llama.cpp/ggml
Jeroen Mostert 46e47417aa
Allow all RDNA2 archs to use sdot4 intrinsic (#8629)
The check gating the use of `__builtin_amdgc_sdot4` specifically checks for gfx1030. This causes a severe perf regression for anything gfx103? that's not gfx1030 and not using `HSA_OVERRIDE_GFX_VERSION` (if you've built ROCm to support it). We already have a generic RDNA2 define, let's use it.
2024-07-23 10:50:40 +02:00
..
cmake llama : reorganize source code + improve CMake (#8006) 2024-06-26 18:33:02 +03:00
include CUDA: fix partial offloading for ne0 % 256 != 0 (#8572) 2024-07-18 23:48:47 +02:00
src Allow all RDNA2 archs to use sdot4 intrinsic (#8629) 2024-07-23 10:50:40 +02:00
.gitignore vulkan : cmake integration (#8119) 2024-07-13 18:12:39 +02:00
CMakeLists.txt cmake : install all ggml public headers (#8480) 2024-07-18 17:47:12 +03:00