[CK][CK Tile] Drop profiler for experimental builder codegen
(#8573)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
## Motivation
Switch to dispatcher profiler for ck tile conv.
## Technical Details
- Switch to dispatcher profiler for ck tile conv.
- Drop profiler for experimental codegen
- Minor fixes for bwd data printing
- Minor fixes for 3d conv in dispatcher codegen
## Test Plan
test_grouped_conv*tile
## Test Result
Passed
## Submission Checklist
- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[CK Tile] Rule-based configuration generation in CK
Dispatcher codegen (#8157)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
## Motivation
The CK Tile Dispatcher code generation for CK Tile Profiler relies on
flat JSON files to list the generated configurations. This approach has
the following problems
- The JSON files are verbose
- The JSON files get easily out of sync with the CK Builder .config
files from which they were generated from.
- The JSON file based configuration make it hard to list explicitly the
rules that govern the instance generation.
## Technical Details
Replaced the JSON files with a rule based configuration. To preserve the
existing functionality, the `profiler` and the `tests` instance sets are
generated directly from the CK Builder config files. The JSON config
files are removed from source control, and the "on-the-fly" generation
guarantees that the Dispatcher codegen uses up to date configurations.
This is PR introduces six different rule sets for the CK Tile Dispatcher
code generation
1. `profiler`: matches with the old JSON set of profiler configurations.
2. `tests`: matches with the old JSON set of tests configurations.
3. `full`: full configuration set created from a rule-based config
selection
4. `full-tests`: a subset of `full` for generating configurations for
convolution integration tests.
5. `tiny`: a subset of `full-tests` to produce the minimal set of
configurations to test the Dispatcher codegen.
6. `default`: the default rules, which corresponds to the existing
heuristic rules for configuration selection. This ensures that ML based
kernel selection doesn't get broken.
The main use of the `full` rule set is to define a reasonable solution
space for the possible implicit GEMM configurations. We start from the
configurations that allowed by the device architecture. The `full` rule
set defines the relevant tile sizes for each convolution direction. From
the tile size we have a curated mapping to the number of waves over the
different GEMM axes, i.e., we describe how many waves each GEMM
dimensions corresponds to. The GEMM-K wave tile dimension can be
computed from the other parameters and does not need to be listed
explicitly.
An orthogonal axis to the tiling strategy is the vectorization strategy.
This mainly defined by the data type and hardware as in general, we want
to use the maximum possible load widths. The maximum sizes for each
convolution direction variant are defined by the implicit GEMM matrix
dimensions. For cases where have a low number of channels per
convolution group, we need smaller vector load sizes. These are captured
by the `VecStrategy` enumeration in the codegen rules.
The problem with the rule based configuration selection is that we "over
generate" configurations. The old JSON configurations compose
approximately 25% of all configuration that the `full` rule set creates.
The additional configurations are valid, but they many not provide any
performance benefits. Hence, we keep the `profiler` and `tests` rule set
for now to avoid building an excessive amount configurations by default.
The `full` rule set can be taken into use by specifying CMake
configuration flag `-D DISPATCHER_RULE_SET=full`. By default, the
`tests` rule set is used, i.e., we don't change the existing bahaviour.
## Test Plan
Added a new stage in the CI/CD pipeline that ensures the Dispatcher
codegen rules are up to date. Otherwise the functionality is covered by
the existing CI/CD tests. There are no functional changes to the
convolution kernels. Only how the different instances are generated.
## Test Result
If the CK Tile conv instances build without errors, the Dispatcher
codegen is generating valid code. If all tests in CI/CD pipeline are
passing, the Dispatcher codegen generates valid instances.
## Submission Checklist
- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[CK] Add support for large tensor index handling into conv
bwd data (#8518)
## Motivation
<!-- Explain the purpose of this PR and the goals it aims to achieve.
-->
## Technical Details
<!-- Explain the changes along with any relevant GitHub links. -->
## Test Plan
<!-- Explain any relevant testing done to verify this PR. -->
## Test Result
<!-- Briefly summarize test outcomes. -->
## Submission Checklist
- [ ] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[CK] Grouped Convolution Global Load/Store instances
## Motivation
Support global load and store in grouped convolutions using instance
factory.
## Technical Details
- add new instances for each direction
- add new tests for large cases
## Test Plan
New test for large cases
## Test Result
pending
## Submission Checklist
- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
AICK-1255
[CK] [MIOPEN] Split convolution library by layout
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
# Split Composable Kernel convolution operations by data layout
TLDR:
1. This is a reorganization of files, folders, and CMakeLists for
convolution kernels and facilitates a splitting of the convolution
library into layouts.
2. The speedup can range anywhere between 15-40% depending on the target
architecture for miopen only builds of CK. For TheRock nightly builds of
CK, which includes both miopen and hip tensor kernel instances, this
constituted in a 10% decrease in compile time for gfx1100.
## Overview
Based on https://github.com/ROCm/composable_kernel/pull/3010/ (except
keeping 1 static library)
## What MIOpen Actually Uses
MIOpen **exclusively uses:
- **NHWGC** for all 2D convolutions
- **NDHWGC** for all 3D convolutions
This is because MIOpen's tensor descriptors natively use channel-last,
group-aware formats.
## Key Changes
### 1. Layout-Based Directory Structure
Reorganized convolution instance files from flat per-operation to
hierarchical layout-based structure. For example:
**Before:**
grouped_conv2d_fwd/
├── device_grouped_conv2d_fwd_xdl_nhwgc_*.cpp (MIOpen-required)
├── device_grouped_conv2d_fwd_xdl_gnhwc_*.cpp (optional)
└── device_grouped_conv2d_fwd_xdl_ngchw_*.cpp (optional)
**After:**
grouped_conv2d_fwd/
├── nhwgc/ ← MIOpen-required
│ ├── xdl/device_grouped_conv2d_fwd_xdl_*.cpp
│ └── wmma/device_grouped_conv2d_fwd_wmma_*.cpp
├── gnhwc/ ← Optional (excluded with MIOPEN_REQ_LIBS_ONLY)
└── ngchw/ ← Optional (excluded with MIOPEN_REQ_LIBS_ONLY)
### 2. Preserved Umbrella Library
As before, all convolution operations are consolidated into a single
static `device_conv_operations` library:
- Aggregates layout-specific instance object files via
`ADD_CONV_LAYOUT_INSTANCES` macro
- **Default build:** Includes all layouts (NHWGC + GNHWC + NGCHW +
NDHWGC + GNDHWC + NGCDHW)
- **MIOpen build (`MIOPEN_REQ_LIBS_ONLY=ON`):** Includes only NHWGC and
NDHWGC layouts
### 3. Binary Size Reduction
When building with `MIOPEN_REQ_LIBS_ONLY=ON`:
**Layouts Included (26 targets):**
- 7× NHWGC instances (2D operations + variants)
- 19× NDHWGC instances (3D operations + variants)
**Layouts Excluded (16 targets):**
- 3× GNHWC instances (2D operations)
- 3× NGCHW instances (2D operations)
- 3× GNDHWC instances (3D operations)
- 3× NGCDHW instances (3D operations)
- 2× GNWC instances (1D operations)
- 1× NWGC instance (1D operations)
- 1× additional NHWGC instance (grouped_conv1d_fwd, not needed by
MIOpen)
This represents a **~38% reduction in instance targets** (16 excluded
out of 42 total
layout-specific targets).
### Testing
- ✅ All existing CK tests link against the umbrella library
- ✅ MIOpen links successfully with the reduced umbrella library
- ✅ Profiler builds with all layout-specific targets explicitly listed
Notes from the Author:
Since this refactor moved most of the convolution files further into
subdirectories, I concentrated on ensuring that no source files were
excluded, including sharded sources: Targets are correctly migrated — no
missing targets, no shard count mismatches.
[CK] Fix latest build issues with staging compiler.
## Motivation
Fixing new warnings with staging compiler.
## Technical Details
<!-- Explain the changes along with any relevant GitHub links. -->
## Test Plan
<!-- Explain any relevant testing done to verify this PR. -->
## Test Result
<!-- Briefly summarize test outcomes. -->
## Submission Checklist
- [ ] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[CK] suppress compiler warnings while building pytorch. (#7760)
## Motivation
Recently added compiler flags that are required to suppress false
warnings by latest staging compiler are not recognized by older compiler
versions and are triggering an avalanche of warnings. Previous attempt
to suppress them by using -Wno-unknown-warning-option flag didn't help,
because that flag wasn't recognized either and just added more warnings.
I've verified that current approach by checking the clang version
actually works as intended and makes the warnings go away.
## Technical Details
<!-- Explain the changes along with any relevant GitHub links. -->
## Test Plan
<!-- Explain any relevant testing done to verify this PR. -->
## Test Result
<!-- Briefly summarize test outcomes. -->
## Submission Checklist
- [ ] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[CK_TILE] Add depthwise conv2d forward kernel (FP16/FP32) (#6838)
## Motivation
CK currently has no kernel optimized for depthwise convolution
(G=C_in=C_out, C=K=1 per group) and existing generic paths perform
poorly for this workload. This PR adds a dedicated depthwise conv
forward kernel in CK Tile.
## Technical Details
Adds a dedicated depthwise conv2d forward op to CK Tile that performs
direct convolution rather than falling back to the generic GEMM path.
The kernel is templatized by filter size, stride, and data type, and
compiled into ~60 instances covering common configurations (kernel
3/5/7/9, stride 1/2, FP16/FP32). Supports both CDNA (gfx942/gfx950) and
RDNA (gfx1100/gfx1200) architectures.
## Test Plan
- [x] Correctness and performance validated on gfx942, gfx950, and
gfx1100, with ckProfiler `grouped_conv_fwd` as baseline.
- [ ] MI300A (gfx942) and gfx1200 validation.
## Submission Checklist
- [x ] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
AICK-1137
---------
Co-authored-by: GenDu <Gen.Du@amd.com>
[CK] add composable kernel support on gfx1250 (#6978)
## Motivation
Add composable kernel support on gfx1250.
## Technical Details
<!-- Explain the changes along with any relevant GitHub links. -->
## Test Plan
<!-- Explain any relevant testing done to verify this PR. -->
## Test Result
<!-- Briefly summarize test outcomes. -->
## Submission Checklist
- [ ] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
---------
Co-authored-by: Qun Lin <qlin@amd.com>
Co-authored-by: jialuo12_amdeng <jia.luo@amd.com>
Co-authored-by: Andriy Roshchenko <andriy.roshchenko@amd.com>
Co-authored-by: hsivasun_amdeng <haresh.sivasuntharampillai@amd.com>
[CK] Suppress new staging compiler errors (#7384)
## Motivation
This should make new builds with staging compiler pass.
## Technical Details
<!-- Explain the changes along with any relevant GitHub links. -->
## Test Plan
<!-- Explain any relevant testing done to verify this PR. -->
## Test Result
<!-- Briefly summarize test outcomes. -->
## Submission Checklist
- [ ] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[CK] Fix latest batch of staging compiler warnings (#7111)
## Motivation
Suppress the new batch of clang lifetimebound and invalidation warnings
with the latest staging compiler.
## Technical Details
<!-- Explain the changes along with any relevant GitHub links. -->
## Test Plan
<!-- Explain any relevant testing done to verify this PR. -->
## Test Result
<!-- Briefly summarize test outcomes. -->
## Submission Checklist
- [ ] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
[CK][CK TILE] Add has hot loop check for pipeline v1
## Motivation
Add has hot loop check for pipeline v1 (v1 basic and v1 basic async).
Enable more tests which have been fixed by this change.
## Technical Details
Hot loop has been executed without num loop check.
## Test Plan
test_grouped_convnd_fwd_tile
## Test Result
Passed
## Submission Checklist
- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
AICK-651
AICK-663
[CK] CK Tile grouped convolution direct load
## Motivation
CK Tile grouped convolution forward direct load support.
## Technical Details
Basic pipeline for direct load and new instances for forward for v1 and
v4 pipelines.
## Test Plan
test_grouped_convnd_fwd_tile
## Test Result
CI pending
## Submission Checklist
- [x] Look over the contributing guidelines at
https://github.com/ROCm/ROCm/blob/develop/CONTRIBUTING.md#pull-requests.
AICK-130
* Adding remaining flavors for grouped conv fwd
As titled. Following variants are added:
- grouped_conv2d_fwd_dynamic_op
- grouped_conv3d_fwd_dynamic_op
- grouped_conv3d_fwd_bilinear
- grouped_conv3d_fwd_convscale
- grouped_conv3d_fwd_convinvscale
- grouped_conv3d_fwd_convscale_add
- grouped_conv3d_fwd_convscale_relu
- grouped_conv3d_fwd_scale
- grouped_conv3d_fwd_combconvscale
- grouped_conv3d_fwd_scaleadd_scaleadd_relu
* Fix incomplete parsing of types from source names in add_instance_library() cmakelists function so we don't build f8 on RDNA3.
* Do not build f8 / bf8 only flavor tests on RDNA3
* Make sure we have proper generic instances for all instance lists related to the post-ces extra flavors, with scalarPerVector = 1. Then disable all but one generic instance per instance list to reduce compile time.
* Post rebase fix: Template parameters for Grouped Conv Fwd Device Impl got tweaked upstream.
* adding int8 and fp16 overloads to the elementwise operations
* fixed copilot nits
* Addressing review comments:
- removed unnecessary examples for dynamic op
- removed unnecessary conv specalizations for all the flavors
- removed spurious bilinear and scale source files
* clang-format
* reduced no of tests
---------
Co-authored-by: Wojciech Laskowski <wojciech.laskowski@streamhpc.com>
* ck-builder: restructure testing conv
In order to prepare for bwd of conv testing, this commit moves some
files and types around so that we can reuse ckt::Args for both forward
and backwards convolution.
* ck-builder: decouple fwd_ck.hpp and fwd_reference.hpp from fwd.hpp
This will allow us to more easily include fwd.hpp from backwards
definitions, which is required for initializing bwd values.
* ck-builder: fix layout of test_ckb_conv_bwd_weight_xdl_cshuffle_v3
Turns out that the supplied layout isn't actually supported...
* ck-builder: ck and reference conv integration for bwd weight
* ck-builder: ck bwd weight execution test
* ck-builder: ckt::run support for ck-tile bwd weight
* ck-builder: ck tile bwd weight execution test
* ck-builder: extra debug printing in MatchesReference
* ck-builder: make ckt::run return RunResult
This type is more convenient than std::tuple, as it will allow us to
use google test matchers with this in the future.
* ck-builder: RunResult matcher
Using EXPECT_THAT(..., SuccessfulRun()) will generate a check and a nice error
message about how and why running an algorithm failed.
* ck-builder: doc fixes
* ck-builder: add missing headers
* Updated the set of tests for FP16
* Fix typo
* Moved f16xi4 test under the correct data layout group
* example for gemm_universal_bf16
* Adding examples for gemm_wmma instances
* Added the missing parameters
* Fixed review comments and added executable to cmakeLists
* Fixing clang format
* Fixing build erros
* Fixed compilation failure.
* Modified some code as per gemm_universal_examples
* Fixed the gemm specialization error
* Fixed the build errors.
* Fix strides of a/b_thread_desc
The descriptors are larger than needed (even though the compiler don't alloc registers for unused values).
* Load in M/NRepeat dims with thread copy's slice instead of a loop
* Clone BlockwiseGemmXdlops_pipeline_v1 for WMMA implementation
* Implement Intrawave and Interwave variants of pipeline v1
* Add instances for Interwave and Intrawave v1
* Add instances with ABlockLdsExtraM and BBlockLdsExtraN = 0
* Remove instances that are too slow (mostly because of register spilling)
* Add a workaround for fp8/bf8->f32 packed conversion issue
* Add instances for Interwave and Intrawave v1
* Enable profiling of mixed precision with f8 and int4 on WMMA
* Fix segfault in profiler when B is pk_i4_t
b_device_buf's size in bytes is larger than b_k_n_permute so b_device_buf.ToDevice reads out-of-bounds.
* Remove instances that are too slow (mostly because of register spilling)
* Add missing add_device_gemm_wmma_universal_f8_f8_bf16 declarations
* Add test case for bf16_i4
* Add missing Regular tests
* Add test_gemm_universal_xdl/wmma_fp16 to REGRESSION_TESTS
They take more than 30 seconds
* Fix a bug that fp16_i4 validation passes only with PermuteB
A permutation required by conversion from pk_i4_t to half_t does not
depend on PermuteB, they can be used independently.
* Use PermuteB with f16_i4 in most instances (as xdl)
Some instances use PermuteB = false for checking correctness.
See also the previous commit.
* Fix cache flushing for pk_i4
* Add mixed precision examples
* Disable all tests and instances with f8 on gfx11
Even though f8_f16 and f16_f8 don't require f8 WMMA instructions,
gfx11 still lacks hardware instructions for fast f8->f32 conversion.
* Add FP16 KM_NK and KM_KN test suites for XDL
These tests were added to common .inc for better testing of WMMA instances
* Support multiple D in GridwiseGemm_wmma_cshuffle_v3
DeviceGemm_Wmma_CShuffleV3 is changed for new template parameters.
* Use ThreadGroupTensorSliceTransfer_v7r3
* Clone for device_gemm_wmma_cshuffle_v3.hpp for future Multiple D support
* Clone example/65_gemm_multiply_multiply/gemm_add_add_xdl_fp16.cpp for wmma
* Implement DeviceGemmMultipleD_Wmma_CShuffleV3
* Make gemm_add_add_wmma to work with DeviceGemmMultipleD_Wmma_CShuffleV3
* Prepare gemma_add tests for adding wmma
* Add gemm_add_fastgelu instances and test
* Add a special wrapper to use DeviceGemmMultipleD_Wmma_CShuffleV3 with old API
ckProfiler uses DeviceGemmMultipleD (tests also call its functions), the wrapper allows to use
DeviceGemmMultipleDSplitK instances there.
* removed unnecessary ck parts from compilation
* initial gemm_add_multiply instance implementations
* fixed profiler help message for gemm_add_multiply
* improved multiply_add profiler layout help
* fixed template arguments for test instances
* added test for gemm_add_multiply
* Support multiple D in GridwiseGemm_wmma_cshuffle_v3
DeviceGemm_Wmma_CShuffleV3 is changed for new template parameters.
* Use ThreadGroupTensorSliceTransfer_v7r3
* Clone for device_gemm_wmma_cshuffle_v3.hpp for future Multiple D support
* Clone example/65_gemm_multiply_multiply/gemm_add_add_xdl_fp16.cpp for wmma
* Implement DeviceGemmMultipleD_Wmma_CShuffleV3
* Make gemm_add_add_wmma to work with DeviceGemmMultipleD_Wmma_CShuffleV3
* Prepare gemma_add tests for adding wmma
* Add gemm_add_fastgelu instances and test
* Add a special wrapper to use DeviceGemmMultipleD_Wmma_CShuffleV3 with old API
ckProfiler uses DeviceGemmMultipleD (tests also call its functions), the wrapper allows to use
DeviceGemmMultipleDSplitK instances there.
* switched to splitK interface
* log print added to splitk benchmarks
* revert main cmake comments
* newline change reverted
* added add_fastgelu instances
* revert unintended change in xdl add_fastgelu
* created gemm_add_add_fastgelu instances
* created fastegelu instances
* added tests for all splitk fastgelus
* Added tests.
* multiply_add instances created
* updates to add_multiply splitk instances
* splitk xdl test fixes
* added wmma multiply_multiply instances
* fixed ONLY_XDL_AND_WMMA_KERNELS tag
* Added gemm_add examples for wmma v1 and v3
* fixed / workarounded i8 instances
* Modified the v3 code to added one fp16 bxdl instance.
* added bf16 xdl instance.
* adding gemm_add wmma_cshuffle and other support
(cherry picked from commit ec447e7f564095ea969eddc39ec77b843aa52976)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* add instances into camkelists
(cherry picked from commit 23bf2d2771c939ea3ca7f493433c55255bffd08e)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* This is work in progress, edited the template parameters in order to build
(cherry picked from commit b4fde8a3314cb44659c4bbda35f1a0133c63dc41)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* temp work saved, changed the BDataType to f16 or bf16 since wmma currently not support non-equal A and B datatype
(cherry picked from commit 22fbd68f1db458ab50780a394ee2544c7a1484d1)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* added datatype and use clang-format-12
(cherry picked from commit ae4e853682ef1bb27784b2f965b4a66b3751ceec)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* Fixing build errors
* Added instances for v3
* Adding instances and executables
* Code update of template parameters modified.
* Renamed file.
* Added tests.
* resolved error tests.
* Fixing build errors
* Updated comments
* removed the changes as per the MR review comment.
* Updated tests.
* fp8 instances - not tested
* Restored the Cmake file that was reverted by mistake during rebase.
* fixed wmma_op test
* Updated comments.
* Updated the template parameter description
* fixed rdna4 instances
* fixed back compatibility on gfx11
* cleanups
* fix ckProfiler
* one more cmake fix
* added fp8 instances
* Updated tests to ad BF16 instances as per review comment
* Added include file and cleaned up(as per review comment)
* Updated and optimized the example code for all types.
* Fixed clang format
* Resolve "Implement `device_gemm_bilinear` for RDNA4"
* test generalization to handle FP16 shuffle better
* added missing changes
* Added bf16 wmma instance for add_relu
* Added f16 wmma instance and corrected bf16 instance errors.
* Added instances to Cmake
* Modified the template parameters to make the instances work.
* Fixed typo in profiler
* Added v3 instances for gemm_add_relu
* addressed core review comments
* Added test for gemm_add_relu wmma instance
* Cleaned up the code.
* Added examples for gemm_add_relu
* Fixing typo to resolve build errors.
* Fixes applied to fix the precision loss.
* fix billinear test after merge
* Removed the old wmma instances.
* Added wrapper and renamed the wmma_v3 instances
* Updated copyrights and added wrappers.
* Fixes applied according to review comments
* Apply 1 suggestion(s) to 1 file(s)
Co-authored-by: Robin Voetter <robin@streamhpc.com>
* Removed the old wmma instances.
* Updated wrapper for the v3 instances
* removed the old wmma examples
* Renamed the v3 instances
* Deleted the gtest file added by mistake.
* Updated thge profiler with wrapper
* Fixed test errors.
* Fixed the review comments
* Fixed the if condition MACROS.
* REVERTED THE PROFILER CHANGES
* Revert "REVERTED THE PROFILER CHANGES"
This reverts commit 21cb98546c.
* Revert "Fixed test errors."
This reverts commit 13efcc6fe1.
* Revert "Updated thge profiler with wrapper"
This reverts commit 536f86661d.
* Added missing wrapper instances
* Updated copyrights.
* Fixed typo.
* Fixed copyrights.
* Updated copyrights.
* updated copyrights.
* comments on the atomics workaround
* fixed cmake comment
* Fix bug from merge
* clang-format-18
* Fix compilation error
* multi_abd wmma support:
- Add multiple A and B support to multiple D implementation (gridwise level)
- Add multi_abd GEMM (device level)
- Add instances (xdl parity)
- Add tests (both xdl and wmma)
- Add examples
- Add ckProfiler support (both xdl and wmma)
* Fix bug in device print function
* Fix unused template parameter
* Add support for fwd conv in gridwise implementation. Identical to run function for bwd data.
* Initial device implementation for grouped conv fwd multiABD wmma cshuffleV3. Functional but needs some fixups and extra features in the future.
* Make relevant profilers print the number of valid instances to aid testing.
* Add instances for all vanilla 2D and 3D flavors for f16 and bf16, only one instance per instance list to save compile time for now. Also added incomplete set of comp instances and bias_clamp for f16 2D, just to make sure the multiple-D aspects of the device implementation are working.
* Reset output buffer after each run in profile_grouped_conv_fwd_impl().
* Disable sharding for the new instances for now, has tendency to lead to linker errors on repeat builds.
* Add CTranspose optimization for NCHW cases just like in xdl cshuffle non-v3 device implementation.
* Add instances for all 8-bit 3D vanilla grouped conv fwd types, including mixed types but with the exception of deprecated f16 comp fp8. Adapt test so we can test 8-bit and mixed types.
* Add int8 instances for 2D vanilla grouped conv fwd all layouts.
* Implement merged groups in device impl and add instances for merged groups 3D vanilla conv fwd
* Add merged groups instances for all 2D vanilla grouped conv fwd types and layouts.
* Implement multi-AB support for grouped conv fwd and add example.
* Add 1D instances
* Add D layout tests to IsSupportedArgument()
* Add comp and mem instances for all vanilla 2D grouped conv fwd types. Skipping "x2" and "part2" instance lists, can be added later without special names if necessary.
* 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.
* Add some more tests for vanilla grouped conv fwd
* Add 2D bias clamp instances and tests
* Add 3D bias clamp instances and tests
* Add 2D and 3D clamp instances and tests
* Unify problem sizes across vanilla and clamp flavor tests
* Clean up device implementation: remove old todos, remove unnecessary comments and print statements, tweak description, wrap all prints in env check.
* Implement rotating memory and flush cache. Requires ad-hoc buffer size calculations.
* Remove wmma fp8 and bf8 instances when not targetting gfx12
* Add newer instances to DEVICE_INSTANCES so the main ckProfiler can build
* Remove old years for newly created files.
* No need to time kernels for now.
* Fixup comments
* Pass struct args to Gridwise Run() function by reference.
* Don't use workspace memory in the case where A needs explicit transposition but B does not.
* Move calculation of rotating memory buffer sizes to Argument member functions.
* After the convolution to gemm transformation, the resulting 2D tensor descriptors are not necessarily RowMajor or ColumnMajor, so things should not rely on this distinction. Therefore, pass all RowMajor to the Gridwise and use a special version of CheckValidity that does not rely on 2D tensor layouts.
* Unify xdl and wmma example code for grouped conv fwd scaleadd ab
* Go back to passing RCR 2D tensor layouts to gridwise gemm, and use CRC for the CTranspose case. Also remove the special convolution version of checkValidity(). It seems like no matter what 2D tensor layouts you pass to the gridwise gemm, and no matter if you are using extraMN, and no matter if you are using the convolution version of checkvalidity, the results of all tests are the same.
* 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 inside the test. Furthermore for XDL only the (NDHWGC, GKZYXC, NDHWGK) layout combination existed in the instance factory so that is the only one I added for wmma cshufflev3 and the gtest test as well. Another layout is tested in example 62, for xdl and wmma cshufflev3.
* 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.
* Small post-merge fixup, everything seems to work.
* Do not build or run Xdl operations with Wmma backend for now. Will be reverted before upstreaming.
* Extend scaleadd_ab instance lists
* Extend merged groups instance lists, including adaptations of xdl "2x" instances.
* Extend "comp" instance lists, including "2x" and "part2" instances. 2x instances disabled for now since they do not compile.
* Extend "mem" instance lists.
* Extend regular instance lists.
* Fixup comments and ignored kernel arg name
* Properly use the splitN offsets for D tensors in the gridwise Run() function. Was necessary to pass the bias_clamp_large_cases test.
* Make sure all strides in ComputePtrOffset are at least value initialized to avoid undefined strides. Not convinced this struct is properly initialized in other code / future code.
* Re-enable sharding for wmma cshufflev3 instances
* Post merge fix to vanilla test
* Optionally allow num_k_loop <= PrefetchStages in gridwise CheckValidity. Use this for grouped conv fwd but not in general.
* Remove spurious ck_tile changes that were presumably introduced somewhere in the repeated merging from develop.
* Post-merge fixes. Make sure the new gridwise gemm wmma v3 common Run function can be used. Remove splitK, and forceThreadTileTransfer for now. Also add CShuffle epilogue argument.
* Disable FP8 / BF8 testing on CDNA1/2, it doesn't work anymore and needs to be either fixed or removed.
* Re-enable old wmma instances
* Re-enable Linqun's Xdl Wmma instances
* Small post-merge fixes
* Fix copyright headers
* Remove commented code snippet in gridwise
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com>
* Limit the explicit cast added in threadwise_tensor_slice_transfer_v7r3 to only be used for f8, just in case it hurts performance.
* Adding tuned instace list for groupoed conv fwd (#3288)
Following flavors are updated with tuned instance list:
- grouped_conv2d_fwd
- grouped_conv2d_fwd_bias_clamp
- grouped_conv2d_fwd_clamp
- grouped_conv3d_fwd
- grouped_conv3d_fwd_bias_clamp
- grouped_conv3d_fwd_clamp
- grouped_conv3d_fwd_scaleadd_ab
Re-factored instance selection:
- removed all the unnecessary instance tuples (comp/mem/16x16/generic)
- removed all unnecessary layouts and data types
* Do not use std::remove_cvref_t, does not exist in C++17, use custom one.
* Splitting grouped conv fwd instances (#3449)
* Disable unnecessary and failing tests related to experimental CK builder
* Disable unnecessary ck builder experimental tests fully
* Adding extra flavors for grouped conv fwd
As titled. Following variants are added:
- grouped_conv3d_fwd_bilinear
- grouped_conv3d_fwd_scale
* fix cmake error
* Fix failing int8 test for DeviceGroupedConvFwdMultipleABD_Xdl_CShuffle
---------
Co-authored-by: Anca Hamuraru <anca@streamhpc.com>
Co-authored-by: apoorva <apoorva@streamhpc.com>
Co-authored-by: Anton Gorenko <anton@streamhpc.com>
Co-authored-by: Zoltan Lakatos <zoltan.lakatos@streamhpc.com>
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
Co-authored-by: Robin Voetter <robin@streamhpc.com>
Co-authored-by: Enrico Degregori <enrico@streamhpc.com>
Co-authored-by: Kiefer van Teutem <kiefer.van.teutem@streamhpc.com>
Co-authored-by: Kiefer van Teutem <50830967+krithalith@users.noreply.github.com>
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com>
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
Co-authored-by: illsilin_amdeng <Illia.Silin@amd.com>
* Padding not supported for when BDataType is pk_i4_t. Added fix for correct check and removed padding instances.
* Fixed typos
* Updated the set of tests for FP16
* Updated the set of tests for FP16
* Fix typo
* Moved f16xi4 test under the correct data layout group
* example for gemm_universal_bf16
* Adding examples for gemm_wmma instances
* Added the missing parameters
* Fixed review comments and added executable to cmakeLists
* Fixing clang format
* Fixing build erros
* Fixed compilation failure.
* Modified some code as per gemm_universal_examples
* Fixed the gemm specialization error
* Fixed the build errors.
* Fix strides of a/b_thread_desc
The descriptors are larger than needed (even though the compiler don't alloc registers for unused values).
* Load in M/NRepeat dims with thread copy's slice instead of a loop
* Clone BlockwiseGemmXdlops_pipeline_v1 for WMMA implementation
* Implement Intrawave and Interwave variants of pipeline v1
* Add instances for Interwave and Intrawave v1
* Add instances with ABlockLdsExtraM and BBlockLdsExtraN = 0
* Remove instances that are too slow (mostly because of register spilling)
* Add a workaround for fp8/bf8->f32 packed conversion issue
* Add instances for Interwave and Intrawave v1
* Enable profiling of mixed precision with f8 and int4 on WMMA
* Fix segfault in profiler when B is pk_i4_t
b_device_buf's size in bytes is larger than b_k_n_permute so b_device_buf.ToDevice reads out-of-bounds.
* Remove instances that are too slow (mostly because of register spilling)
* Add missing add_device_gemm_wmma_universal_f8_f8_bf16 declarations
* Add test case for bf16_i4
* Add missing Regular tests
* Add test_gemm_universal_xdl/wmma_fp16 to REGRESSION_TESTS
They take more than 30 seconds
* Fix a bug that fp16_i4 validation passes only with PermuteB
A permutation required by conversion from pk_i4_t to half_t does not
depend on PermuteB, they can be used independently.
* Use PermuteB with f16_i4 in most instances (as xdl)
Some instances use PermuteB = false for checking correctness.
See also the previous commit.
* Fix cache flushing for pk_i4
* Add mixed precision examples
* Disable all tests and instances with f8 on gfx11
Even though f8_f16 and f16_f8 don't require f8 WMMA instructions,
gfx11 still lacks hardware instructions for fast f8->f32 conversion.
* Add FP16 KM_NK and KM_KN test suites for XDL
These tests were added to common .inc for better testing of WMMA instances
* Support multiple D in GridwiseGemm_wmma_cshuffle_v3
DeviceGemm_Wmma_CShuffleV3 is changed for new template parameters.
* Use ThreadGroupTensorSliceTransfer_v7r3
* Clone for device_gemm_wmma_cshuffle_v3.hpp for future Multiple D support
* Clone example/65_gemm_multiply_multiply/gemm_add_add_xdl_fp16.cpp for wmma
* Implement DeviceGemmMultipleD_Wmma_CShuffleV3
* Make gemm_add_add_wmma to work with DeviceGemmMultipleD_Wmma_CShuffleV3
* Prepare gemma_add tests for adding wmma
* Add gemm_add_fastgelu instances and test
* Add a special wrapper to use DeviceGemmMultipleD_Wmma_CShuffleV3 with old API
ckProfiler uses DeviceGemmMultipleD (tests also call its functions), the wrapper allows to use
DeviceGemmMultipleDSplitK instances there.
* removed unnecessary ck parts from compilation
* initial gemm_add_multiply instance implementations
* fixed profiler help message for gemm_add_multiply
* improved multiply_add profiler layout help
* fixed template arguments for test instances
* added test for gemm_add_multiply
* Support multiple D in GridwiseGemm_wmma_cshuffle_v3
DeviceGemm_Wmma_CShuffleV3 is changed for new template parameters.
* Use ThreadGroupTensorSliceTransfer_v7r3
* Clone for device_gemm_wmma_cshuffle_v3.hpp for future Multiple D support
* Clone example/65_gemm_multiply_multiply/gemm_add_add_xdl_fp16.cpp for wmma
* Implement DeviceGemmMultipleD_Wmma_CShuffleV3
* Make gemm_add_add_wmma to work with DeviceGemmMultipleD_Wmma_CShuffleV3
* Prepare gemma_add tests for adding wmma
* Add gemm_add_fastgelu instances and test
* Add a special wrapper to use DeviceGemmMultipleD_Wmma_CShuffleV3 with old API
ckProfiler uses DeviceGemmMultipleD (tests also call its functions), the wrapper allows to use
DeviceGemmMultipleDSplitK instances there.
* switched to splitK interface
* log print added to splitk benchmarks
* revert main cmake comments
* newline change reverted
* added add_fastgelu instances
* revert unintended change in xdl add_fastgelu
* created gemm_add_add_fastgelu instances
* created fastegelu instances
* added tests for all splitk fastgelus
* Added tests.
* multiply_add instances created
* updates to add_multiply splitk instances
* splitk xdl test fixes
* added wmma multiply_multiply instances
* fixed ONLY_XDL_AND_WMMA_KERNELS tag
* Added gemm_add examples for wmma v1 and v3
* fixed / workarounded i8 instances
* Modified the v3 code to added one fp16 bxdl instance.
* added bf16 xdl instance.
* adding gemm_add wmma_cshuffle and other support
(cherry picked from commit ec447e7f564095ea969eddc39ec77b843aa52976)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* add instances into camkelists
(cherry picked from commit 23bf2d2771c939ea3ca7f493433c55255bffd08e)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* This is work in progress, edited the template parameters in order to build
(cherry picked from commit b4fde8a3314cb44659c4bbda35f1a0133c63dc41)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* temp work saved, changed the BDataType to f16 or bf16 since wmma currently not support non-equal A and B datatype
(cherry picked from commit 22fbd68f1db458ab50780a394ee2544c7a1484d1)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* added datatype and use clang-format-12
(cherry picked from commit ae4e853682ef1bb27784b2f965b4a66b3751ceec)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* Fixing build errors
* Added instances for v3
* Adding instances and executables
* Code update of template parameters modified.
* Renamed file.
* Added tests.
* resolved error tests.
* Fixing build errors
* Updated comments
* removed the changes as per the MR review comment.
* Updated tests.
* fp8 instances - not tested
* Restored the Cmake file that was reverted by mistake during rebase.
* fixed wmma_op test
* Updated comments.
* Updated the template parameter description
* fixed rdna4 instances
* fixed back compatibility on gfx11
* cleanups
* fix ckProfiler
* one more cmake fix
* added fp8 instances
* Updated tests to ad BF16 instances as per review comment
* Added include file and cleaned up(as per review comment)
* Updated and optimized the example code for all types.
* Fixed clang format
* Resolve "Implement `device_gemm_bilinear` for RDNA4"
* test generalization to handle FP16 shuffle better
* added missing changes
* Added bf16 wmma instance for add_relu
* Added f16 wmma instance and corrected bf16 instance errors.
* Added instances to Cmake
* Modified the template parameters to make the instances work.
* Fixed typo in profiler
* Added v3 instances for gemm_add_relu
* addressed core review comments
* Added test for gemm_add_relu wmma instance
* Cleaned up the code.
* Added examples for gemm_add_relu
* Fixing typo to resolve build errors.
* Fixes applied to fix the precision loss.
* fix billinear test after merge
* Removed the old wmma instances.
* Added wrapper and renamed the wmma_v3 instances
* Updated copyrights and added wrappers.
* Fixes applied according to review comments
* Apply 1 suggestion(s) to 1 file(s)
Co-authored-by: Robin Voetter <robin@streamhpc.com>
* Removed the old wmma instances.
* Updated wrapper for the v3 instances
* removed the old wmma examples
* Renamed the v3 instances
* Deleted the gtest file added by mistake.
* Updated thge profiler with wrapper
* Fixed test errors.
* Fixed the review comments
* Fixed the if condition MACROS.
* REVERTED THE PROFILER CHANGES
* Revert "REVERTED THE PROFILER CHANGES"
This reverts commit 21cb98546c.
* Revert "Fixed test errors."
This reverts commit 13efcc6fe1.
* Revert "Updated thge profiler with wrapper"
This reverts commit 536f86661d.
* Added missing wrapper instances
* Updated copyrights.
* Fixed typo.
* Fixed copyrights.
* Updated copyrights.
* updated copyrights.
* comments on the atomics workaround
* fixed cmake comment
* Fix bug from merge
* clang-format-18
* Fix compilation error
* multi_abd wmma support:
- Add multiple A and B support to multiple D implementation (gridwise level)
- Add multi_abd GEMM (device level)
- Add instances (xdl parity)
- Add tests (both xdl and wmma)
- Add examples
- Add ckProfiler support (both xdl and wmma)
* Fix bug in device print function
* Fix unused template parameter
* Add support for fwd conv in gridwise implementation. Identical to run function for bwd data.
* Initial device implementation for grouped conv fwd multiABD wmma cshuffleV3. Functional but needs some fixups and extra features in the future.
* Make relevant profilers print the number of valid instances to aid testing.
* Add instances for all vanilla 2D and 3D flavors for f16 and bf16, only one instance per instance list to save compile time for now. Also added incomplete set of comp instances and bias_clamp for f16 2D, just to make sure the multiple-D aspects of the device implementation are working.
* Reset output buffer after each run in profile_grouped_conv_fwd_impl().
* Disable sharding for the new instances for now, has tendency to lead to linker errors on repeat builds.
* Add CTranspose optimization for NCHW cases just like in xdl cshuffle non-v3 device implementation.
* Add instances for all 8-bit 3D vanilla grouped conv fwd types, including mixed types but with the exception of deprecated f16 comp fp8. Adapt test so we can test 8-bit and mixed types.
* Add int8 instances for 2D vanilla grouped conv fwd all layouts.
* Implement merged groups in device impl and add instances for merged groups 3D vanilla conv fwd
* Add merged groups instances for all 2D vanilla grouped conv fwd types and layouts.
* Implement multi-AB support for grouped conv fwd and add example.
* Add 1D instances
* Add D layout tests to IsSupportedArgument()
* Add comp and mem instances for all vanilla 2D grouped conv fwd types. Skipping "x2" and "part2" instance lists, can be added later without special names if necessary.
* 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.
* Add some more tests for vanilla grouped conv fwd
* Add 2D bias clamp instances and tests
* Add 3D bias clamp instances and tests
* Add 2D and 3D clamp instances and tests
* Unify problem sizes across vanilla and clamp flavor tests
* Clean up device implementation: remove old todos, remove unnecessary comments and print statements, tweak description, wrap all prints in env check.
* Implement rotating memory and flush cache. Requires ad-hoc buffer size calculations.
* Remove wmma fp8 and bf8 instances when not targetting gfx12
* Add newer instances to DEVICE_INSTANCES so the main ckProfiler can build
* Remove old years for newly created files.
* No need to time kernels for now.
* Fixup comments
* Pass struct args to Gridwise Run() function by reference.
* Don't use workspace memory in the case where A needs explicit transposition but B does not.
* Move calculation of rotating memory buffer sizes to Argument member functions.
* After the convolution to gemm transformation, the resulting 2D tensor descriptors are not necessarily RowMajor or ColumnMajor, so things should not rely on this distinction. Therefore, pass all RowMajor to the Gridwise and use a special version of CheckValidity that does not rely on 2D tensor layouts.
* Unify xdl and wmma example code for grouped conv fwd scaleadd ab
* Go back to passing RCR 2D tensor layouts to gridwise gemm, and use CRC for the CTranspose case. Also remove the special convolution version of checkValidity(). It seems like no matter what 2D tensor layouts you pass to the gridwise gemm, and no matter if you are using extraMN, and no matter if you are using the convolution version of checkvalidity, the results of all tests are the same.
* 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 inside the test. Furthermore for XDL only the (NDHWGC, GKZYXC, NDHWGK) layout combination existed in the instance factory so that is the only one I added for wmma cshufflev3 and the gtest test as well. Another layout is tested in example 62, for xdl and wmma cshufflev3.
* 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.
* Small post-merge fixup, everything seems to work.
* Do not build or run Xdl operations with Wmma backend for now. Will be reverted before upstreaming.
* Extend scaleadd_ab instance lists
* Extend merged groups instance lists, including adaptations of xdl "2x" instances.
* Extend "comp" instance lists, including "2x" and "part2" instances. 2x instances disabled for now since they do not compile.
* Extend "mem" instance lists.
* Extend regular instance lists.
* Fixup comments and ignored kernel arg name
* Properly use the splitN offsets for D tensors in the gridwise Run() function. Was necessary to pass the bias_clamp_large_cases test.
* Make sure all strides in ComputePtrOffset are at least value initialized to avoid undefined strides. Not convinced this struct is properly initialized in other code / future code.
* Re-enable sharding for wmma cshufflev3 instances
* Post merge fix to vanilla test
* Optionally allow num_k_loop <= PrefetchStages in gridwise CheckValidity. Use this for grouped conv fwd but not in general.
* Remove spurious ck_tile changes that were presumably introduced somewhere in the repeated merging from develop.
* Post-merge fixes. Make sure the new gridwise gemm wmma v3 common Run function can be used. Remove splitK, and forceThreadTileTransfer for now. Also add CShuffle epilogue argument.
* Disable FP8 / BF8 testing on CDNA1/2, it doesn't work anymore and needs to be either fixed or removed.
* Re-enable old wmma instances
* Re-enable Linqun's Xdl Wmma instances
* Small post-merge fixes
* Fix copyright headers
* Remove commented code snippet in gridwise
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com>
* Limit the explicit cast added in threadwise_tensor_slice_transfer_v7r3 to only be used for f8, just in case it hurts performance.
* Adding tuned instace list for groupoed conv fwd (#3288)
Following flavors are updated with tuned instance list:
- grouped_conv2d_fwd
- grouped_conv2d_fwd_bias_clamp
- grouped_conv2d_fwd_clamp
- grouped_conv3d_fwd
- grouped_conv3d_fwd_bias_clamp
- grouped_conv3d_fwd_clamp
- grouped_conv3d_fwd_scaleadd_ab
Re-factored instance selection:
- removed all the unnecessary instance tuples (comp/mem/16x16/generic)
- removed all unnecessary layouts and data types
* Do not use std::remove_cvref_t, does not exist in C++17, use custom one.
* Splitting grouped conv fwd instances (#3449)
* Disable unnecessary and failing tests related to experimental CK builder
* Disable unnecessary ck builder experimental tests fully
* Added large tensor support for grouped conv fwd wmma
---------
Co-authored-by: Anca Hamuraru <anca@streamhpc.com>
Co-authored-by: apoorva <apoorva@streamhpc.com>
Co-authored-by: Anton Gorenko <anton@streamhpc.com>
Co-authored-by: Zoltan Lakatos <zoltan.lakatos@streamhpc.com>
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
Co-authored-by: Robin Voetter <robin@streamhpc.com>
Co-authored-by: Enrico Degregori <enrico@streamhpc.com>
Co-authored-by: Kiefer van Teutem <kiefer.van.teutem@streamhpc.com>
Co-authored-by: Kiefer van Teutem <50830967+krithalith@users.noreply.github.com>
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com>
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.com>
* Fixed typos for padded instances
* Added tests for fp16, KM_KN and KM_NK
* Padding not supported for when BDataType is pk_i4_t. Added fix for correct check and removed padding instances.
* Fixed typos
* Updated the set of tests for FP16
* Updated the set of tests for FP16
* Fix typo
* Moved f16xi4 test under the correct data layout group
* example for gemm_universal_bf16
* Adding examples for gemm_wmma instances
* Added the missing parameters
* Fixed review comments and added executable to cmakeLists
* Fixing clang format
* Fixing build erros
* Fixed compilation failure.
* Modified some code as per gemm_universal_examples
* Fixed the gemm specialization error
* Fixed the build errors.
* Fix strides of a/b_thread_desc
The descriptors are larger than needed (even though the compiler don't alloc registers for unused values).
* Load in M/NRepeat dims with thread copy's slice instead of a loop
* Clone BlockwiseGemmXdlops_pipeline_v1 for WMMA implementation
* Implement Intrawave and Interwave variants of pipeline v1
* Add instances for Interwave and Intrawave v1
* Add instances with ABlockLdsExtraM and BBlockLdsExtraN = 0
* Remove instances that are too slow (mostly because of register spilling)
* Add a workaround for fp8/bf8->f32 packed conversion issue
* Add instances for Interwave and Intrawave v1
* Enable profiling of mixed precision with f8 and int4 on WMMA
* Fix segfault in profiler when B is pk_i4_t
b_device_buf's size in bytes is larger than b_k_n_permute so b_device_buf.ToDevice reads out-of-bounds.
* Remove instances that are too slow (mostly because of register spilling)
* Add missing add_device_gemm_wmma_universal_f8_f8_bf16 declarations
* Add test case for bf16_i4
* Add missing Regular tests
* Add test_gemm_universal_xdl/wmma_fp16 to REGRESSION_TESTS
They take more than 30 seconds
* Fix a bug that fp16_i4 validation passes only with PermuteB
A permutation required by conversion from pk_i4_t to half_t does not
depend on PermuteB, they can be used independently.
* Use PermuteB with f16_i4 in most instances (as xdl)
Some instances use PermuteB = false for checking correctness.
See also the previous commit.
* Fix cache flushing for pk_i4
* Add mixed precision examples
* Disable all tests and instances with f8 on gfx11
Even though f8_f16 and f16_f8 don't require f8 WMMA instructions,
gfx11 still lacks hardware instructions for fast f8->f32 conversion.
* Add FP16 KM_NK and KM_KN test suites for XDL
These tests were added to common .inc for better testing of WMMA instances
* Support multiple D in GridwiseGemm_wmma_cshuffle_v3
DeviceGemm_Wmma_CShuffleV3 is changed for new template parameters.
* Use ThreadGroupTensorSliceTransfer_v7r3
* Clone for device_gemm_wmma_cshuffle_v3.hpp for future Multiple D support
* Clone example/65_gemm_multiply_multiply/gemm_add_add_xdl_fp16.cpp for wmma
* Implement DeviceGemmMultipleD_Wmma_CShuffleV3
* Make gemm_add_add_wmma to work with DeviceGemmMultipleD_Wmma_CShuffleV3
* Prepare gemma_add tests for adding wmma
* Add gemm_add_fastgelu instances and test
* Add a special wrapper to use DeviceGemmMultipleD_Wmma_CShuffleV3 with old API
ckProfiler uses DeviceGemmMultipleD (tests also call its functions), the wrapper allows to use
DeviceGemmMultipleDSplitK instances there.
* removed unnecessary ck parts from compilation
* initial gemm_add_multiply instance implementations
* fixed profiler help message for gemm_add_multiply
* improved multiply_add profiler layout help
* fixed template arguments for test instances
* added test for gemm_add_multiply
* Support multiple D in GridwiseGemm_wmma_cshuffle_v3
DeviceGemm_Wmma_CShuffleV3 is changed for new template parameters.
* Use ThreadGroupTensorSliceTransfer_v7r3
* Clone for device_gemm_wmma_cshuffle_v3.hpp for future Multiple D support
* Clone example/65_gemm_multiply_multiply/gemm_add_add_xdl_fp16.cpp for wmma
* Implement DeviceGemmMultipleD_Wmma_CShuffleV3
* Make gemm_add_add_wmma to work with DeviceGemmMultipleD_Wmma_CShuffleV3
* Prepare gemma_add tests for adding wmma
* Add gemm_add_fastgelu instances and test
* Add a special wrapper to use DeviceGemmMultipleD_Wmma_CShuffleV3 with old API
ckProfiler uses DeviceGemmMultipleD (tests also call its functions), the wrapper allows to use
DeviceGemmMultipleDSplitK instances there.
* switched to splitK interface
* log print added to splitk benchmarks
* revert main cmake comments
* newline change reverted
* added add_fastgelu instances
* revert unintended change in xdl add_fastgelu
* created gemm_add_add_fastgelu instances
* created fastegelu instances
* added tests for all splitk fastgelus
* Added tests.
* multiply_add instances created
* updates to add_multiply splitk instances
* splitk xdl test fixes
* added wmma multiply_multiply instances
* fixed ONLY_XDL_AND_WMMA_KERNELS tag
* Added gemm_add examples for wmma v1 and v3
* fixed / workarounded i8 instances
* Modified the v3 code to added one fp16 bxdl instance.
* added bf16 xdl instance.
* adding gemm_add wmma_cshuffle and other support
(cherry picked from commit ec447e7f564095ea969eddc39ec77b843aa52976)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* add instances into camkelists
(cherry picked from commit 23bf2d2771c939ea3ca7f493433c55255bffd08e)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* This is work in progress, edited the template parameters in order to build
(cherry picked from commit b4fde8a3314cb44659c4bbda35f1a0133c63dc41)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* temp work saved, changed the BDataType to f16 or bf16 since wmma currently not support non-equal A and B datatype
(cherry picked from commit 22fbd68f1db458ab50780a394ee2544c7a1484d1)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* added datatype and use clang-format-12
(cherry picked from commit ae4e853682ef1bb27784b2f965b4a66b3751ceec)
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
* Fixing build errors
* Added instances for v3
* Adding instances and executables
* Code update of template parameters modified.
* Renamed file.
* Added tests.
* resolved error tests.
* Fixing build errors
* Updated comments
* removed the changes as per the MR review comment.
* Updated tests.
* fp8 instances - not tested
* Restored the Cmake file that was reverted by mistake during rebase.
* fixed wmma_op test
* Updated comments.
* Updated the template parameter description
* fixed rdna4 instances
* fixed back compatibility on gfx11
* cleanups
* fix ckProfiler
* one more cmake fix
* added fp8 instances
* Updated tests to ad BF16 instances as per review comment
* Added include file and cleaned up(as per review comment)
* Updated and optimized the example code for all types.
* Fixed clang format
* Resolve "Implement `device_gemm_bilinear` for RDNA4"
* test generalization to handle FP16 shuffle better
* added missing changes
* Added bf16 wmma instance for add_relu
* Added f16 wmma instance and corrected bf16 instance errors.
* Added instances to Cmake
* Modified the template parameters to make the instances work.
* Fixed typo in profiler
* Added v3 instances for gemm_add_relu
* addressed core review comments
* Added test for gemm_add_relu wmma instance
* Cleaned up the code.
* Added examples for gemm_add_relu
* Fixing typo to resolve build errors.
* Fixes applied to fix the precision loss.
* fix billinear test after merge
* Removed the old wmma instances.
* Added wrapper and renamed the wmma_v3 instances
* Updated copyrights and added wrappers.
* Fixes applied according to review comments
* Apply 1 suggestion(s) to 1 file(s)
Co-authored-by: Robin Voetter <robin@streamhpc.com>
* Removed the old wmma instances.
* Updated wrapper for the v3 instances
* removed the old wmma examples
* Renamed the v3 instances
* Deleted the gtest file added by mistake.
* Updated thge profiler with wrapper
* Fixed test errors.
* Fixed the review comments
* Fixed the if condition MACROS.
* REVERTED THE PROFILER CHANGES
* Revert "REVERTED THE PROFILER CHANGES"
This reverts commit 21cb98546c.
* Revert "Fixed test errors."
This reverts commit 13efcc6fe1.
* Revert "Updated thge profiler with wrapper"
This reverts commit 536f86661d.
* Added missing wrapper instances
* Updated copyrights.
* Fixed typo.
* Fixed copyrights.
* Updated copyrights.
* updated copyrights.
* comments on the atomics workaround
* fixed cmake comment
* Fix bug from merge
* clang-format-18
* Fix compilation error
* multi_abd wmma support:
- Add multiple A and B support to multiple D implementation (gridwise level)
- Add multi_abd GEMM (device level)
- Add instances (xdl parity)
- Add tests (both xdl and wmma)
- Add examples
- Add ckProfiler support (both xdl and wmma)
* Fix bug in device print function
* Fix unused template parameter
* Add support for fwd conv in gridwise implementation. Identical to run function for bwd data.
* Initial device implementation for grouped conv fwd multiABD wmma cshuffleV3. Functional but needs some fixups and extra features in the future.
* Make relevant profilers print the number of valid instances to aid testing.
* Add instances for all vanilla 2D and 3D flavors for f16 and bf16, only one instance per instance list to save compile time for now. Also added incomplete set of comp instances and bias_clamp for f16 2D, just to make sure the multiple-D aspects of the device implementation are working.
* Reset output buffer after each run in profile_grouped_conv_fwd_impl().
* Disable sharding for the new instances for now, has tendency to lead to linker errors on repeat builds.
* Add CTranspose optimization for NCHW cases just like in xdl cshuffle non-v3 device implementation.
* Add instances for all 8-bit 3D vanilla grouped conv fwd types, including mixed types but with the exception of deprecated f16 comp fp8. Adapt test so we can test 8-bit and mixed types.
* Add int8 instances for 2D vanilla grouped conv fwd all layouts.
* Implement merged groups in device impl and add instances for merged groups 3D vanilla conv fwd
* Add merged groups instances for all 2D vanilla grouped conv fwd types and layouts.
* Implement multi-AB support for grouped conv fwd and add example.
* Add 1D instances
* Add D layout tests to IsSupportedArgument()
* Add comp and mem instances for all vanilla 2D grouped conv fwd types. Skipping "x2" and "part2" instance lists, can be added later without special names if necessary.
* 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.
* Add some more tests for vanilla grouped conv fwd
* Add 2D bias clamp instances and tests
* Add 3D bias clamp instances and tests
* Add 2D and 3D clamp instances and tests
* Unify problem sizes across vanilla and clamp flavor tests
* Clean up device implementation: remove old todos, remove unnecessary comments and print statements, tweak description, wrap all prints in env check.
* Implement rotating memory and flush cache. Requires ad-hoc buffer size calculations.
* Remove wmma fp8 and bf8 instances when not targetting gfx12
* Add newer instances to DEVICE_INSTANCES so the main ckProfiler can build
* Remove old years for newly created files.
* No need to time kernels for now.
* Fixup comments
* Pass struct args to Gridwise Run() function by reference.
* Don't use workspace memory in the case where A needs explicit transposition but B does not.
* Move calculation of rotating memory buffer sizes to Argument member functions.
* After the convolution to gemm transformation, the resulting 2D tensor descriptors are not necessarily RowMajor or ColumnMajor, so things should not rely on this distinction. Therefore, pass all RowMajor to the Gridwise and use a special version of CheckValidity that does not rely on 2D tensor layouts.
* Unify xdl and wmma example code for grouped conv fwd scaleadd ab
* Go back to passing RCR 2D tensor layouts to gridwise gemm, and use CRC for the CTranspose case. Also remove the special convolution version of checkValidity(). It seems like no matter what 2D tensor layouts you pass to the gridwise gemm, and no matter if you are using extraMN, and no matter if you are using the convolution version of checkvalidity, the results of all tests are the same.
* 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 inside the test. Furthermore for XDL only the (NDHWGC, GKZYXC, NDHWGK) layout combination existed in the instance factory so that is the only one I added for wmma cshufflev3 and the gtest test as well. Another layout is tested in example 62, for xdl and wmma cshufflev3.
* 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.
* Small post-merge fixup, everything seems to work.
* Do not build or run Xdl operations with Wmma backend for now. Will be reverted before upstreaming.
* Extend scaleadd_ab instance lists
* Extend merged groups instance lists, including adaptations of xdl "2x" instances.
* Extend "comp" instance lists, including "2x" and "part2" instances. 2x instances disabled for now since they do not compile.
* Extend "mem" instance lists.
* Extend regular instance lists.
* Fixup comments and ignored kernel arg name
* Properly use the splitN offsets for D tensors in the gridwise Run() function. Was necessary to pass the bias_clamp_large_cases test.
* Make sure all strides in ComputePtrOffset are at least value initialized to avoid undefined strides. Not convinced this struct is properly initialized in other code / future code.
* Re-enable sharding for wmma cshufflev3 instances
* Post merge fix to vanilla test
* Optionally allow num_k_loop <= PrefetchStages in gridwise CheckValidity. Use this for grouped conv fwd but not in general.
* Remove spurious ck_tile changes that were presumably introduced somewhere in the repeated merging from develop.
* Post-merge fixes. Make sure the new gridwise gemm wmma v3 common Run function can be used. Remove splitK, and forceThreadTileTransfer for now. Also add CShuffle epilogue argument.
* Disable FP8 / BF8 testing on CDNA1/2, it doesn't work anymore and needs to be either fixed or removed.
* Re-enable old wmma instances
* Re-enable Linqun's Xdl Wmma instances
* Small post-merge fixes
* Fix copyright headers
* Remove commented code snippet in gridwise
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com>
* Limit the explicit cast added in threadwise_tensor_slice_transfer_v7r3 to only be used for f8, just in case it hurts performance.
* Adding tuned instace list for groupoed conv fwd (#3288)
Following flavors are updated with tuned instance list:
- grouped_conv2d_fwd
- grouped_conv2d_fwd_bias_clamp
- grouped_conv2d_fwd_clamp
- grouped_conv3d_fwd
- grouped_conv3d_fwd_bias_clamp
- grouped_conv3d_fwd_clamp
- grouped_conv3d_fwd_scaleadd_ab
Re-factored instance selection:
- removed all the unnecessary instance tuples (comp/mem/16x16/generic)
- removed all unnecessary layouts and data types
* Do not use std::remove_cvref_t, does not exist in C++17, use custom one.
* Splitting grouped conv fwd instances (#3449)
* Disable unnecessary and failing tests related to experimental CK builder
* Disable unnecessary ck builder experimental tests fully
---------
Co-authored-by: Anca Hamuraru <anca@streamhpc.com>
Co-authored-by: apoorva <apoorva@streamhpc.com>
Co-authored-by: Anton Gorenko <anton@streamhpc.com>
Co-authored-by: Zoltan Lakatos <zoltan.lakatos@streamhpc.com>
Co-authored-by: Cenxuan <cenxuan@streamhpc.com>
Co-authored-by: Robin Voetter <robin@streamhpc.com>
Co-authored-by: Enrico Degregori <enrico@streamhpc.com>
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com>
Co-authored-by: Wojciech Laskowski <77888887+wj-laskowski@users.noreply.github.com>
Refactor and integrate CK GPU references into ckProfiler.
- All convolution layouts and groupings supported for all three directions
- Unit tests verifying GPU and CPU reference is the same
- Support added to profiler (do_verification = 2 enables GPU reference)
- One profiler-based test per direction changed to GPU reference to demonstrate usag
Closes AICK-427
* Parallelization in dataset generation
* Parallelizable tests for fwd, bwd data, bwd weight with datasets
* .gitignore generated datasets
* Test parallelization script with round-robin GPU scheduling
* Parallelization updates to test generation and running
* Dataset paths relative to executable
* Update output from test generation
* Default to one GPU in test generation
* Add small dataset tests to Jenkins
* Update copyright lines
* Update test_data/generate_test_dataset.sh
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* Move trap disable
* Common get path function
---------
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
* [CK] Add command option instance_index and param_mask to run partial ck test
Many CK test are instance test. it will loop all instance in the instance library. It causes test often out-of-time if we run test on simulator/emulator.
This PR add option instance_index and param_mask to reduce the workload of instance test
instance_index: only run test 1 available instance with specified index.
param_mask: filter the embedded parameter with specified mask
* fix CI error
* fix clang format
---------
Co-authored-by: illsilin_amdeng <Illia.Silin@amd.com>
* Add CSV-driven convolution test pipeline
- Add test_grouped_convnd_fwd_dataset_xdl.cpp with CSV reader functionality
- Add complete dataset generation toolchain in test_data/
- Add Jenkins integration with RUN_CONV_COMPREHENSIVE_DATASET parameter
- Ready for comprehensive convolution testing with scalable datasets
* Update convolution test dataset generation pipeline
* add 2d, 3d dataset csv files
* Remove CSV test dataset files from repository
* Update generate_test_dataset.sh
* Fix channel division for MIOpen to CK conversion
* Remove unnecessary test files
* Fix clang-format-18 formatting issues
* TEST: Enable comprehensive dataset tests by default
* Fix test_data path in Jenkins - build runs from build directory
* Add Python dependencies and debug output for CSV generation
* Remove Python package installation - not needed
* Add better debugging for generate_test_dataset.sh execution
* Fix Jenkinsfile syntax error - escape dollar signs
* Add PyTorch to Docker image for convolution test dataset generation
- Install PyTorch CPU version for lightweight model execution
- Fixes Jenkins CI failures where CSV files were empty due to missing PyTorch
- Model generation scripts require PyTorch to extract convolution parameters
* Add debugging to understand Jenkins directory structure and CSV file status
- Print current working directory
- List CSV files in test_data directory
- Show line counts of CSV files
- Will help diagnose why tests fail in Jenkins
* Fix clang-format-18 formatting issues
- Applied clang-format-18 to test file
- Fixed brace placement and whitespace issues
* Add detailed debugging for CSV dataset investigation
- Check generated_datasets directory contents
- List all CSV files with line counts
- Show first 5 lines of main CSV file
- Applied clang-format-18 formatting
- This will help identify why CSV files are empty in Jenkins
* keep testing add pytorch installation in shell script
* Use virtual environment for PyTorch installation
- Jenkins user doesn't have permission to write to /.local
- Create virtual environment in current directory (./pytorch_venv)
- Install PyTorch in virtual environment to avoid permission issues
- Use PYTHON_CMD variable to run all Python scripts with correct interpreter
- Virtual environment will be reused if it already exists
* Remove debug code and reduce verbose logging in Jenkins
- Remove bash -x and debug commands from Jenkinsfile execute_args
- Remove all debug system() calls and getcwd from C++ test file
- Remove unistd.h include that was only needed for getcwd
- Remove debug print in CSV parser
- Add set +x to generate_test_dataset.sh to disable command echo
- Redirect Python script stdout to /dev/null for cleaner output
This makes Jenkins logs much cleaner while still showing progress messages.
* install gpu torch
* Clean up and optimize comprehensive dataset test pipeline
- Reorder Jenkinsfile execution: build -> generate data -> run test
- Remove commented-out debug code from generate_test_dataset.sh
- Ensure all files end with proper newline character (POSIX compliance)
- Keep useful status messages while removing development debug prints
- Set MAX_ITERATIONS=0 for unlimited test generation in production
* Add configuration modes to reduce test execution time
- Add --mode option (half/full) to generate_model_configs.py
- half mode (default): ~278 configs (224 2D + 54 3D) -> ~1,058 total tests
- full mode: ~807 configs (672 2D + 135 3D) -> ~3,093 total tests
- Update generate_test_dataset.sh to use CONFIG_MODE environment variable
- Keeps all model types but reduces parameter combinations intelligently
- Fixes Jenkins timeout issue (was running 3,669 tests taking 17+ hours)
- Default half mode should complete in ~4-5 hours instead of 17+ hours
* Add small mode for quick testing of comprehensive dataset
* jenkins pipeline test done
* jenkins test done
* Trigger CI build
* remove test comment and update data generation option as half
---------
Co-authored-by: Bartłomiej Kocot <barkocot@amd.com>
* Support 64 bit indexing
* Add new grouped conv fwd kernel for large tensors
* Add instances large tensor
* Fixes for transform conv to gemm
* Fixes
* fixes
* Remove not needed instances
* examples fixes
* Remove not need ds arrays
* Fix tests
* Add 2GB check in gridwise dl
* Fixes
* parse examples inside the add_example_executable function
* fix the example 64 cmake file
* add xdl flag to the gemm_bias_softmax_gemm_permute example
* add filtering of tests based on architecture type
* enable test_grouped_gemm for gfx9 only
* enable test_transpose only for gfx9
* only linnk test_transpose if it gets built
* split the gemm instances by architectures
* split gemm_bilinear,grouped_conv_bwd_weight instances by targets
* split instances by architecture
* split grouped_conv instances by architecture
* fix clang format
* fix the if-else logic in group_conv headers
* small fix for grouped convolution instances
* fix the grouped conv bwd weight dl instances
* fix client examples
* only enable client examples 3 and 4 on gfx9
* set the gfx9 macro
* make sure the architecture macros are set by cmake
* use separate set of xdl/wmma flags for host code
* sinmplify the main cmake file
* add conv_fwd_bf8 instance declaration
* Introduce multiABD api and deprecate multiD
* Replace multiD with multiABD
* Mark structures as deprecated
* Change doxygen deprecated to note to avoid warnings
* Support multi AB for grouped conv fwd xdl
* Add instances
* Add client example
* Add example
* Add interface test
* Minor fixes
Minor fixes
Minor fixes
* Comment fixes
* Fixes
* Reference fix
* Test xdl fixes
* Improve multi_ab interface test
* Enable grouped conv with small K or C
* Add missing instances
* Refactor grouped conv fwd instances
* Fix fp16 instances since it supports src_per_vec %2 = 0
* Add generic instances
