mirror of
https://github.com/ROCm/composable_kernel.git
synced 2026-05-18 20:09:25 +00:00
259055e53f
* Add GEMM MX BF6 example
* Fix BF6 type_convert
* Add type_convert for bf16x6
* Add compare operator to f4x2_pk_t
* Update README for 67_gemm_microscaling
* Fix host tensor initialization with integer values for FP8
[ROCm/composable_kernel commit: 518dc21ae8]
27 lines
693 B
Markdown
27 lines
693 B
Markdown
# GEMM Examples for Microscaling Formats
|
|
|
|
## example_gemm_mx_fp8
|
|
|
|
Custom verification parameters:
|
|
```bash
|
|
# arg1: verification (0=no, 1=CPU)
|
|
# arg2: initialization (0=constant values, 1=integer values, 2=decimal values)
|
|
# arg3: time kernel (0=no, 1=yes)
|
|
# arg4: verbosity (0=no info, 1=verbose info)
|
|
# arg5 to 10: M(256x), N(256x), K(512x), StrideA, StrideB, StrideC
|
|
# arg11: KBatch
|
|
# arg12: warmup runs pre-timing
|
|
# arg13: repeat run count for timing
|
|
./bin/example_gemm_mx_fp8 1 1 0 1
|
|
```
|
|
|
|
Custom tensor shapes:
|
|
```bash
|
|
./bin/example_gemm_mx_fp8 1 2 1 0 256 256 512 -1 -1 -1 1 10 10
|
|
```
|
|
|
|
Default invocation:
|
|
```bash
|
|
# Implies: ./bin/example_gemm_mx_fp8 1 2 0 0
|
|
./bin/example_gemm_mx_fp8
|
|
``` |