composable_kernel
composable_kernel copied to clipboard
Composable Kernel: Performance Portable Programming Model for Machine Learning Tensor Operators
## Proposed changes Please describe the motivation behind the pull request, whether it enables a new feature or fixes a bug. If there are associated pull requests or issues, please...
## Proposed changes Please describe the motivation behind the pull request, whether it enables a new feature or fixes a bug. If there are associated pull requests or issues, please...
### Problem Description when build with `make tile_example_gemm_basic -j` gives: ```sh In file included from /workspace/composable_kernel/example/ck_tile/03_gemm/gemm_basic.cpp:12: In file included from /workspace/composable_kernel/include/ck_tile/host.hpp:7: In file included from /workspace/composable_kernel/include/ck_tile/host/check_err.hpp:16: In file included from...
### Problem Description hi, ck team, I am trying with `CK_TILE_PIPELINE_MEMORY 2` used in universal_gemm example, the default V3 pipeline works well, while this gives: ```c++ In file included from...
## Proposed changes This change should be refactored. It will be better to add support for multiple ABD and use scale as B1. It will be developed after support for...
Either at compile-time or run-time determine which FP8 interpretation to use. ## Checklist Please put an `x` into the boxes that apply. You can also fill these out after creating...
### Problem Description the `__attribute__((address_space([x])))` (where `[x]` is `0` or `1`, `2`, `3`) defined in `include/ck_tile/core/config.hpp`, line:55~60+ is not working. PR:#1930 fixed this issue, but indeed introduced compile error. steps...
### Problem Description in [example/ck_tile/03_gemm/gemm_basic.cpp](https://github.com/ROCm/composable_kernel/tree/develop/example/ck_tile/03_gemm) with not supported args, e..g -m=128, -n=128, -k=32, it will hangs at : ```c++ if(!Kernel::IsSupportedArgument(kargs)) { throw std::runtime_error("Wrong! Arguments not supported! Skipping gemm!\n"); } ```...
## Proposed changes Please describe the motivation behind the pull request, whether it enables a new feature or fixes a bug. If there are associated pull requests or issues, please...
### Problem Description I'm getting `invalid operand for instruction` error while compiling inline assembly `v_cmpx_le_u32` and `buffer_load_dword`. I tried with `-DDL_KERNELS=On` and Off, but nothing changed. Am I doing something...