mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-14 18:17:44 +00:00
25cca7c462
* 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
[ROCm/composable_kernel commit: 53ea4713af]
Instructions for batchnorm nhwc Example
Run batchnorm forward nhwc
# -D <xxx> : input 4-d tensor lengths
# -v <x> : verification (0=no, 1=yes)
#arg1: data type (0: fp16, 1: fp32, 3: int8, 5: bp16, 6: fp64)
#arg2: 1/0 to indicate whether to update the moving average and variance (0=no, 1=yes)
#arg3: 1/0 to indicate whether to save result mean/invVariance (0=no, 1=yes)
#arg4: initialization (0=no init, 1=single integer value, 2=scope integer value, 3=decimal value)
#arg5: time kernel (0=no, 1=yes)
./bin/example_batchnorm_forward -D 128,16,16,1024 -v 1 0 0 1 2 1
Result
./bin/example_batchnorm_forward -D 128,16,16,1024 -v 1 0 0 1 2 1
launch_and_time_kernel: grid_dim {64, 1, 1}, block_dim {256, 1, 1}
Warm up 1 time
Start running 10 times...
launch_and_time_kernel: grid_dim {120, 1, 1}, block_dim {256, 1, 1}
Warm up 1 time
Start running 10 times...
launch_and_time_kernel: grid_dim {120, 1, 1}, block_dim {256, 1, 1}
Warm up 1 time
Start running 10 times...
Perf: 2.08231 ms, 354.519 GB/s
Result
./bin/example_batchnorm_forward -D 128,16,16,1024 -v 1 0 1 0 2 0
echo $?
0
Run batchnorm infer nhwc
# -D <xxx> : input 4-d tensor lengths
# -v <x> : verification (0=no, 1=yes)
#arg1: data type (0: fp16, 1: fp32, 3: int8, 5: bp16, 6: fp64)
#arg2: initialization (0=no init, 1=single integer value, 2=scope integer value, 3=decimal value)
#arg3: time kernel (0=no, 1=yes)
./bin/example_batchnorm_infer -D 128,16,16,1024 -v 1 0 2 1
Result
./bin/example_batchnorm_infer -D 128,16,16,1024 -v 1 0 2 1
launch_and_time_kernel: grid_dim {120, 1, 1}, block_dim {256, 1, 1}
Warm up 1 time
Start running 10 times...
Perf: 1.28235 ms, 523.329 GB/s