ROCm/composable_kernel
1
0
Fork 0
mirror of https://github.com/ROCm/composable_kernel.git synced 2026-05-03 05:01:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

Reduction for int8 and bfloat16 (#125)

Browse Source 
* Use thread cluster descriptor and explicit M_K 2d descriptor to simply Blockwise Reduction

* Change by replacing ReduceDims by NumReduceDims as Device Reduce interface template parameter

* Rename the folder name for the pool2d and reduce examples

* Update to reduction test scripts

* Add Readme for pool2d_fwd and reduce_blockwise examples

* Add support for int8_t reduction (ADD/AVG, MIN/MAX/AMAX)

* Tiny fix in reduce profiler and tiny update in reduce testing scripts

* Tiny fix in testing script profile_reduce_no_index.sh

* Tiny fix in testing script profile_reduce_no_index.sh

* Add support for bfp16 reduction (using bhalf_t = ushort)

* Tiny fix in amd_buffer_addressing.hpp

* Tiny change in script/profile_reduce_with_index.sh

* Use AccDataType for Beta value and use element_wise::PassThrough

* Use type_convert for type converting in host layer reduction

* Renaming and refining in Reduction profiler/device layer/examples

* Renaming and refining in Reduction profiler/device layer/examples

* Renaming all NumReduceDims to NumReduceDim

* Fix the leaked type_convert in ThreadwiseTensorSliceTransfer_v2

* Update to testing scripts to add bf16 support

* added more static_assert

* Remove buggy tunable configurations defined in device_reduce_instance_xxx.hpp

* Add static_assert to give compile-time warning for incorrect thread slice-size/vector-size configurations

* minor change

* Refine and fix (in GetWorkspaceSizeInBytes of MultiBlockPartialReduce) to make int8 completely pass

* Tiny renaming in gridwise_2d_reduction_multiblock_partial_reduce.hpp

* Tiny fix in script/profile_reduce_no_index.sh

* Refine in DeviceReduce layer with regard to using NumInvariantDim/NumReduceDim or InvariantDims/ReduceDims

* Generic renaming in host reduction and DeviceReduce layer

* Add support for 4-d all dimension reduction in the profiler and add_device_reduce_xxx instances

* Use multi-thread and simplification for host Reduction implementation

* Add ctest for reduction

* Update to clarify the using of data init method in produce_reduce/example_reduce/test_reduce/

* Update to the reduce CTest executables to enable default testing behavior when no command argument

* Renaming

Co-authored-by: Jianfeng yan <jfyan008@gmail.com>
This commit is contained in:
Qianfeng
2022-03-23 03:35:14 +08:00
committed by GitHub
parent cb87b049de
commit 9a8ee8a39a
113 changed files with 4036 additions and 973 deletions
Download Patch File Download Diff File Expand all files Collapse all files

34
include/ck/tensor_operation/gpu/element/element_wise_operation.hpp
View File

@@ -1,6 +1,5 @@
#ifndef CK_ELEMENT_WISE_OPERATION_HPP
#define CK_ELEMENT_WISE_OPERATION_HPP
#include "data_type.hpp"
#include "data_type.hpp"
@@ -19,6 +18,8 @@ struct PassThrough
__host__ __device__ void operator()(int32_t& y, const int32_t& x) const { y = x; }
__host__ __device__ void operator()(int8_t& y, const int8_t& x) const { y = x; }
__host__ __device__ void operator()(double& y, const double& x) const { y = x; }
};
struct Add
@@ -239,6 +240,24 @@ struct UnaryIdentic<int32_t, int32_t, false>
__host__ __device__ void operator()(int32_t& y, const int32_t& x) const { y = x; };
};
template <>
struct UnaryIdentic<int32_t, int32_t, true>
{
__host__ __device__ UnaryIdentic(const int32_t divider = 1) { divider_ = divider; };
__host__ __device__ void operator()(int32_t& y, const int32_t& x) const { y = x / divider_; };
int32_t divider_ = 1;
};
template <>
struct UnaryIdentic<int8_t, int8_t, false>
{
__host__ __device__ UnaryIdentic(const int8_t divider = 1) { (void)divider; };
__host__ __device__ void operator()(int8_t& y, const int8_t& x) const { y = x; };
};
template <typename Y, typename X, bool HasDividing = false>
struct UnarySquare;
@@ -311,6 +330,19 @@ struct UnaryAbs<double, double>
__host__ __device__ void operator()(double& y, const double& x) const { y = abs(x); };
};
template <>
struct UnaryAbs<int8_t, int8_t>
{
__host__ __device__ UnaryAbs(const int32_t divider = 1) { (void)divider; };
__host__ __device__ void operator()(int8_t& y, const int8_t& x) const
{
int8_t sgn = x >> (8 - 1);
y = (x ^ sgn) - sgn;
};
};
template <typename Y, typename X>
struct UnarySqrt;