* modify comment
* trim unnecessary check
* add gemm spec in kernel name
* add TNTT gemm_gemm + atten kernel instances
* refactor attention padding to better fit in unit tests
This streamlines usage where "ResetNaNToMinusInf" is now hidden from user facing device op.
Also added compile-time conditionals that load OOB value as NaN only after padding is enabled
* add adhoc padding test for atten
* shrink input value range for attention kernel validation to avoid occasional error by 1e-3
Still unsure whether this kind of deterministic floating point accurary issue is expected
or not. May want to try exact same approach as the GPU kernel in the host reference
GEMM+Softmax+GEMM function to see if the accuracy discrepancy goes away. Until then,
shrink the input value range as it is less likely to produce errors of around ~1e-3.
* attention kernel proper granular padding for all 4 dims
* IsSupportedArgument checks
* test more padded cases
* block PadK specialization in attention kernels
* workaround clang crash for gfx908
(gfx908 only) workaround for compiler crash in fused kernels on mainline #9110; #10738 seems ok
error message was "fatal error: error in backend: Error while trying to spill VGPR0 from class
VGPR_32: Cannot scavenge register without an emergency spill slot!"
this fall back to less ideal way of handle NPadding in fused attention kernel
* comment out kernels giving wrong results on MI100; MI200 doesn't seem affected
* Refactor the design of DeviceGemmMultipleDMultipleR_Xdl_CShuffle
* Add 'DeviceGroupedConvFwdMultipleDMultipleR' interface
* Add DeviceGroupedConvFwdMultipleDMultipleR_Xdl_CShuffle
* Remove 'GridwiseConvFwdMultipleDMultipleR_xdl_cshuffle'
* Add 'TransformConvFwdToGemm<>' utility class (from Chao)
* Use 'TransformConvFwdToGemm<>' to shorten code
* Fix ill-formed method declaration
* Re-implement MakeRGridDescriptor_M() function
* Change problem description
* Use macro to define layout types
* Define K-reduced output tensor layout types
* Let user to decide R output tensor layout
* Rename variables
* Add padding to the reduced output tensor if necessary
* Extract common code as helper method
* Remove debug message
* Add missing include directive
* Add partial fp16 Conv + Reduction example
* Add example verification code for 2D Conv problem
* Use type alias to simplify code
* Share code across different-dimension Conv problems
* Rename file/functions from run_conv_fwd* to run_convnd_fwd*
* Make example code more verbose
* Add code to support 1D & 3D Conv + Reduction on host
* Add more examples for data type: bf16, fp32
* Add example for int8
* Add custom target to group examples
* Use more general custom target name
* Change the description in error message
* Disable testing for example other than fp32
* Add examplel for int4 (just copy from int8)
* Fix wrong data type
* Use larger data type for intermediate tensors
* Finish int4 example
* Undefine macro PP_DEFINE_LAYOUT_TYPE() after use
* Use named variables to replace magic numbers
* Remove debug messages
* Use same A/B data type for host Conv in int4 example
* Add check for the 'RLayout' type argument
* Group same-dim-layouts together in 'LayoutSetting<>'
* Add 'final' specifier to utility classes
* Use different initialization method for examples
* Remove macro PP_DEFINE_LAYOUT_TYPE()
* Fix code-comment mismatch
* Use more reasonable initialization value for all data types
* Default use init_method=1 for all examples
* Remove never-used code
* Remove confusing out-of-date comments
* clean
Co-authored-by: Chao Liu <chao.liu2@amd.com>
Co-authored-by: Chao Liu <lc.roy86@gmail.com>
* GEMM + Reduce max fp16+fp32
* GEmm + Max bf16 + int8
* Refactor common definitions.
* Refactor common func of mean meansquare example.
* More examples for mean meansquare.
* Update int8 examples and skip them cause of random errors.
* Int4 examples.
* Fix examples for max int4/8
* Tensor conversion for int4 input data for mean meansquare example.
* Remove int4 mean_meansquare example
* Fix int8 mean_meansquare example.
-All ReductionAccData and R<N>DataType have to be F32. The INT32 data
type is giving wrong results.
* Guard int4 with ifdef
* Change int8 example to add_addsquare due to div rounding err.
* Clang format
* Change the return type of common function.
* Get back int8 example with division.
* Remove int8 mean meansquare.
* Use proper cast for BF16 data type.
* Use ck::literals.
* Use proper data type for host tensors & reference.
- Use ReduceAccDataType for reference gemm output data type.
- Cast host reference output tensor to EDataType
- Fix ifdefs for int4.
Co-authored-by: Adam Osewski <aosewski@amd.com>
* add padding algo for bmm+scale+softmax+bmm. Version for verification
* remove verification code
* remove comments
* add padded bmm scale softmax bmm example
* format
* refactor
* add comments for usages of padding bmm+scale+softmax+bmm
Co-authored-by: Chao Liu <lc.roy86@gmail.com>
* More int4 UT.
* Disable BitwiseRepresentation UT.
* Add UT with static_cast
* Surround cout statements with #if
Co-authored-by: Adam Osewski <aosewski@amd.com>
* Add int4 example for convnd_fwd_bias_relu_add
* Fix AddReluAdd for building without int4 support
* Update CMakeLists.txt
* Format
* Convert int4 tensors for int8 kernel
* Fix device memory allocation
* Format
* Format
* comment on specialization for TensorSpecialization::Packed
* gemm_softmax_gemm with output permutation
* scaling
* refactor MatrixPadder; rename to GemmPadder
* remove old sanity check
* restore original gemm_softmax_gemm
* revise comment in gemm_softmax_gemm example
* use GetElementSpaceSize()
* remove extra header
* typo
* remove archaic DeviceOpPtr
* add examples into grouped/batched_gemm
* adding splitK examples
* fixed splitK
* add bfp16 int8 example into splitK
* formatting
* use static_cast
* added common for batched_gemm
* add commons for examples of splitK/batched/grouped_gemm
* return true
* adjust splitK check tol
* update example
Co-authored-by: Chao Liu <lc.roy86@gmail.com>
* Add custom target to bundle examples together
* Add int4 example conditionally (just copy from int8 example)
* Extract common code into common.hpp
* Move ref gemm type alias into data-type-specific sources
* Add #error directive to prevent compile with wrong setting
* Let AddAddFastGelu support int4 parameter type
* Let check_err() support int4 parameter type
* Add wrapper function to hide value conversion while copying memory
* Finish int4 example for GEMM + AddAddFastGelu
* Add new DeviceMem API to copy memory
* Use new DeviceMem API to implement examples
* Fix wrongly use of macro 'CK_EXPERIMENTAL_BIT_INT_EXTENSION_INT4'
* Revert "Add new DeviceMem API to copy memory"
This reverts commit e26e7af71e.
* Add conversion ctor for Tensor<>
* Add 'const' specifier to Tensor<>::CopyAsType()
* Convert Tensor<> values before/after transfer between host & device
* GemmPadder and GemmGemmPadder
* proper padding using GemmGemmPadder
* test gemm_gemm padding
* properly check size K in IsSupportedArgument()
* properly check size requirement given SrcScalarPerVector in IsSupportedArgument()
* comment
* format
* Introduce int4 data type.
* Add unit-tests for int4
* Compile int4 UT only when int4 enabled.
* clang-format
Co-authored-by: Adam Osewski <aosewski@amd.com>
* Implement multiple-reduction in one kernel (kernels, device ops, examples)
* Add generic elementwise kernel and device interface
* Add generator for normal-distributed data initialization
* Add host refer implementation of batchnorm-forward and batchnorm-infer
* Add examples for implementing batchnorm-forward and batchnorm-infer using generic kernels
* Remove un-needed including in batchnorm example
* Renaming generic_elementwise to elementiwise in kernel and device classes/functions
* Change in gemm_layernorm examples to use DeviceElementwise instead of Device5AryElementwise
* Change in exampe 19_binary_elementwise to use DeviceElementwise instead of DeviceBinaryElementwise
* Change in device_cgemm_4gemm_xdl_cshuffle.hpp to use kernel_elementwise instead of kernel_binary_elementwise
* Add DeviceElementwiseBase and use it in device_normalize_instance.cpp
* Removing and renaming files
* Update to synchronize gemm_layernorm client example to the generic element-wise device op API
* Update to synchronize with the latest headers directory and HostTensorDescriptor interface renaming
* Merge two static member functions in device_elementwise.hpp
* Remove unary_elementwise_1d kernel and device
* Add threadwise and blockwise welford
* Rename gridwise op, prepare to add welford version
* implement welford and integrate welford into layernorm
* Take care of tail loop
* Fix buf when ThreadSliceK > 1
* Fix bug of merging of two empty set
* Rename clip to clamp
* 1. Fix type of count
2. Remove useless static_assert
* Do not inherit Reduction::Argument
* [What] replace __syncthreads() with block_sync_lds()
[Why] __syncthreads might wait both lgkmcnt(0) and vmcnt(0)
* Add y stride
* Rename.
DeviceLayernorm -> DeviceLayernormImpl
DeviceNormalization2 -> DeviceLayernorm
* Move literal ""_uz & ""_zu into namespace 'literals'
* Move namespace 'literals' as 'ck::literals'
Co-authored-by: Po-Yen, Chen <PoYen.Chen@amd.com>
Co-authored-by: Chao Liu <chao.liu2@amd.com>
* initial stub for gemm_gemm_xdl_cshuffle
* set up example code
* compiles
* prevent integer overflow
* harmonize interface between ref_gemm and ref_batched_gemm
* batched_gemm_gemm
* fix example
* host tensor gen: diagonal pattern in lowest two-dimensions only
* make c descriptors containing only integral constants
* clean up
* add BlockwiseGemmXdlops_v2 while exploring an unified approach
* implement proper interface
* tidy up example
* fix compilation warnings
* coarsely controlled 2nd gemm padding
* remove rocm-cmake's hard requirement for certain revision
* clang-format
* resolve merge conflict
* fix compilation error on gfx10
* adds acc0 elementwise op to interface
* add gemm_gemm instances and tests
* avoid LDS data hazard
* fix build
Co-authored-by: Chao Liu <chao.liu2@amd.com>
* initial stub for gemm_gemm_xdl_cshuffle
* set up example code
* compiles
* prevent integer overflow
* harmonize interface between ref_gemm and ref_batched_gemm
* batched_gemm_gemm
* fix example
* host tensor gen: diagonal pattern in lowest two-dimensions only
* make c descriptors containing only integral constants
* clean up
* add BlockwiseGemmXdlops_v2 while exploring an unified approach
* implement proper interface
* tidy up example
* fix compilation warnings
* coarsely controlled 2nd gemm padding
* remove rocm-cmake's hard requirement for certain revision
* clang-format
* resolve merge conflict
* fix compilation error on gfx10
* adds acc0 elementwise op to interface
* attention host validation
* add blockwsie softmax v1
* iteratively update softmax+gemm
* transpose both gemm0 and gemm1 xdl output so as to avoid broadcasting softmax max/sum
* add init method for easier debugging
* do away with manual thread cluster calculation
* generalize blockwise softmax interface
* row-wise softmax sum & max
* format
* rename to DeviceBatchedGemmSoftmaxGemm
* add gemm_softmax_gemm instances and tests
* comment
Co-authored-by: ltqin <letao.qin@amd.com>
Co-authored-by: Chao Liu <chao.liu2@amd.com>
* [LWPCK-359] Initial commit
* Working version for fp16, add results to readme
* Update according to PR #341
* Update results in readme
* Add fp32 example
* Add bf16 example
* Update fp16 and fp32 examples
* Add int8 example
* Add separate lengths and strides tensors for D tensors
Co-authored-by: Rosty Geyyer <rosty.geyyer@amd.com>
* Add always_false<> util to delay symbol resolution
* Use always_false<> to prevent trying instantiate unwanted method
* Add new specializations of AddAddFastGelu::operator() method
* Add GEMM + AddAddFastGelu examples for data types: int8, bf16, fp32
* Use floating point literal to simplify code
* Remove unnecessary capture in lambda expressions
* Extract fast GeLU calculation as standalone method
* Mark methods as 'constexpr'
* Add constraint for HostTensorDescriptor templated ctors
* Simplify HostTensorDescriptor ctor calls
* Add C++23 std::size_t literal suffix
* Use _uz suffix to shorten example code
* Remove unnecessary conversion to std::array<>
* Re-order include directives
* Remove C-style casting by literal suffix
* Remove unnecessary statements in main()
* Remove unused type parameter of always_false<>
* Remove unused include directive
* Exit main() by returning meaningful value
* Use 'if constexpr' to switch example flow
* Use std::is_same_v<> to shorten example code
* Add 'inline' specifier to literal functions
* Unify output methods in example
* Move common codes into .inc file
* Add type check in type_convert<>()
* Add type_convert<float>() before computation
* Merge AddAddFastGelu method specializations
* Remove always_false<>
* Add constraint to AddAddFastGelu::operator() parameter types
* Add int8 specialization for elementwise Add and Subtract.
* CGEMM examples bf16, fp32, int8
* Add convert reference output to CDataType.
* Skip BF16 data type during testing.
* Lower K value to get rid of accumulation error.
* Fix merge artifact.
* Fix changed function name: GetElementSpaceSize()
* Fix merge artifact.
Co-authored-by: Adam Osewski <aosewski@amd.com>
* Implement layernorm kernel and deviceOp
* verify gpu kernel with host code
* 1. Separate gamma aand beta from affine
2. Check if argument is valid
* clean
* Sync the naming
* Support sweep once mode if we can put k dimension data inside one block
* [What] Get length from upper length.
[Why] if we get length directly, we may get length after padding.
* We only use one block in K dimension.
Hence, we can simplify the indexing of global R/W.
* Use 1d descriptor for gamma and beta
* Add accElementwiseOp
* Extract layernorm host code
* Support different YVectorDim in GridwiseLayernorm
* Rename XSrcVectorDim to XYSrcVectorDim. Because we use same parameter in deviceOp
* Gamma and beta can share the VGPR.
* Add test for fp32 and fp16
* Fix bug of concurrency and add test case which may fail orignally
* Propagate NaN for layernorm
Co-authored-by: Chao Liu <chao.liu2@amd.com>
* format
* improving pipeline
* fix typo
* format
* adding thread group
* adding thread group
* adding thread group
* adding gemm pipeline
* tweak
* refactor
* refactor
* add missing type convert
* refactor
* refactor
* refactor
* clean
* fix build
* refactor
* format
* clean up
* use remove_cvref_t
* clean
* use pipeline_v2 for gemm kernel
* Remove inconsistent indent
* Fix compilation errors due to incomplete merge process
* Add missing include directives
* Fix compilation errors in currently unused files
* Add license in newly added files
* Re-format touched files by clang-format-10
* Fix wrong template argument count of DeviceGemm<>
* Use language construct to choose between types
* Use language construct to choose GEMM example instance
* Fix compilation error due to interface change
* Re-use type alias to avoid duplication
* Unify type alias usage in source file
* Only use v2 pipeline in one gridwise GEMM type
* Remove no-longer used include directives
* Add static_assert() to check pipeline type requirements
* Revert "Add static_assert() to check pipeline type requirements"
This reverts commit f0985f0a13.
* clean
* clean
* clean
* clean
Co-authored-by: Chao Liu <chao.liu2@amd.com>
Co-authored-by: shaojiewang <wsjmessi@163.com>
* dump lds content in appropriate precision type
* add squared add reduction op; allows sq sum
* initial stub from regular gemm impl
* layernorm example code & host verification
* initial layernorm implementation
* tidy up
* make C0 precision type consistent with C
* clang-tidy and additional comments
* tighten up example code
* account for extra flops/bytes from normalization
* clang-format
* c0 bias/beta/gamma now have its own precision type
* AccElemOp for gemm outputs prior to feeding to layernorm
* update workgroup mapping
* rename kernel template param to reflect its dual use
* use LDS mem pool for reduction workspace
* change cshuffle precision type to f16; clean up
* clang-format
* correct naming
* explicit cast
* fully implemented gemm + bias + activation + add + norm
* activation in correct order
* reflect reduction API's recent change
* amend
* clean up; add comment
* keep up with recent changes in reduction API
* format
* resolve merge conflicts
Co-authored-by: Chao Liu <chao.liu2@amd.com>
* use 'sweep once' softmax kernel where applicable
* threadwise copy's dst buffer can specify invalid element value
* add int8 in/out float compute softmax support
give a bit of leeway for int absolute tolerance as there's a single data point of all test cases showing off-by-1 error
* format
* softmax inherits DeviceNormalization
* softmax profiler stub
* tighten up reference softmax interface
* example prints tensor dimension
* add fp32 to softmax profiler
* rename header
* hook with ckProfiler
* format
* resolve merge conflict
* resolve merge conflicts
* update normalization profiler help string
* resolve conflict
* typo
* remove residual
* softmax profiler: address feedback
* test for mixed precision input/output
* fully qualify ck::math::isnan
* add comment for device normalization interface
* revise wording
* constness for alpha/beta scaler pointer
