* Add indexing support to pooling operator
- Add IndexDataType template parameter to pooling problem and kernel
definitions
- Enable pooling kernel to output indices of selected elements during
max/absmax pooling
- Add overloaded operators for Max and AbsMax that track when values
change using bool changed parameter
- Support optional index buffer allocation and management in device
memory
- Modify BlockReduce2d classes to handle index tensors alongside value
tensors
- Add separate shared memory allocation for index data in cross-warp
reductions
- Create validate_pool_indices function to verify index correctness
- Modify pool3d.cpp example to demonstrate index output functionality
- Add tests for index output
* fixes
* Refactor BlockReduce2D functions to get rid auxiliary private types.
* comment resolutions and some changes to block_reduce2d
- index reference implementation improved
- reduce_operator.hpp cleanedup
- updated the block_reduce2d.hpp to have index calculation for
BlockReduce2dLinearCrossWarpSync as well
* conditionally used variable declaration improvement
- the conditionally used vairbales are used only when indexing is
enabled. To inform the compiler that they may be unused and declare them
with least size possible. This may allow it to be optimized compared to
the previous declarations
* comment resolutions
* lexical ordering of the indicies
- introduced accumulate methods that handle the intermediate steps if
needed to order the indexes
* add reduce_operator_accumulate.hpp to core.hpp
---------
Co-authored-by: Adam Osewski <Adam.Osewski@amd.com>
* Implement argument passing to element-wise functions for fwd convolution
* Add files for fwd + bias + clamp example
* Implement Bias
* Implement Clamp
* Elementwise function composition
* Composition unit test
* Implement fwd + bias + clamp example
* Simplify argument passing and composition
* elfunc -> bias_and_clamp
* Rename function to specify example
* Move element-wise function instantiation to kernel
* Make bias a runtime tensor
* No ugly namespace aliasing
* Initialize element-wise function on host
* Remove function initialization helper, simplify Compose initialization
* Remove unintended LSP compatibility patch
* Clean up includes and unused code
* Switch names in cshuffle epilogue
* Move CDElementwise to conv traits
* Re-add required include
* Initialize bias in same way as other tensors
* Better type specification for ds pointer
* Disable 1D convolution
* Add warning for non-group-constant bias
* [CK_TILE] Correct BlockWarps calculation and fix smoke-test in rmsnorm
* Update rmsnorm host reference
* Update tree reduction of rmsnorm for reference host
* Fix cross warp for m > 1 cases
* Add RMSNorm model selectable option for host reference
* Fix save_unquant cases
* Update reference rmsnorm forward function to use enum for model sensitivity
* Update reference rmsnorm calculation for model sensitivity
* Fix m warp for layernorm
* Adjust parameter of reference for twoPass
* Fix clang format
* Run clang-format-overwrite.sh to fix formating issue
* fix clang format
---------
Co-authored-by: MHYang <mengyang@amd.com>
Co-authored-by: illsilin_amdeng <Illia.Silin@amd.com>
Co-authored-by: ThomasNing <thomas.ning@amd.com>
* Initial commit. create batched_contraction_kernel file
* initial problem definition
* implement initial example to launch kernel
* add universal gemm to contraction. initial phase
* complete implementation for special case all Dims are 1 and no Ds
* clean code
* initial changes to support multi dimensional G
* more progress in implementing multiple G
* tmp commit
* manage dynamic NumDimG in kernel
* improving example for multi M,N,K,G handling. start generalizing kernel. it is a temporary commit
* implement the example for general Multi dimension G M N K and test different reference calculation algorithms
* 2 functions for reference using multi dimensional and flat indexing
* clean the code for muti dimentional G, M, N, K contraction and add some logs
* Add Make descriptor function in kernel for merging Ms, Ns, Ks for A, B, E
* some cleaning on kernel
* clean the code for calculating the offsets from flatten batch number
* Start adding MultiD support to kernel and example
* more changes to manage multi D in kernel and example
* manage passing multi d to kernel and testing.
* complete multi D support in kernel. modify example code to support it
* Correct algorithm to calc the correct offset values for D tensor batches and some code cleaning
* Minor fix
* Generalize example code for variable NumD tensors and apply cleanup based on review feedback
* Refactored code and addressed review feedback
* refactoring, cleaning, add documents, in kernel side and example codes
* Optimize batch offset calculation in kernel
* Inline CalculateBatchOffset in batched contraction kernel, update CHANGELOG.md
---------
Co-authored-by: Adam Osewski <19374865+aosewski@users.noreply.github.com>
* Pooling 2D/3D with refernce
* Tests & cleanup
- added test for ppoling
- cleanup
- removed 2d example
* Comment resolution
- README added
- example target name rectified
- appropriate arg description and comments added
* clang-format
* appropriate blocksize calc
* modifications for future indexing addition
- instead of transforming views we now transform the descriptors, so
that the same descriptor can be re-used for index tensor in the future
* some basic fixes
* comment resolutions
* comment resolutions
---------
Co-authored-by: Illia Silin <98187287+illsilin@users.noreply.github.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
* rename gemm_group_quant to gemm_quant
* Add TensorWise quant mode
* Cshuffle epilogue tests with tensor scaling
* Add tensor quant to example
* Don't use readfirstlane for reading scales - doesn't work for some reason
* Add to changelog
* revert include - from a merge problem?
* revert common.hpp include
* revert host.hpp include
* remove unused utility function
* rename quant pipeline problem
* refactor quant tests
* remove aquant utils
* use TEST_F
* fix all tests by changing gemm config
* Use typed tests
* fix copyright
* Improve random number generation
* use different seed for each input (Q, K, V...);
* use deterministic generation of:
* seqstart_q/k (for group mode);
* block_table (for paged-kvcahe);
* cache_batch_idx (for kvcache);
* Extract arg_parser-related code from run functions to use them as tests
* Split examples into main programs and fmha runners, build instances separately
* Add dummy tests that use instances and runners
* Fix a missed corner case of f32->f8 conversion
When value if < min f8 denormal but > min f8 denormal / 2, it must be
rounded to min f8 denormal (i.e. 0b1), not to 0.
* Fix incorrect fp8 scales for P and O in validation code
DataTypeConfig was incorrectly compared with fp8_t.
* Add host generation of dropout random values and use it for validation
Previously host validation (reference_batched_dropout) used random
numbers generated by BlockDropout of the kernel, meaning that incorrect
generation on device (bad distribution, repeated numbers, too many zeros,
etc.) would not trigger any validation errors.
* Implement tests from smoke_test_bwd.sh
* Return result as enum to distinguish failure and missing instance
* Add tests for bwd features: bias, alibi, dropout
* Implement tests from smoke_test_fwd.sh
* Pass seqlen_q/k as vectors to fwd and bwd runners
* Add tests for fwd features: bias, alibi, dropout
* Add tests for pagedkv and splitkv
* Fix conditions when to use splitkv and pagedkv kernels
splitkv was executed only when use_kvcache which == (need_append_kvcache || use_cache_batch_idx || 0 < page_block_size).
In the SplitKV tests: the regular fwd kernel was executed if use_cache_batch_idx was not requested even when num_splitkv > 1.
In the AppendKV tests: the pagedkv kernel was executed but it often failed to find an instance.
* Add tests for appendkv
* Use is_v_rowmajor = true because there are no instances with column layout anymore
* Split public and private compile options for instances
Tests and examples need to know only about CK_TILE_FMHA_FWD_*_API.
* Improve parsing validation in bias and mask
* Pass bias as string for consistency with mask
* Catch parsing and other exceptions
* Add bwd test for deterministic flag
* Initialize fp8 tensors (-init=ufq) similarly to uf
* Fix splitkv/pagedkv invocation: use padded sk when seqlen_k_ptr is not null
seqlen_k cannot be used to determine padding when seqlen_k_ptr is
provided. The actual seqlen_k is taken from seqlen_k_ptr[b].
Even seqlen_k values (% bn0 == 0) use padded seqlen_k while seqlen_k_ptr
may contain arbitrary values.
In the example or tests this produces incorrect results with appendkv
(for example, -d=32 -s=1 -s_k=64 -s_knew=7 -vlayout=c -b=8).
* Fix use_pagedkv value when kvcache = true but page_block_size = 0
In this case block_table_ptr is nullptr which is accessed in the kernel.
* Clean up bwd tests
* Unify fwd tests for f16/bf16 and fp8
* Use better explicit instantiation declaration for fmha_bwd<2>
* Use the same seed for all tests, allow to override it with env variable
* Undo clang-format of one irrelevant file
For some reason my local clang-format-18 and the one in CI work differently.
* Do not build instances and tests on unsupported archs
* Build instance libraries as OBJECT library
* CI: Enable sccache for HIP
There are source files with LANGUAGE HIP, they need
-DCMAKE_HIP_COMPILER_LAUNCHER=sccache
* Add tests to REGRESSION_TESTS
* Fix OOB accesses in deterministic bwd due to incorrectly assumed kN0
The runner assumes kN0 = (hdim_q <= 128) ? 128 : 64 but there are
smaller tiles (for tr_load or fp32). This can create too small dq_acc_buf.
* Pass CK_TILE_FMHA_FWD_*_API as INTERFACE compile options
The instances don't actually depend on them, only examples and tests do.
Passing these definitions as INTERFACE allows to change FMHA_FWD_ENABLE_APIS
without recompiling instances that are already in ccache.
* Fix formatting and names
- Add support for tensor A/B in both fp16+pk_int4_t and fp8+pk_int4_t formats
- Implement A(bf8) B(i4) support in universal GEMM
- Use new implementation for i4 to fp8 conversion in Block Scale
* Add cshuffle epilogue test
* add the poc implementation to the epilogue and tests
* refactor cshuffle epilogue
* WIP: adding tensor/tile usage to scale_tile
* fix usage of tile_elementwise_inout
* add gemm_quant_kernel for generalizing gemm quant kernel
* Add problem specific to different quants, add QuantType to Traits
* Add quant_type to quant_kernel template parameters
* Create aq/bq_block_windows and views depending on QuantType
* Use tile windows as inputs in cshuffle epilogue
* Fix some issues in epilogue
* initial new example code for new general gemm quant kernel test
* Fix issues in kernel
* Add verification check for rowcol Quantmode
* use AccDataType instead of AQ in pipeline
* fix aquant preshuffle
* fix formatting
* some cleanup
* remove gemm_aquant_basic.cpp
* remove gemm_aquant_kernel.hpp
* fix tests for the renamed quant kernel
* fix formatting
* clean example files
* fix some merge conflicts
* fix preshufflequant rename issue
* fix some templates after merging with develop
* fix test preshuffle parameter
* fix formatting
* Unify bquant kernel to the common quant kernel
* remove bquant kernel also from common header
* fix formatting
* clean up commented code
* fix formatting config hpp
* fix merge mistake
* Non-const for movable windows
* fix formatting
* Fix grammar in README
Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>
* Remove #include<bit> and clean up example
* fix strides
* Add some descriptions for move_windows
---------
Co-authored-by: Mohsen Saffari <mohsen.saffari@amd.com>
Co-authored-by: spolifroni-amd <Sandra.Polifroni@amd.com>
* base working version for single groupped conv bwd data
* Fix 2d descriptor
* fix groups
* Add 3d support
* fixes
* fixes
* fixes
---------
Co-authored-by: Jakub Piasecki <jakpia21@gmail.com>
* General 2D Reduction Kernel
* Move the reduction kernel from the example
* Split the code and add the necessary policy, problem, shape files as
per ck_tile convention
* Add/modify the headers
* Modified the example to work with the 'new' kernel
* Added tests for the kernel
* N-D refernce reduce
* Added support for N-D input with transform to 2D
* Added padding to support various input sized tensors
* Bug fix in the thread buffer constructor
* Some comments to explain the reduce2d block kernel
* comments resolution
* clang-format
* comments resolution
* clang-format
* clang-format
* comments resolution
* clang-format
Resolves R_X86_64_32 relocation out of range errors in grouped conv2d instances
by splitting debug information into separate .dwo files.
Add explicit cast to avoid signed/unsigned comparison warning.
* Elementwise kernel implementation
Co-authored-by: Sami Aario <samaario@amd.com>
Co-authored-by: Mohsen Saffari <mohsen.saffari@amd.com>
Co-authored-by: yashagar <yashagar@amd.com>
* Elementwise with generalized nDims
* Adding the n-ary input tensor feature
* Generalize dimensions on top of inputs
* Add TFLOPS + remove std usage for tuples
* 1D basecase optimization
* Cleanup code + refactoring to a common interface
* Generalize to unary and add an example
* Cleanup, refactoring and commenting
* Suggestions for LWPCK-3170: elementwise kernel improvements
* Clang-format: remod.py
* Replace InputTensorType with XDataType as the type of input_tensors
* Add Tuple::apply and use it in ElementWiseKernel::operator to call operation with the exact number of arguments in xs
* Move examples to folder 19_elementwise
* Add missing copyright headers and fix some existing ones
* Replace an assert with throw std::runtime_error in elementwise example
* Avoid reading the output by using make_static_distributed_tensor for y_tile
* Removed two unused includes
* No need to move windows to the next block when each workgroup processes a single tile
* Only copy input tensors to the device
* Use get_warp_size to obtain warp size, and use ceiling division for grid size also for the unary example
* Adding output strides to the kernel, transposition example and update the other examples
* Changes made by remod.py
* Use default template parameter values for memory operation and coherence in a call to make_naive_tensor_view
* Move binary operations to include/ck_tile/ops/elementwise/binary_elementwise_operation.hpp
* Reuse generic reference binary/unary operation in examples + refactoring the transpose reference
* Fix comments in elementwise_example.cpp
- Refer to AMD terminology except when suggesting NVIDIA alternatives in parentheses
- ElementWiseTraits was renamed to ElementWiseShape
- Adopt suggestions made by Copilot when prompted to check for factual or typographical errors
* Simplify CMakeLists.txt and remove the unused variables this uncovers
* Rename a file and fix some copyright statements
* Changes made by script/clang-format-overwrite.sh
* Add basic unit test for ElementWiseKernel
* Remove left-over uninformative comment in apply unit test
* Changes made by clang-format-overwrite.sh
* fixup! Use default template parameter values for memory operation and coherence in a call to make_naive_tensor_view
* Clean up test_tuple_apply.cpp and test_elementwise_1d.cpp
* Use make_uniform_array_with_factory to define h_xs and d_xs_mems_owner as type std::array
* Use a DeviceMem constructor that calls get_element_space_size_in_bytes internally
* Move examples to folder 20_elementwise
* Reduced register pressure on the CK tile elementwise kernel + add 4d input example to be able benchmark against old CK
* Fix CLang formating
* Bump up the elementwise example folder number
* Elementwise: add padding + minor cleanup
* Add Vector Size inference + fix issue with wrong vectorization due to missing GuaranteedLastDimensionVectorStride setting in make_naive_tensor_view
* Add isSupportedArg to Elementwise kernel + addapt example and unit tests
* Fix clang-format on the unit test file
---------
Co-authored-by: Damien Lejeune <damien.lejeune@amd.com>
Co-authored-by: Sami Aario <samaario@amd.com>
Co-authored-by: Mohsen Saffari <mohsen.saffari@amd.com>
Co-authored-by: Aviral Goel <aviral.goel@amd.com>
* ck_tile kernel for gemm with groupwise quantized A or B tensor.
This change introduces new pipelines with Intrawave scheduler and block gemm primitives that loads the scale tensor to registers to perform dequantization post MFMA on C tensor in registers.
Scale tensor data, AQ/BQ is spliced across threads in registers and not stored in LDS.
Current support is for the following combinations, but it should be fairly straightforward to extend support to more formats.
1. fp8, fp8 -> f32
2. bf8, bf8 -> f32
3. i4, fp8 -> f32
4. i4, bf8 -> f32
Group size can go down to as low as K length of underlying WarpGemm primitive.
For Gemm problems with quantized B tensor, this change also introduces preliminary support for flatmm pipeline which loads B tensor directly into registers.
* [Block Scale Gemm] Only run gemm quant examples on __gfx94__
- Only run gemm quant examples on __gfx94__ for usage of
`v_cvt_pk_fp8_f32`
- Format the code
* [Block Scale Gemm] Remove Bquant Gemm BlockScale
This cleanup is in preparation for future development of bquant. By
isolating Aquant-related code, we can streamline the codebase and make
it easier to add and maintain bquant functionality in subsequent
updates.
* [Block Scale Gemm] Format code with clang-format-12
The latest clang-format (v19) in ROCm 7.0 generate different result than
clang-format-12 which is used in CK CI.
Format code with clang-format-12 for consistency.
* [Block Scale Gemm] Split the k direction loop
- Split the k direction loop in block_universal_gemm_as_quant_bs_cr.hpp
to make the logic clearer.
- Disable C transposition.
* [Block Scale Gemm] Move block scale gemm example to 38_block_scale_gemm
* [Block Scale Gemm] Update copyright
* test
* Add TailHandler
* Move TileDistributionEncodingPatternAQ
* Refactor
* refactor
* fix bug
* fix bug
* help solve the PR comment
* Format the code
* [Block Scale Gemm] Add unit tests
* [Block Scale Gemm] Add support to 16x16x32 MFMA
- Add support to 16x16x32 MFMA
- Fix a bug when exchange data crossing lanes
---------
Co-authored-by: Vijay Krishnamoorthy <vjkrish@meta.com>
Co-authored-by: Cong MA <congma13@ctr2-alola-ctrl-01.amd.com>
Co-authored-by: ThomasNing <thomas.ning@amd.com>
* Multiple d, initial commit
* Check Ds Layout
* Readme and clang format
* Update branch & conflicts
* Multiple D - fix clang-formatter
* Rename elemetwise_op
* Fix CI
* Code review part1
* Remove printf
* Remove unnecessary comment
* Add new tests with Col layout
* Review part 2
* Added support for Multiple D GEMM
* Update comment
* Remove maybe_unused
* Clang-format
* Review part 3
* Add comment to function
* Add comment to function: another
* Take number of params for a refrence function
* Remove additional d param for 0 tensor
* Change name of function
* Fix CI fails
* 50ms -> 28ms
* Fix bug in non fuse_add_store cases
* Fine tuned setting for 2 pass pipeline
* adjust workload
* remove unnecessary change
* add layernorm
* Adding output quant and unquant results at the same time.
* fix test
* fix format
* tune for cases 128x640 and 128x1024
* bug ifx
* Support bf16/fb8/bf8 datatypes for ck_tile/gemm
* remove commented out code.
* Addressing code review comments and enabling universal_gemm for all the supported data types.
* Merge conflict resolution.
* Solve the memory pipeline compilation error. Merge with the new change of CShuffle
* finish the feature, pass the tests
* Fix the pipeline and add the benchmark script for other data types
---------
Co-authored-by: ThomasNing <thomas.ning@amd.com>
* Add shortcut to RMSNorm
* Modify test for adding shortcut for RMSNorm
* Add fused parameter into tests
* 1. Add YDataType. 2. rmsnorm2d_fwd_traits_ from rmsnorm2d_fwd.hpp to rmsnorm2d_fwd_api.cpp and rmsnorm2d_fwd_instance_common.hpp
* 1. Supports various stride and percisions.
* Add support of Epilogue
* Add fuse and epilogue support to rmsnorm ref
* Modify rmsnorm example
* Refactor tests/examples
* Bug fix for newly added tests/examples
* Bug fix for new tests 2
* Modify smoke test scripts
remove dbg code
* Supports non-smooth dyanmic quant
* Update Rmsnorm2dFwd::GetName()
* rename xscale and prec_sx to smoothscale and prec_sm
Bug fix after rename
Remove files
* change example_rmsnorm2d_fwd.cpp
* update performance calculator
* Fix issue in two-pass when fuse add is enabled
* Remove comment of beta
---------
Co-authored-by: rocking <ChunYu.Lai@amd.com>
* add prenorm/postnorm support, refactor using generate.py
* update README
* update README
* fix format
* update some description and fix format
* update format
* format
* use non-raw for loading
* format and update n4096
* dynamic-quant ready
* update readme
* support fused dynamic-quant
* update fused-quant, with smooth
* update README
* update args
* update some based on comment
* CK-Tile GEMM with memory bound pipeline.
* Memory bound gemm pipeline.
* Fix not closed namespace.
* Block gemm mem pipeline draft.
* Do not use ck_tile:: within ck_tile namespace.
* Refactoring & Move Layout info to pipeline problem.
* Get hot loop and TailNum information before lunching kernel.
* Fixes in pipeline.
* Add comment to load_tile_raw and change variable naming style.
* Few small changes & formatting.
* Do not use macro.
* Add gtests.
* Use AccDataType for Output of MFMA instruction.
* Formatting.
* Refactor gemm examples.
* Switch over to current block gemm.
* Use currently available pipeline policy.
* Refactoring and review comment.s
* Fixes after merge.
* Add missing include.
* Add load tile overload which accepts output tensor as parameter.
* This give 8% perf boost at the cost of using more registers.
* Rename example.
* Small changes.
* Fix compilation err and lower K.
* Support different layouts for A/B
* Fix vector size for different layouts.
* Rename Alignment into VectorSize
* Unblock tests.
* Add reduce2d new api
* Prevent user use cross warp reduction
* Fix bug of std caculation
* Add rmsnorm2d
* Add rmsnorm small example
* Remove static assert to prevent compile fail
* Add script to test performance and correctness
* Add missing cmake change
* refine naming
* refine example of rmsnorm
* Fix bug of rmsnorm
* Refine naming
* Fix cmake
* clang format
* Refine pipeline name
* Add add_rmsnorm2d_rdquant kernel
* Add reduce op
* host verification
* Fix bug of one pass pipeline
* Refine tile size
* Add two pass pipeline
* Rename two pass to three pass
* Fix bug of kSaveX == false
* Add instance library
* Add test script
* Fix bug of x verification
* Add save_x to trait
* Add README
* Move reduce2d into reduce folder
* Fix bug of welford when number of m warp > 1
* remove reduncant comment
* 1. move 06_rmsnorm2d to 10_rmsnorm2d
2. move 07_add_rmsnorm2d_rdquant to 11_add_rmsnorm2d_rdquant
* clang format and add missing header
* Add host validation of add + layernorm2d + rsquant
* Revert "Add host validation of add + layernorm2d + rsquant"
This reverts commit 936cb45797.
* Remove deprecated flag
* port layernorm
* change warp_welford.hpp
* Update warpshuffle
* 1. Add save mean and save std back
2. Move construction of tensor_view and tile_window to operator()
* refine welford max count calculation
* unify layernorm api
* Rename file
* Remove save mean and inv std
* Revert "refine welford max count calculation"
This reverts commit 022365802b.
* Fix order of parameter
* refine welford max count calculation again
* Remove fp32 instances
* Fix bug of padding
* refactor api
* Support bf16
* Extract common function
* Refine arg of operator()
* Add kMThreadPerBlock to template parameter
* clang format
* Refine variable name
* Refine file name
* remove redundant line
* refactor layernorm2d pipeline and add block-per-block utility
* fix name
* rename more
* add more block-per-tile instance
* remove duplicated define
* update instance for 2048, 1024 case
* support up to 2048 now
* opt loading
* add n1536
* Add two pass pipeline
* format
* Fix incorrect type
* parallel compilation
* Use smaller N
* fix 2p pass
* Support Repeat_M in distribution
* Refine nameing
* Add reduce example
---------
Co-authored-by: letaoqin <letaoqin@amd.com>
Co-authored-by: aska-0096 <haocwang@amd.com>
Co-authored-by: rocking <ChunYu.Lai@amd.com>
Co-authored-by: carlushuang <carlus.huang@amd.com>
* ake the cshuffle compilable
* modify Mhe reference on gpu and cpu. Correaccess of cshuffle
* fix the cpu reference code
* Complete the in tile shuffle logic
* restructure the kernel template input
* change the naming pattern of ck_tile gemm pipeline
* Re-format files using remod.py
* Solve the fmha conflict with gemm
* Comment Addressed from Carlus
---------
Co-authored-by: Po Yen, Chen <PoYen.Chen@amd.com>
* Finished the feature of gpu verification
* Add the ck_tile_gemm test in the CI CD
* add the include of tensor_layou in reference_gemm
* Comment Addressed
* split ck_tile fhma and gemm tests into separate stages
* restructure the reference gemm
* restructure a new reference_gemm api that could read the device mem
---------
Co-authored-by: carlushuang <carlus.huang@amd.com>
Co-authored-by: illsilin <Illia.Silin@amd.com>
* Checkpoint: Finished with the tile example & kernel verification, working on the different matrix layout
* Finished the Matrix Layout feature set up. Note: Need to modify the inner block to solve the shuffle problem in the future.
* Fix: Clang Format, API fixed from fmha
* fix with better naming convention
* revert back the pipeline code of fmha
* Fixed: Addressed the comments and merge the GEMM shape of GEMM Operator and FMHA Operator to one.
* clang format with the reference_gemm file
* convert the clang format with the remod.py
* Changed the format and variable name of the kernel gemm_shape and partitioner
---------
Co-authored-by: thomasning <thomasning@banff-cyxtera-s70-4.ctr.dcgpu>
* Use dictionary to config all the functions
* Add init codegen logic for fmha fwd appendkv
* Call HIP_CHECK_ERROR() macro to get real source info
* Setup meaningfull arguments
* Sync kernel name with the codegen
* Add knew/vnew tensors to the kernel argument
* Fix wrong K values after appending
* Fix vnew append errro
* Extract common logics
* Fix Vnew tile dstr for row major case
* Conditionally add fwd_splitkv API in fmha_fwd example
* Conditionally add call to fmha_fwd_splitkv()
* Remove "EXAMPLE_" prefix of cmake variables
* Regsiter API handlers automatically
* Early return if 0 < s_k_new is not supported
* Show message if we are ignoring option
* Unify CMakeLists.txt coding style
* Set num_splits=1 if split-kv is not supported
* Add length/stride getters for HostTensor
* Add RoPE example utilities
* Add reference_rotary_position_embedding() (not implemented)
* Finish reference_rotary_position_embedding() impl
* Fix typo of HostTensor<>::get_length()
* Fix compilation errors
* Fix wrong answer when interleaved=false
* Fix wrong answer when interleaved=true
* Append K/V in the host verification code
* Simplify K appending logics
* Simplify v_host_ref definition
* Reduce input/output dimensions
* Rename function: add "batched" prefix
* Apply RoPE on host side
* Rename RoPE utility function
* Fix wrong tensor size
* Avoid invoking deprecated method 'find_module'
* Pass RoPE kernel args
* Create Rotary Cos/Sin tile windows in kernel
* Add compute data type alias for RoPE
* Randomly generate seqlen_knew if needed
* Fix seqlen_knew enabling check logic
* Add minimum seqlen_k to generate compliance kvcache
* Fix compilation error in debug mode
* Fix wrong boundaries
* Fix wrong seqlen_k for kvcache
* Rename variables used in distributio encoding
* Fix rotary cos/sin tensor/tile size
* Add constraint to the rotary_dim option
* Remove unused inner namespace
* Add dram distribution for rotary_cos/rotary_sin (interleaved)
* Only apply interleaved RoPE on Knew for now
* Fix wrong thread starting offset
* Instantiate multiple kernels for RoPE approaches
* Clean-up pipeline
* Fix error in RoPE host reference
* Handle RoPE half-rotated logics
* Support 8x rotary_dim under half-rotated RoPE
* Add comment
* Apply elementwise function to the loaded tiles
* Unify parameter/variable naming style
* Remove constness from q_ptr
* Add code blocks for q_tile
* Apply RoPE to q_tile
* Remove debug print code in kernel
* Fix wrong knew/vnew appending positions
* Use better naming for tile indices
* Add make_tile_window() for adding distribution only
* Skip code if # of block is more than needed
* Move thread locating logics into policy
* Remove always true static_assert()
* Rename header
* Rename RotaryEmbeddingEnum
* Extract rotary embedding logic out
* Re-order parameters
* Align naming of some tile size constants
* Rename more tile size constants
* Fix wrong grid size
* Fix wrong shape of knew_host/vnew_host
* Fix wrong index into knew_host/vnew_host
* Fix wrong rotary_cos/rotary_sin memory size for Q
* Extract Q/Knew vector size to helper methods
* Use different rotary_cos/rotary_sin distr for Q/Knew
* Update host/device specifiers
* Fix wrong data type for Q rotary_cos/rotary_sin
* Remove RoPEComputeDataType type alias
* Shift rotary_cos/rotary_sin by cache_seqlen_k
* Add comment for why I just 't' for all padding flags
* Align commit message to the real comment
* Fix wrong pipeline
* Rename utility function
* Disable host verification if API not exist
* Fix wrong rope key for fp8 pipeline
* Allow only apply RoPE on Q (without append KV)
* Add append-kv smoke tests
* Remove debug statements
* Remove more debug statements
* Re-arrange the 'set +x' command
* Remove no-longer used method in pipeline
* Add missing init code
* Refine pipeline padding settings
* Enlarge rotary_dim limit (8 -> 16)
* Enlarge KPerThread for rotary_interleaved=false
* Update rotary_dim range in smoke_test_fwd.sh
* Add template argument 'kIsPagedKV' for splitkv kernels
* Launch splitkv kernel if given page_block_size
* Fix wrong kernel name
* Fix seqlen_k_min for pre-fill case (1 -> 0)
* Add copy_const<> type trait
* Add another make_tile_window()
* Introduce 'TileWindowNavigator' types
* Simplify TileWindowNavigator interfaces
* Fix tile window navigation bugs
* Disable calling fmha_fwd()
* Remove ununnecessary data members
* Simplify more make_tile_window() overloads
* Move V tile through TileWindowNavigator
* Fix uneven split checking logic
* Move code after decide seqlen_q/seqlen_k
* Make sure we always start reading complete tile
* Use 128 as minimus page_block_size
* Fix wrong origin for bias
* Add batch_stride_k/batch_stride_v in group mode
* Unify origin
* Add missing kernel arguments for group mode
* Add paged-kv codegen logic for appendkv kernels
* Add block_table kernel args for appendkv kernel
* Add tile navigators to the appendkv kernel
* Fix wrong tensor descriptor lengths
* Pass re-created tile window to pipeline
* Fix wrong strides for appendkv kernel
* Allow transit tile_window to another page-block
* Handle cross-page-block write
* Donot perform write again if already in last page-block
* Always add fmha_fwd() api
* Add missing group mode argument
* Remove debug macro usages
* Rename option s_k_new to s_knew
* Separate splitkv/non-splitkv args/traits
* Remove fmha_fwd_dispatch()
* Fix compilation errors
* Remove dropout code in splitkv kernel
* Allow problem types without define kHasDropout attr
* Use generic lambda to init traits objects
* Separate more non-splitkv & splitkv traits/args
* Display more info for specific kernels
* Show more detailed warning message
* Rename 'max_num_blocks' to 'max_num_page_blocks'
* Remove no-longer used pipeline files
* Wrap code by #if directives
* Move functors to the begining of validation code
* Use generic lambda to init all the api traits/args
* Fix wrong seqlen for kvcache
* Add missing comment
* Rename TileWindowNavigator to PageBlockNavigator
* Only expose necessary methods (not attributes)
* Re-order pipeline paremeters
* Refine smoke_test_fwd.sh
* Fix wrong arugment count
* Make tile window directly via PageBlockNavigator
* Remove unused template paremeter
* Remove group mode from appendkv kernel
* Fix skcheck logic
* Fix wrong syntax in skcheck expr
* Use meaningful options in smoke test
* Remove options
* Fix formatting
* Fix more format
* Re-organize bash functions
* Pass cache_batch_idx to kernels
* Support cache_batch_idx in example
* Fix compilation error
* Add more appendkv test
* Add more case for appendkv
* Fix unexisted attribute
* Remove 0 < seqlen_knew constraint
* Clarify the case in warning message
* Remove macro checking
* Force batch mode when invoking appendkv & splitkv apis
* Fix mode overriding logics
* Fix wrong parameter name
* Randomize seqlen_k if use kvcache
* Use randomized seqlen_k for kvcache
* Avoid using too small rotary_cos & rotary_sin
* Rename parameter
* Add seqlen_q & seqlen_k rules
* Add comment
* Add more comments
* Fix compilation errors
* Fix typo in comment
* Remove type argument
* Avoid seqlen_k=0 for kvcache
* Revert "Avoid seqlen_k=0 for kvcache"
This reverts commit 21c4df89e4.
* Fix wrong uneven split checking logics
* Only randomize kvcache seqlen_k if 1 < batch
* Return earlier if split is empty
* Revert "Only randomize kvcache seqlen_k if 1 < batch"
This reverts commit b9a4ab0d7e.
* Re-order seqlen_k_start adjustment logics
* Fix compilation errors
* Re-format script
* Find executable from folder automatically
* Fix kvcache seqlen_k generating logic
* Make comment more clear
* Fix wrong knew/vew appending logic on host
* Add s_barrier to sync threads
* Revert "Add s_barrier to sync threads"
This reverts commit d3f550f30c.
* Support only using 1 row of rotary_cos/rotary_sin
* Rotate Q in different way
* Unify tensor view creation logics
* Fix wrong argument
* Add mask to switch how we use the rotary_cos/sin
* Move attr from traits to problem
* Move has_mask to fmha_fwd_appendkv_args
* Support use uint32_t as SAD operand in Alibi<>
* Use sad_u32() in splitkv kernels
* Store tensor views in PageBlockNavigator
* Use stored tensor view to update tile windows
* Enlarge tensor view size
* Remove debug code
* Fix wrong tensor view size
* Wrap tensor view into PageBlockNavigator
* Add DataType member to PageBlockNavigator
* Remove unnecessary member functions
* Refind macro use
* Fix typo
* Add blank line between directives and actual code
* Re-format files
* Remove type in comment
---------
Co-authored-by: carlushuang <carlus.huang@amd.com>
Co-authored-by: rocking <ChunYu.Lai@amd.com>
* Add layernorm2d forward
* Refind file path
* clang format
* Exclude ck_tile op from all
* use add_executable instead
* refactor layernorm2d_fwd example
---------
Co-authored-by: carlushuang <carlus.huang@amd.com>
