79, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016667113 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204148, index: 91, mask: {92, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016686389 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204155, index: 98, mask: {99, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016699894 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204162, index: 105, mask: {106, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016726965 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204169, index: 112, mask: {113, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016766520 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204176, index: 119, mask: {120, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016774535 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204183, index: 126, mask: {127, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.016817867 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204134, index: 77, mask: {78, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017014197 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204140, index: 83, mask: {84, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017034154 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204147, index: 90, mask: {91, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017044053 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204154, index: 97, mask: {98, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017068298 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204161, index: 104, mask: {105, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017100900 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204167, index: 110, mask: {111, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017129223 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204175, index: 118, mask: {119, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017141737 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204168, index: 111, mask: {112, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017160723 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204133, index: 76, mask: {77, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017196991 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204139, index: 82, mask: {83, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017216397 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204146, index: 89, mask: {90, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017240202 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204153, index: 96, mask: {97, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017259619 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204160, index: 103, mask: {104, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017313711 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204174, index: 117, mask: {118, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017340311 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204181, index: 124, mask: {125, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017590432 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204138, index: 81, mask: {82, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017608166 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204145, index: 88, mask: {89, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017643642 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204152, index: 95, mask: {96, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017656737 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204132, index: 75, mask: {76, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017659482 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204158, index: 101, mask: {102, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017698095 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204172, index: 115, mask: {116, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017720317 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204165, index: 108, mask: {109, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017727480 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204179, index: 122, mask: {123, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017774449 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204137, index: 80, mask: {81, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017774429 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204131, index: 74, mask: {75, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017797041 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204144, index: 87, mask: {88, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017831116 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204151, index: 94, mask: {95, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.018048044 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204127, index: 70, mask: {71, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.018321590 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204122, index: 65, mask: {66, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017905285 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204166, index: 109, mask: {110, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017906878 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204173, index: 116, mask: {117, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017907850 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204159, index: 102, mask: {103, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017909834 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204180, index: 123, mask: {124, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017909904 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204123, index: 66, mask: {67, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017915685 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204125, index: 68, mask: {69, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017916326 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204128, index: 71, mask: {72, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017911196 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204124, index: 67, mask: {68, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017909874 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204126, index: 69, mask: {70, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017914312 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204182, index: 125, mask: {126, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017935763 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204129, index: 72, mask: {73, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.017934721 [E:onnxruntime:Default, env.cc:226 ThreadMain] pthread_setaffinity_np failed for thread: 204130, index: 73, mask: {74, }, error code: 22 error msg: Invalid argument. Specify the number of threads explicitly so the affinity is not set.�[m �[1;31m2026-03-14 04:01:30.965621586 [E:onnxruntime:, inference_session.cc:2544 operator()] Exception during initialization: /onnxruntime_src/onnxruntime/core/graph/graph_utils.cc:29 int onnxruntime::graph_utils::GetIndexFromName(const onnxruntime::Node&, const std::string&, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_onnx::Pow_1551for node: Mul_33/SimplifiedLayerNormFusion/ �[m Traceback (most recent call last): File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 359, in main() File "/src/AMDMIGraphX/tools/accuracy/accuracy_checker.py", line 278, in main sess = ort.InferenceSession(model_name, File "/usr/local/lib/python3.10/dist-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 485, in init self._create_inference_session(providers, provider_options, disabled_optimizers) File "/usr/local/lib/python3.10/dist-packages/onnxruntime/capi/onnxruntime_inference_collection.py", line 584, in _create_inference_session sess.initialize_session(providers, provider_options, disabled_optimizers) onnxruntime.capi.onnxruntime_pybind11_state.RuntimeException: [ONNXRuntimeError] : 6 : RUNTIME_EXCEPTION : Exception during initialization: /onnxruntime_src/onnxruntime/core/graph/graph_utils.cc:29 int onnxruntime::graph_utils::GetIndexFromName(const onnxruntime::Node&, const std::string&, bool) itr != node_args.end() was false. Attempting to get index by a name which does not exist:InsertedPrecisionFreeCast_onnx::Pow_1551for node: Mul_33/SimplifiedLayerNormFusion/ |
There was a problem hiding this comment.
Pull request overview
This PR introduces an Eigen-backed implementation for reference GEMM execution in MIGraphX, replacing the previous header-only naive GEMM and wiring it through dot/quant_dot evaluation paths to improve ref-target performance.
Changes:
- Replaces the header-only
gemmtemplate with a compiledgemm(const argument&, ...)implementation and updatesdot/ref-lowering call sites. - Adds optional Eigen support (
MIGRAPHX_USE_EIGEN) to accelerate ref GEMM when Eigen is available. - Updates a reshaping ONNX test fixture and adds a changelog entry.
Reviewed changes
Copilot reviewed 8 out of 9 changed files in this pull request and generated 5 comments.
Show a summary per file
| File | Description |
|---|---|
| test/onnx/reshape_test.onnx | Updates an ONNX test model fixture used by reshape parsing tests. |
| src/targets/ref/lowering.cpp | Removes ref-specific dot lowering and simplifies quant_dot ref compute to call the new gemm. |
| src/include/migraphx/op/dot.hpp | Switches dot compute to use the new gemm(argument, ...) entry point. |
| src/include/migraphx/gemm.hpp | Replaces the template GEMM with a declared compiled function. |
| src/gemm.cpp | Adds the new GEMM implementation with optional Eigen acceleration. |
| src/CMakeLists.txt | Adds gemm.cpp to the build and wires in optional Eigen discovery/linking/defines. |
| requirements.txt | Adds Eigen as a third-party dependency source. |
| CMakeLists.txt | Adds a top-level CMake option to enable/disable Eigen usage. |
| CHANGELOG.md | Adds a changelog entry for Eigen-based ref GEMM support. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
| template <class T, class U> | ||
| void gemm_eigen(tensor_view<T> cmat, tensor_view<U> amat, tensor_view<U> bmat) | ||
| { | ||
| if constexpr(std::is_same<T, U>{} and (std::is_same<T, float>{} or std::is_same<T, double>{})) | ||
| { | ||
| eigen_multiply(cmat, amat, bmat); | ||
| } | ||
| else | ||
| { | ||
| std::vector<float> a_buf(amat.get_shape().elements()); | ||
| std::copy(amat.begin(), amat.end(), a_buf.begin()); | ||
| auto amat_flat = make_view( | ||
| amat.get_shape().as_standard().with_type(shape::float_type), a_buf.data()); | ||
|
|
||
| std::vector<float> b_buf(bmat.get_shape().elements()); | ||
| std::copy(bmat.begin(), bmat.end(), b_buf.begin()); | ||
| auto bmat_flat = make_view( | ||
| bmat.get_shape().as_standard().with_type(shape::float_type), b_buf.data()); | ||
|
|
||
| std::vector<float> c_buf(cmat.get_shape().elements(), 0.0f); | ||
| auto c_shape_std = cmat.get_shape().as_standard().with_type(shape::float_type); | ||
| auto cmat_flat = make_view(c_shape_std, c_buf.data()); | ||
|
|
||
| eigen_multiply(cmat_flat, amat_flat, bmat_flat); | ||
|
|
||
| std::copy(c_buf.begin(), c_buf.end(), cmat.begin()); | ||
| } |
There was a problem hiding this comment.
In the Eigen-enabled path, the fallback for non-float/double inputs converts A/B to float and computes in float before copying into C. This breaks exactness for integer GEMMs (e.g., quant_dot with int8/uint8 inputs produces an int32 output per quant_dot::compute_shape), because accumulation in float can round for larger K / values. Please keep integer GEMMs on the naive/integer-accumulation path (or cast to int32/int64 for the multiply) when C is integral or when A/B are integral types.
| if(MIGRAPHX_USE_EIGEN) | ||
| find_package(Eigen3) | ||
| if(EIGEN3_FOUND) | ||
| target_compile_definitions(migraphx PUBLIC MIGRAPHX_USE_EIGEN=1) | ||
| target_compile_definitions(migraphx PRIVATE EIGEN_MPL2_ONLY) | ||
| target_link_libraries(migraphx PRIVATE Eigen3::Eigen) | ||
| else() | ||
| message(STATUS "Eigen not found, disabling MIGRAPHX_USE_EIGEN") | ||
| target_compile_definitions(migraphx PUBLIC MIGRAPHX_USE_EIGEN=0) | ||
| endif() |
There was a problem hiding this comment.
The Eigen enablement check uses if(EIGEN3_FOUND), but find_package(Eigen3) in config mode sets Eigen3_FOUND (and the most robust check is if(TARGET Eigen3::Eigen)). As written, Eigen can be found but still be treated as missing, leaving MIGRAPHX_USE_EIGEN disabled unintentionally.
| @@ -240,15 +217,12 @@ struct ref_quant_gemm | |||
| argument compute(context&, const shape& output_shape, std::vector<argument> args) const | |||
| { | |||
There was a problem hiding this comment.
ref_quant_gemm::compute only multiplies args[0] and args[1]. However op::quant_dot::compute_shape explicitly supports a 4-input form {A, B, scale_A, scale_B} and returns float output for that case. As-is, compiling/evaluating a 4-input quant_dot on the ref target will silently ignore the scale inputs and produce incorrect results; handle args.size()==4 here (or delegate to a shared quant_dot reference implementation).
| { | |
| { | |
| // Handle 4-input form {A, B, scale_A, scale_B} via the op::quant_dot reference implementation | |
| if(args.size() == 4) | |
| { | |
| return op.compute(output_shape, args); | |
| } | |
| // Fallback to existing behavior for 2-input (and other) forms |
| * Added `auto_pad` attribute support for the ONNX `ConvTranspose` operator, supporting `SAME_UPPER`, `SAME_LOWER`, and `VALID` padding modes for static shapes (#4638). | ||
| * Added a dedicated logger for MIGraphX. | ||
| * [Linux] Use HSA API to query number of chiplets for architectures where this is applicable (ex. gfx90a). | ||
| * Added Eigen third party headers for ref GEMMs (#4631). |
There was a problem hiding this comment.
This changelog entry has trailing whitespace at the end of the line, which can trigger formatting/lint issues in some setups. Please remove the extra space after the period.
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
Motivation
speedup ref implementation of GEMMs using Eigen library
Technical Details
Similar to how blaze was used, but Eigen is still currently maintained. It does a copy for datatypes smaller than fp32.
Changelog Category
Add a
CHANGELOG.mdentry for any option other thanNot Applicable