* Use vectorized stores for dropout randvals
With no kPadSeqLenK the kernel uses 2 buffer_store_dwordx2 instead of
16 buffer_store_byte. This requires less registers and reduces spilling.
* Calculate dropout randvals for storing and applying only once
Even though it may add a small overhead when storing is not required,
it uses significantly less registers and hence no spilling.
Fixes LDS bank conflicts on gfx950 for universal gemm v3 pipeline
Replaces hardcoded LDS layer calculations with dynamic computation using the new architecture helpers
Adds architecture-specific helper function get_n_lds_banks()
Changes function attributes from CK_TILE_HOST_DEVICE to CK_TILE_DEVICE in universal gemm policy
BlockFmhaPipelineQRKSVS reuses LDS for K and dropout so there must be
block_sync_lds between loading k_lds_window by gemm_0 and storing
dropout randval.
* Pass hdim to tile_example_fmha_fwd in fp8 tests
* Add WMMA support to fwd FMHA pipelines
* Tune tile sizes a bit for less spilling
fp16 256 is still quite slow
* Fix Q grad tile distribution for warp size = 32 and hdim >= 256
With AccDataType = float and warp size = 32, K0 becomes 0, K repeat is required to correcty distribute the tile.
* Use code based on BlockDropout in BlockDropoutBwd
* Fix split KV combine kernel for gfx12 (warp size 32) and make it more universal
* Fix LSE LDS tensor descriptors: kMaxSplits and kM0 were swapped, it worked on gfx9
because they both equal to 8 while on gfx12 they are 8 and 4;
* Fix Oacc LDS tensor descriptor: it was transposed even though its shape=[4 * kM0, kN1],
it worked on gfx9 because 4 * kM == kN1 == 32;
* Removing these hidden dependecies allows to support:
* any number of warps (power-of-2), not only 4;
* kN1 = 16, not only 32;
* any number of splits;
* Rename ids like o_acc_4 and Oacc4 to eliminate confusion: kNumWarps doesn't have to be 4 now
* Replace hard-coded kN1 in dispatch code with the requested tile size
* Add gfx12-specific tile sizes for split KV
* Pass GPU architecture to kernel generation scripts
This is still a temporary solution.
* Build and run FMHA CI tests for gfx12
* Fix issue after merging
* Fix bwd tile sizes
The current pipelines always read only one tile K and V tile, this
requires bk0 == bhdq and bk2 == bhdv (kK0 == kQKHeaddim and
kK2 == kVHeaddim).
* Use hardware f32->f8 on gfx12, remove v_perm
__builtin_amdgcn_perm is not needed because
__builtin_amdgcn_cvt_pk_fp8_f32 allows to specify which word (16 bit of
32-bit dword) is used to store results (two f8 values).
* Update changelog
* Add WMMA support to pagedkv
* Fix scripts after rebasing
* Support 16x16 (MFMA, WMMA) and 32x32 (MFMA) tiles in fwd and bwd BlockDropout
Add comments with dropout implementation details
Fix performance regression of fwd+dropout
* Remove some usage of type punning (reinterpret_cast with ref or ptr) in Philox;
* "scalarize" seed and offset, they may come either from kernel args or from device memory
(presumably loaded with vector loads).
These changes help the compiler to procude more optimal code and reduce register spilling.
Use WarpGemmDispatcher instead of explicit WarpGemmMfma... to get CWarpDstrEncoding
Use code based on BlockDropout in BlockDropoutBwd
Refactor BlockDropout (fwd)
Implement BlockDropout (fwd) for WMMA
Originally BlockDropout only supported 32x32 tiles (IsWG32 = true),
this version supports 16x16 tiles.
If MPerBlock > MWarp * 16, it can generate numbers for two 16x16 tiles, similarly
to BlockDropoutBwd.
Implement BlockDropoutBwd for WMMA
Remove MakeRandValLds* functions unused in BlockDropoutBwd
Remove unused Run overload from BlockDropoutBwd
* Fix regression with philox seed and offset when they exceed 32-bit int
__builtin_amdgcn_readfirstlane works with 32-bit values, seed and offset
are 64-bit so they get truncated.
* Fix names after cherry-picking
* Fix selection of a fallback tile based on bm0
The assumption that the largest bm0 == 128 is not always true for
current fp32 tiles.
* Do not use filters related to qr_async_trload
They disable tiles/pipelines which are valid for gfx12.
* Use different dstr encoding when C is transposed
* Do not call GetQKBlockGemm (and hence WarpGemmDispatcher) in host code
Some WarpGemmDispatcher instantiations are defined only
for specific archs and undefined on host.
Calculations related to sched barriers are moved from Pipeline's public
fields into pipeline's operator().
* Fix incorrect name WarpGemmMfmaFp8Fp8F32M32N32K16SwizzleBTransposedCDistribution
Correct name is WarpGemmMfmaFp8Fp8F32M32N32K32SwizzleBTransposedCDistribution
because it's 32x32x16 with IterateK = 2 so K = 32, also all tiles used
in codegen scripts are 32, 32, 32.
* Generalize usages of WarpGemmDispatcher for MFMA and WMMA
WarpGemmMfmaFp8Fp8F32M32N32K32SwizzleBTransposedCDistribution is still
used explicitly becaus of swizzle factor = 4.
* Mark has_load_tr as maybe_unused
There are no transpose loading for RDNA.
* Remove CK_TILE_USE_MFMA/WMMA from fmha-related code
* Detect BlockSize on host based on warp size of the current device
If kBlockSize == kNumWarps * get_warp_size(), the kernel is launched with
kBlockSize / 2 because on host get_warp_size() == 64 always.
* Fix calculation of grid size for combine kernel with warp size = 32
* Add missing includes and header
* Support multiple archs in one binary for fwd
* Support multiple archs in one binary for fwd_splitkv, fwd_appendkv, pagedkv_prefill
* Support multiple archs in one binary for bwd
* trload kernels are compiled only for gfx950;
* instances with padding are checked after instances without padding so
they can be used as fallbacks (similarly to fwd);
* Extract common code from register_traits
* Revert "Fix regression with philox seed and offset when they exceed 32-bit int"
To simplify merging , the proper fix is in develop already.
* Support new numerical d paddings in trait ordering checks
* Build fp32 tests only on gfx9
* Do not use hardcoded M0 = 64 for dot bwd kernel
* Use textwrap.indent from standard library
* Make fp8 pipelines on gfx12 consistent with gfx9
* Update tests for current pipelines
* Make ninja check more responsive in CI
ninja buffers output so this job looks hanging.
* Support fp8fp32 by limiting O vector size
The fp32 output type requires storing 8 * sizeof(float) = 32 bytes,
which is not implemented (here 8 is the number of C values per lane for
v_wmma_f32_16x16x16...).
* Remove unused cmake options
* Unify including amd_buffer_addressing.hpp/_builtins.hpp
* Temporarily use amd_buffer_addressing.hpp on >=gfx10
amd_buffer_addressing_builtins.hpp uses inline asm for loads/stores
which is not compatible with >=gfx10:
* 1 scalar for exec masks instead of 2,
* gfx12 uses different instruction names etc.
* Update asm in bf16 conversions to work with warp 32
* Do not generate splitkv/appendkv with vlayout=col for consistency with fwd
* Add arch tags to kernels/host funcs, compile for each arch separately
* Add kM0 to fmha_bwd_dot_do_o kernel name to match filename
* Add workaround for miscompilation of bwd with padded hdim
SWDEV-559729: v_wmma instructions can be incorrectly placed in divergent
branches used to store padded tensors (when some lanes are inactive due
to padding). Inline asm with dummy dependencies on VGPRs of the tensors
prevents the compiler doing this.
* Fix add_gtest_executable for absolute paths
Some tests (like gemm_tile_engine) pass absolute paths to source files.
In CI the branch name is a part of the root dir, and if the branch name
contains "wmma", "xdl" etc., files can be incorrectly excluded.
* Run only hdim 128 smoke tests for fp8fp32
There are no instances for hdim 64 and 256.
* Format py with ruff to simplify merging develop
* Fix incorrect var name
* Codegen for gfx9,gfx950 when --targets is not specified
Aiter and Pytorch require changes for passing their targets to the codegen scripts.
With this temporary solution the files are generated but not all of them
have to be really built (depending on the used --offload-arch=).
* Combine arch-related values into ArchTrait
This more centralized approach removes duplication of various formatting templates.
* Try a workaround for Jenkins error "groovyjarjarasm.asm.MethodTooLargeException: Method too large"
Some code is extracted into a function.
* [CK_TILE] fmha: Add query padding support to backward pass
Introduces support for query sequence padding (q_padding) in the FMHA backward pass kernels.
- Passing `seqlen_q_ptr` to the backward kernels to distinguish logical from physical sequence lengths.
- Updating `OGradDotO`, `ConvertQGrad`, and `DQDKDV` kernels to respect logical lengths and handle zero-length sequences.
- Aligning LSE indexing in the forward kernel with the padded layout for consistency.
- Adding a new GTest suite (`test_fmha_bwd_kernel_padding.cpp`) with comprehensive tests for various padding scenarios, including zero-length
sequences and deterministic mode.
* fix clang format
* Adapt fmha_bwd_runner.cpp to new q, kv sequence padding
Add backward q/kv sequence padding unit tests.
* [CK_TILE] fmha: Unify sequence length and padding handling
Refactor the handling of sequence lengths and padding in the
FMHA forward and backward kernels to provide a more unified and flexible
interface.
- Replaced `seqstart_padded_*_ptr` with a more robust system that uses
`seqstart_*_ptr` for physical sequence lengths and introduces
`seqlen_*_ptr` and `cu_seqlen_*_ptr` for logical (unpadded) lengths.
- Established a clear order of precedence for determining sequence
length: cumulative lengths (`cu_seqlen_*_ptr`) take priority,
followed by per-sequence lengths (`seqlen_*_ptr`), and finally
physical lengths derived from `seqstart_*_ptr`.
- Clarified the distinction between "group mode" and "batch mode" and
how sequence lengths are handled in each case.
- Renamed `cu_seqlen_kv_ptr` to `cu_seqlen_k_ptr` for consistency.
- Updated comments and documentation to reflect the new argument
structure and usage.
---------
Co-authored-by: illsilin_amdeng <Illia.Silin@amd.com>
* Support 16x16 (MFMA, WMMA) and 32x32 (MFMA) tiles in fwd and bwd BlockDropout
Add comments with dropout implementation details
Fix performance regression of fwd+dropout
* Remove some usage of type punning (reinterpret_cast with ref or ptr) in Philox;
* "scalarize" seed and offset, they may come either from kernel args or from device memory
(presumably loaded with vector loads).
These changes help the compiler to procude more optimal code and reduce register spilling.
Use WarpGemmDispatcher instead of explicit WarpGemmMfma... to get CWarpDstrEncoding
Use code based on BlockDropout in BlockDropoutBwd
Refactor BlockDropout (fwd)
Implement BlockDropout (fwd) for WMMA
Originally BlockDropout only supported 32x32 tiles (IsWG32 = true),
this version supports 16x16 tiles.
If MPerBlock > MWarp * 16, it can generate numbers for two 16x16 tiles, similarly
to BlockDropoutBwd.
Implement BlockDropoutBwd for WMMA
Remove MakeRandValLds* functions unused in BlockDropoutBwd
Remove unused Run overload from BlockDropoutBwd
* Fix regression with philox seed and offset when they exceed 32-bit int
__builtin_amdgcn_readfirstlane works with 32-bit values, seed and offset
are 64-bit so they get truncated.
* Add F32 MFMA warp gemms
* Support f32 in fwd FMHA
* Implement transpose_vectors for 4-byte types (float)
* Fix unexpected implicit f32->uint32 cast in buffer_store<4>
__builtin_amdgcn_raw_buffer_store_b32 expects unsigned int but float was passed (implicitly casted to uint).
mbuf_t types in other buffer_store<> are changed for consistency.
* Support F32 in bwd FMHA
hdim = 256 is disabled for now because it uses too much memory on gfx90a
* Support Headdim = 48 (divisible by 16) in fwd
* Add fp32-specific receipts (800 and 801)
* Tune fwd tiles
* Tune bwd tiles
* Use small tiles only for small seqlen_q
* Fix after rebasing
* Fix selection of a fallback tile based on bm0
The assumption that the largest bm0 == 128 is not always true for
current fp32 tiles.
* Remove constraints and adjust filtering for fp32
Custom constraints are no longer needed because now the smallest tile
is selected automtically based on seqlen_q.
Filters related to qr_async_trload disabled valid fp32 tiles.
* Add fp32 tests
* Make splitkv and appendkv compile for fp32 only
There are no instances yet, but API still must compile when only fp32 is
requested.
* Remove unimportant f32 instances
* Add test_ck_tile_fmha_*_fp32 to REGRESSION_TESTS
* Replace magic numbers with a constant, improve comments for dropout
* Update changelog
* Fix condition that dq_acc must be set to zero when mask is used
The change was introduced in #2799
* Replace warp_uniform with recently added amd_wave_read_first_lane
* Add hdim = 96 and 192 to fwd
* Fix validation of rotary embedding with time_kernel_
When rotary embedding is used, the appendkv kernel modifies the q tensor
(multiple times when time_kernel_ is set). We need to reset the q buffer
and rerun all kernels.
* Fix synchronization issue in splitkv combine pipeline
Different warps can read and then rewrite the same values of lse_acc_lds.
Sometimes warps progress at different speeds, one warp can rewrite
values that are still being read by another warp.
Running the tests multiple times and, preferably, with multiple
processes on the same GPU helps to trigger this issue:
bin/test_ck_tile_fmha_fwd_fp16 --gtest_repeat=-1 --gtest_shuffle --gtest_throw_on_failure --gtest_filter="TestCkTileFmhaFwd/*KV*"
* * [CK_TILE] Add sequence padding and variable length support in fmha (and v3)
- Group Mode Padding: Introduces the `-s_qpad` argument to support
physically padded layouts. Kernels now use padded start pointers
(`seqstart_padded_*_ptr`) for memory addressing.
- Batch Mode Variable Length: Adds `-q_eff_lens` and `-kv_eff_lens`
arguments for efficient processing of variable-length sequences by
passing cumulative effective lengths (`cu_seqlen_*_ptr`) to the kernel.
- FMHA examples: Support padding and variable length both in
group and batch mode. Dispatcher is updated as well (dispatch to
kPadSeqLenK enabled pipeline).
- New padding test cases: Add padding test cases to `smoke_test_fwd.sh` and
`test_fmha_fwd.inc`, and add benchmarks to `benchmark_fwd.sh` and
`benchmark_fwd_v3.sh` as well. These test cases and benchmarks that
specifically validate/benchmark the new padding and variable-length
functionalities in both group and batch modes.
* [CK_TILE] Fix build error in fmha unit tests
* [CK_TILE] add mqa, gqa to sequence padding unit tests
* [CI_TILE] Reduce the number of padding seqlen unit tests in FMHA to avoid timeouts in CI
* [CK_TILE] remove unnecessary MageKArgs overload in FmhaFwdV3Kernel and FmhaFwdKernel
* disable cast_tile_pk_fp16_fp32 on gfx950
* fix wrong encoding when hdim is not exponentiation of 2
---------
Co-authored-by: asleepzzz <hanwen.chang@amd.com>
* Have a workable version for SGPR
* have a workable version for atomic add
* Revert "have a workable version for atomic add"
This reverts commit 792377a590c26cfff9c8f545d9a9e8484a7422eb.
* substitute with the new sgpr read api
* update the CHANGELOG
* have a workable version for atomic add
* Revert "have a workable version for atomic add"
This reverts commit 792377a590c26cfff9c8f545d9a9e8484a7422eb.
* change to static for logic
* have a workable version for atomic add
* Revert "have a workable version for atomic add"
This reverts commit 792377a590c26cfff9c8f545d9a9e8484a7422eb.
* [CK_TILE] Add sequence padding and variable length support in fmha (and v3)
- Group Mode Padding: Introduces the `-s_qpad` argument to support
physically padded layouts. Kernels now use padded start pointers
(`seqstart_padded_*_ptr`) for memory addressing.
- Batch Mode Variable Length: Adds `-q_eff_lens` and `-kv_eff_lens`
arguments for efficient processing of variable-length sequences by
passing cumulative effective lengths (`cu_seqlen_*_ptr`) to the kernel.
- FMHA examples: Support padding and variable length both in
group and batch mode. Dispatcher is updated as well (dispatch to
kPadSeqLenK enabled pipeline).
- New padding test cases: Add padding test cases to `smoke_test_fwd.sh`,
and add benchmarks to `benchmark_fwd.sh` and `benchmark_fwd_v3.sh` as well.
These test cases and benchmarks that specifically validate/benchmark the
new padding and variable-length functionalities in both group and batch modes.
* [CK_TILE] Fix build error in fmha unit tests
---------
Co-authored-by: Po Yen Chen <PoYen.Chen@amd.com>
Co-authored-by: Yi DING <yi.ding@amd.com>
* Run ctest with --output-on-failure
* Fix synchronization issues in bwd pipelines
The bwd kernel reuses the same area of LDS for ds (SGrad), bias and
dbias (BiasGrad). This means that there must be block_sync_lds between
loading one tensor and storing another to the same area.
Heavy instructions like MFMA/WMMA and global loads are executed between
reuses of the same memory so in MOST cases loading is finished by all
warps before storing is started. However, sometimes warps progress at
different speeds.
Running the tests multiple times and, preferably, with multiple
processes on the same GPU helps to trigger this issue:
bin/test_ck_tile_fmha_bwd_bf16 --gtest_repeat=-1 --gtest_shuffle --gtest_throw_on_failure
* change host using fp16 to check
* fp8 to fp8 compare
* rewrite input parameters
* add not squant
* remove some output code
* for scale = 1
* format
* saturates only for fp8
* add fp8bf16 data type
* add fp8bf16 data type
* fix test fp8 code
* add run_fp8bf16_tests
* change fmha fwd example parameter(adding fp8bf16)
* Support fp8bf16 for Aiter
* Support aiter fp8bf16 in c++
* fix comment about fp8 in readme.md
* add fp8fp32
* add fp8fp32 test
* remove range_q etc.
* format
* fix test parameters about squant and fmha example input fp8bf16 fp8fp32 data type
* add fp8bf16 to data_type function
* change colmajor to rowmajor in test_ck_tile_fmha_fwd_fp8
* format
* reset atol for fp8
* fix bug for atol
---------
Co-authored-by: rocking <ChunYu.Lai@amd.com>
Co-authored-by: asleepzzz <hanwen.chang@amd.com>
