llama.cpp/ggml
fxzjshm 3ec9fd4b77
HIP: force max threads per block to be 1024 (#11621)
Some old/vendor forked version of llvm still use 256. Explicitly set it to 1024 to align with upstream llvm.

Signed-off-by: fxzjshm <fxzjshm@163.com>
2025-02-04 19:18:38 +01:00
..
cmake cmake: add ggml find package (#11369) 2025-01-26 12:07:48 -04:00
include CUDA: use mma PTX instructions for FlashAttention (#11583) 2025-02-02 19:31:09 +01:00
src HIP: force max threads per block to be 1024 (#11621) 2025-02-04 19:18:38 +01:00
.gitignore vulkan : cmake integration (#8119) 2024-07-13 18:12:39 +02:00
CMakeLists.txt cmake: Add ability to pass in GGML_BUILD_NUMBER (ggml/1096) 2025-02-04 12:59:15 +02:00