mozempk/ggml-org-llama-cpp-mirror
1
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2026-11-11 22:32:07 +01:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: experimental native mxfp4 support for blackwell (#17906)

Browse Source 
* CUDA: experimental native mxfp4 support for blackwell

* optimize load_tiles

* optimize quantize_mxfp4

* cleanup

* first pass review: formatting

* use interleaved layout for mma

* mmq: add assert for size

* use __nv_fp4x4_e2m1

* use iter_k as 512, cleanup

* Use 1200 as blackwell instead of 1000

* address review comments

* mmq: fix stride

* quantize.cu: use reference impl of e8m0 scale

* address review comments

* add 120f-virtual + minor fixes

---------

Co-authored-by: Aman Gupta <aman>
This commit is contained in:
Aman Gupta
2025-12-24 22:28:26 +08:00
committed by GitHub
co-authored by Aman Gupta
parent 54132f1b1f
commit c8a2417d7b
8 changed files with 426 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files
ggml/src/ggml-cuda/vendors/cuda.h
Vendored
+4
View File
@@ -10,6 +10,10 @@
#include <cuda_fp8.h>
#endif // CUDART_VERSION >= 12050
#if CUDART_VERSION >= 12080
#include <cuda_fp4.h>
#endif // CUDART_VERSION >= 12080
#if CUDART_VERSION < 11020
#define CU_DEVICE_ATTRIBUTE_VIRTUAL_MEMORY_MANAGEMENT_SUPPORTED CU_DEVICE_ATTRIBUTE_VIRTUAL_ADDRESS_MANAGEMENT_SUPPORTED
#define CUBLAS_TF32_TENSOR_OP_MATH CUBLAS_TENSOR_OP_MATH