ccb87fa3ee
[CUDA] Increase number of output elements per-thread block if the K-dimension is small ( #20635 )
...
* Increase per-thread work if the K-dimension is small
With tensor parallelism, the K-dimension of the FFN-down matrices is split, which makes it quite small, especially for MOEs. For example, Qwen3-30b-A3B has a K-dimension of 768, and Qwen3235B-A22B has k-dimension of 1536.
The current heuristic uses a group of 4 warps irrespective of K-dimension size, resulting in some of the threads being idle. This results in poor performance for these matrices.
This change increases the number of output elements per block for such cases.
* Limit this change to ncols_dst = 1
* tab to space
2026-03-22 16:49:35 +08:00
617db241aa
cuda : add RDNA4-specific MMVQ parameter table for bs=1 decode ( #19478 )
...
* mmvq: add RDNA3/RDNA4-specific parameter table (nwarps=8, rows=1)
* mmvq: add dedicated RDNA3 parameter table
* mmvq: exclude RDNA3.5 (gfx1150/1151) from RDNA3 table
2026-03-15 08:33:39 +01:00
8bece2eb20
CUDA: use mmvq for mul-mat-id for small batch sizes ( #18958 )
...
* CUDA: use mmvq for mul-mat-id for small batch sizes
* add mmvq too
* Fix perf issue on ampere. Use mmvf mm-id only for non-nvidia GPUs
* templatize multi_token_path
2026-02-03 23:31:23 +08:00
d3dc9dd898
CUDA: Remove unneded bias/gate dims in fused mmvq ( #16858 )
...
* CUDA: Remove unneded bias/gate dims in fused mmvq
Pointed out
[here](https://github.com/ggml-org/llama.cpp/pull/16847#discussion_r2476798989 )
that only a single value is needed per target col per thread
* Apply suggestions from code review
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
* Fix "Error 991-D: extra braces are nonstandard" during compilation
---------
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
2025-11-01 13:13:26 +08:00
8b11deea46
Hide latency of bias and gate-loading ( #16847 )
...
This is realised by loading them into registers before computation of
the dot-product, effectively batching them together with said
dot-product. As a lot of threads are alive here, the warp scheduler has
enough threads available to effectively hide the cost of additionally
loading those two floats.
2025-10-30 11:34:15 +08:00
463bbf20bf
CUDA: add unused vars to mmvf and mmvq ( #16807 )
2025-10-28 10:31:21 +08:00
f77c13b91f
CUDA: General GEMV fusion ( #16715 )
2025-10-26 19:28:04 +08:00
0f7c69689f
musa: fix build warnings ( #15611 )
...
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
2025-09-26 02:56:10 +02:00
5143fa895e
CUDA: fastdiv, launch bounds for mmvq + q8_1 quant ( #15802 )
...
* CUDA: fastdiv, launch bounds for mmvq + q8_1 quant
2025-09-05 16:07:02 +02:00
8ad038c0fd
musa: add GGML_UNUSED_VARS ( #15446 )
...
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
2025-08-21 11:06:05 +08:00
fd1234cb46
llama : add gpt-oss ( #15091 )
...
* oai moe
* compat with new checkpoint
* add attn sink impl
* add rope scaling yarn
* logits match with latest transformers code
* wip chat template
* rm trailing space
* use ggml_scale_bias
* rm redundant is_swa_all
* convert interleaved gate_up
* graph : fix activation function to match reference (#7 )
* vocab : handle o200k_harmony special tokens
* ggml : add attention sinks support (#1 )
* llama : add attn sinks
* ggml : add attn sinks
* cuda : add attn sinks
* vulkan : add support for sinks in softmax
remove unnecessary return
* ggml : add fused swiglu_oai op (#11 )
* ggml : add fused swiglu_oai op
* Update ggml/src/ggml-cpu/ops.cpp
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* update CUDA impl
* cont : metal impl
* add vulkan impl
* test-backend-ops : more test cases, clean up
* llama : remove unfused impl
* remove extra lines
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
---------
Co-authored-by: slaren <slarengh@gmail.com>
* repack mxfp4 upon conversion
* clean up a bit
* enable thinking
* add quick hack to render only some special tokens
* fix bf16 conversion
* remove vocab hack
* webui ok
* support chat parsing for gpt-oss
* fix webui
* direct mapping mxfp4, FINALLY
* force using mxfp4
* properly use lazy tensor
* ggml : add mxfp4
ggml : use e8m0 conversion instead of powf
Co-authored-by: Diego Devesa <slarengh@gmail.com>
change kvalues_mxfp4 table to match e2m1 (#6 )
metal : remove quantization for now (not used)
cuda : fix disabled CUDA graphs due to ffn moe bias
vulkan : add support for mxfp4
cont : add cm2 dequant
* ggml : add ggml_add_id (#13 )
* ggml : add ggml_add_id
* add cuda impl
* llama : add weight support check for add_id
* perf opt
* add vulkan impl
* rename cuda files
* add metal impl
* allow in-place ggml_add_id
* llama : keep biases on CPU with --cpu-moe
* llama : fix compile error
ggml-ci
* cuda : add fallback for __nv_cvt_e8m0_to_bf16raw
ggml-ci
* cleanup
ggml-ci
* sycl : fix supports_op for MXFP4
ggml-ci
* fix Unknown reasoning format
* ggml-cpu : fix AVX build
ggml-ci
* fix hip build
ggml-ci
* cuda : add mxfp4 dequantization support for cuBLAS
ggml-ci
* ggml-cpu : fix mxfp4 fallback definitions for some architectures
ggml-ci
* cuda : fix version required for __nv_cvt_e8m0_to_bf16raw
---------
Co-authored-by: Xuan Son Nguyen <son@huggingface.co>
Co-authored-by: slaren <slarengh@gmail.com>
2025-08-05 22:10:36 +03:00
7474e00b34
CUDA: fix crash with partial offloading of MoE ( #13439 )
2025-05-11 16:09:33 +02:00
2356fb1d53
CUDA: fix bad asserts for partial offload ( #13337 )
2025-05-06 13:58:51 +02:00
9070365020
CUDA: fix logic for clearing padding with -ngl 0 ( #13320 )
2025-05-05 22:32:13 +02:00
e1e8e0991f
CUDA: batched+noncont MMQ, refactor bs>1 MoE code ( #13199 )
2025-04-30 23:12:59 +02:00
b10d8bfdb1
CUDA: use switch statements in constexpr functions ( #13095 )
2025-04-24 15:57:10 +02:00
658987cfc9
CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID ( #13014 )
...
* CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID
* fix logic for RoPE support, CUDA graphs
2025-04-22 21:27:40 +02:00
492d7f1ff7
musa: fix all warnings, re-enable `-DLLAMA_FATAL_WARNINGS=ON` in ci and update doc ( #12611 )
...
* musa: fix all warnings
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
* musa: enable -DLLAMA_FATAL_WARNINGS=ON in run.sh
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
* musa: update ci doc (install ccache)
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
* fix Windows build issue
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
* Address review comments
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
* Address review comments
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
---------
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
2025-03-30 10:59:38 +02:00
bd40678df7
HIP: Add support for RDNA4 targets ( #12372 )
2025-03-26 23:46:30 +01:00
10f2e81809
CUDA/HIP: refractor mmqv to unify the calculation of nwarps and rows per block between host and device code. ( #12177 )
...
refactor mmqv to unify the calculation of nwarps and rows per block between host and device code.
---------
Co-authored-by: Johannes Gäßler <johannesg@5d6.de>
2025-03-11 20:16:03 +01:00
9fbadaef4f
rocBLAS: Avoid fp32->fp16->fp32 conversion on cdna ( #11356 )
2025-01-24 17:50:49 +01:00
750cb3e246
CUDA: rename macros to avoid conflicts with WinAPI ( #10736 )
...
* Renames NVIDIA GPU-architecture flags to avoid name clashes with WinAPI. (e.g. CC_PASCAL, GPU architecture or WinAPI pascal compiler flag?)
* Reverts erroneous rename in SYCL-code.
* Renames GGML_CUDA_MIN_CC_DP4A to GGML_CUDA_CC_DP4A.
* Renames the rest of the compute capability macros for consistency.
2024-12-10 18:23:24 +01:00
3ad5451f3b
Add some minimal optimizations for CDNA ( #10498 )
...
* Add some minimal optimizations for CDNA
* ggml_cuda: set launch bounds also for GCN as it helps there too
2024-11-27 17:10:08 +01:00
ae8de6d50a
ggml : build backends as libraries ( #10256 )
...
* ggml : build backends as libraries
---------
Signed-off-by: Xiaodong Ye <xiaodong.ye@mthreads.com>
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
Co-authored-by: R0CKSTAR <xiaodong.ye@mthreads.com>
2024-11-14 18:04:35 +01:00
2b1f616b20
ggml : reduce hash table reset cost ( #8698 )
...
* ggml : reduce hash table reset cost
* fix unreachable code warnings after GGML_ASSERT(false)
* GGML_ASSERT(false) -> GGML_ABORT("fatal error")
* GGML_ABORT use format string
2024-07-27 04:41:55 +02:00
cb5fad4c6c
CUDA: refactor and optimize IQ MMVQ ( #8215 )
...
* CUDA: refactor and optimize IQ MMVQ
* uint -> uint32_t
* __dp4a -> ggml_cuda_dp4a
* remove MIN_CC_DP4A checks
* change default
* try CI fix
2024-07-01 20:39:06 +02:00
f3f65429c4
llama : reorganize source code + improve CMake ( #8006 )
...
* scripts : update sync [no ci]
* files : relocate [no ci]
* ci : disable kompute build [no ci]
* cmake : fixes [no ci]
* server : fix mingw build
ggml-ci
* cmake : minor [no ci]
* cmake : link math library [no ci]
* cmake : build normal ggml library (not object library) [no ci]
* cmake : fix kompute build
ggml-ci
* make,cmake : fix LLAMA_CUDA + replace GGML_CDEF_PRIVATE
ggml-ci
* move public backend headers to the public include directory (#8122 )
* move public backend headers to the public include directory
* nix test
* spm : fix metal header
---------
Co-authored-by: Georgi Gerganov <ggerganov@gmail.com>
* scripts : fix sync paths [no ci]
* scripts : sync ggml-blas.h [no ci]
---------
Co-authored-by: slaren <slarengh@gmail.com>
2024-06-26 18:33:02 +03:00
Gaurav Garg
PikaPikachu
Aman Gupta
Oliver Simons
R0CKSTAR
Johannes Gäßler
Georgi Gerganov
R0CKSTAR
Slobodan Josic
uvos
uvos
Andreas Kieslinger
Diego Devesa