composable_kernel

mirror of https://github.com/ROCm/composable_kernel.git synced 2026-04-19 22:39:03 +00:00

Files

coconutruben fcd4a6f3d1 device_prop.hpp - replace map with compile time hash and switch (#1898 )

* device_prop.hpp - replace map with compile time hash and switch

Summary:

We replace a static const map with a compile time hash function
and a switch statement to achieve the same goal: translate names
to architectures. Most of these are very old, however the function
needs to continue to work.

Why? because the static map can cause issues when compiling into
libraries that get dynamically loaded/unloaded, leading to memory
corruption

Test Plan:

Running pytorch `torch.compile()` with CK enabled, and seeing it not
segfault on the 2nd kernel (1st reload of the library)

Reviewers:

Subscribers:

Tasks:

Tags:

* clang-format

2025-02-24 09:57:55 -08:00

device_prop.hpp

device_prop.hpp - replace map with compile time hash and switch (#1898 )

2025-02-24 09:57:55 -08:00

flush_cache.hpp

add int8 gemm multiply multiply a8w8 (#1591 )

2024-10-26 16:39:34 +08:00

hip_check_error.hpp

[A8W8 GEMM] Optimized weight-preshuffled implementation & add quantization datatype for CK TILE rms_norm (#1862 )

2025-02-20 14:00:27 -08:00

io.hpp

update copyright headers (#726 )