Adding remaining flavors for grouped conv fwd#3436

Draft

wj-laskowski wants to merge 270 commits intodevelopfrom

streamhpc/grouped-conv-fwd-extra-flavors

This pull request is big! We're only showing the most recent 250 commits

Commits on May 15, 2025

Commits on May 16, 2025

Added padding instances for f16xf8xf16, f16xi4xf16
ancahamuraru
committed

Commits on May 19, 2025

Commits on May 22, 2025

example for gemm_universal_bf16

ApoorvaKalyani
authored and
ex-rzr
committed
Adding examples for gemm_wmma instances

ApoorvaKalyani
authored and
ex-rzr
committed
Added the missing parameters

ApoorvaKalyani
authored and
ex-rzr
committed
Fixed review comments and added executable to cmakeLists

ApoorvaKalyani
authored and
ex-rzr
committed
Fixing clang format

ApoorvaKalyani
authored and
ex-rzr
committed
Fixing build erros

ApoorvaKalyani
authored and
ex-rzr
committed
Fixed compilation failure.

ApoorvaKalyani
authored and
ex-rzr
committed
Modified some code as per gemm_universal_examples

ApoorvaKalyani
authored and
ex-rzr
committed
Fixed the gemm specialization error

ApoorvaKalyani
authored and
ex-rzr
committed
Fixed the build errors.

ApoorvaKalyani
authored and
ex-rzr
committed

Commits on May 23, 2025

Commits on May 30, 2025

Make gemm_add_add_wmma to work with DeviceGemmMultipleD_Wmma_CShuffleV3
ex-rzr
committed

Commits on Jun 2, 2025

Commits on Jun 3, 2025

removed unnecessary ck parts from compilation
Zoltan Lakatos
committed
initial gemm_add_multiply instance implementations
Zoltan Lakatos
committed
fixed profiler help message for gemm_add_multiply
Zoltan Lakatos
committed
improved multiply_add profiler layout help
Zoltan Lakatos
committed
fixed template arguments for test instances
Zoltan Lakatos
committed
added test for gemm_add_multiply
Zoltan Lakatos
committed

Commits on Jun 4, 2025

Commits on Jun 5, 2025

log print added to splitk benchmarks
Zoltan Lakatos
committed
revert main cmake comments
Zoltan Lakatos
committed
Merge remote-tracking branch 'origin/feature/multiple-d-gemms' into 8-implement-device_gemm_add_multiply-for-rdna4
Zoltan Lakatos
committed

Commits on Jun 6, 2025

newline change reverted
Zoltan Lakatos
committed

Commits on Jun 10, 2025

added add_fastgelu instances
Zoltan Lakatos
committed

Commits on Jun 11, 2025

revert unintended change in xdl add_fastgelu
Zoltan Lakatos
committed
created gemm_add_add_fastgelu instances
Zoltan Lakatos
committed
created fastegelu instances
Zoltan Lakatos
committed

Commits on Jun 12, 2025

added tests for all splitk fastgelus
Zoltan Lakatos
committed

Commits on Jun 13, 2025

Added tests.
ApoorvaKalyani
committed
multiply_add instances created
Zoltan Lakatos
committed
updates to add_multiply splitk instances
Zoltan Lakatos
committed
splitk xdl test fixes
Zoltan Lakatos
committed

Commits on Jun 17, 2025

added wmma multiply_multiply instances
Zoltan Lakatos
committed
fixed ONLY_XDL_AND_WMMA_KERNELS tag
Zoltan Lakatos
committed

Commits on Jun 18, 2025

Commits on Jun 19, 2025

Commits on Jun 23, 2025

fixed rdna4 instances
Zoltan Lakatos
committed
fixed back compatibility on gfx11
Zoltan Lakatos
committed

Commits on Jun 24, 2025

cleanups
Zoltan Lakatos
committed
fix ckProfiler
Zoltan Lakatos
committed
one more cmake fix
Zoltan Lakatos
committed
added fp8 instances
Zoltan Lakatos
committed

Commits on Jun 26, 2025

Resolve "Implement `device_gemm_bilinear` for RDNA4"
Zoltán Lakatos
committed
Merge branch '63-implement-device_gemm_bilinear-for-rdna4' into 'feature/multiple-d-gemms'
Zoltán Lakatos
committed

Commits on Jun 30, 2025

Merge remote-tracking branch 'origin/feature/multiple-d-gemms' into 8-implement-device_gemm_add_multiply-for-rdna4
Zoltan Lakatos
committed
Merge remote-tracking branch 'origin/feature/multiple-d-gemms' into 64-implement-device_gemm_multiply_multiply_instance-for-rdna4
Zoltan Lakatos
committed
test generalization to handle FP16 shuffle better
Zoltan Lakatos
committed
added missing changes
Zoltan Lakatos
committed

Commits on Jul 1, 2025

Commits on Jul 2, 2025

Commits on Jul 7, 2025

Fixes applied to fix the precision loss.
ApoorvaKalyani
committed

Commits on Jul 8, 2025

Commits on Jul 9, 2025

Commits on Jul 10, 2025

comments on the atomics workaround
Zoltan Lakatos
committed

Commits on Jul 11, 2025

Commits on Jul 14, 2025

Commits on Jul 28, 2025

Merge branch 'develop' into feature/multiple-d-gemms
EnricoDeg
committed

Commits on Aug 4, 2025

Fix bug from merge
EnricoDeg
committed

Commits on Aug 24, 2025

Commits on Aug 26, 2025

Commits on Aug 27, 2025

Commits on Aug 28, 2025

Commits on Aug 29, 2025

Commits on Aug 31, 2025

Add comp and mem instances for vanilla 3D grouped conv fwd. Skipped 2x and part2 instances, can be added later in the same instance lists.
krithalith
committed

Commits on Sep 2, 2025

Commits on Sep 4, 2025

Implement rotating memory and flush cache. Requires ad-hoc buffer size calculations.
krithalith
committed

Commits on Sep 5, 2025

Commits on Sep 15, 2025

Commits on Sep 18, 2025

Commits on Sep 19, 2025

Add wmma scaleadd ab instances to the device factory and add a completely new scaleadd_ab gtest test for wmma cshufflev3 and xdl. Currently there is no profiler for scaleadd_ab so I made my own ins…
krithalith
committed

Commits on Sep 22, 2025

Add support for V3 pipeline (tested). To be able to support num_loop < 3 we need the fixes from the batched gemm gemm MR which was already merged upstream, so just need to rebase or merge.
krithalith
committed

Commits on Sep 23, 2025

Merge remote-tracking branch 'origin/develop' into 65-grouped-conv-fwd-wmma
krithalith
committed

Commits on Oct 1, 2025

Commits on Oct 2, 2025

Commits on Oct 3, 2025

Commits on Nov 10, 2025

Commits on Nov 19, 2025

Commits on Dec 15, 2025