Aman Gupta 55a1c5a5fd CUDA: add softmax broadcast (#14475) ...

* CUDA: add softmax broadcast

* Pass by const ref

* Review: Use blockDims for indexing, remove designated initializers

* Add TODO for noncontigous input/output

2025-07-02 15:48:33 +03:00

..

template-instances

CUDA: FA support for Deepseek (Ampere or newer) (#13306)

2025-05-09 13:34:58 +02:00

vendors

CUDA/HIP: Share the same unified memory allocation logic. (#12934)

2025-04-15 11:20:38 +02:00

acc.cu

llama/ggml: add LLM training support (#10544)

2025-05-12 14:44:49 +02:00

acc.cuh

…

arange.cu

…

arange.cuh

…

argmax.cu

…

argmax.cuh

…

argsort.cu

…

argsort.cuh

…

binbcast.cu

…

binbcast.cuh

…

clamp.cu

cuda: unary ops as float + de-duplicate (ggml/1130)

2025-03-03 18:18:11 +02:00

clamp.cuh

…

CMakeLists.txt

CUDA: FA support for Deepseek (Ampere or newer) (#13306)

2025-05-09 13:34:58 +02:00

common.cuh

musa: enable fp16 mma (all) and cublas on qy2 (#13842)

2025-06-26 12:11:59 +08:00

concat.cu

musa: fix all warnings, re-enable -DLLAMA_FATAL_WARNINGS=ON in ci and update doc (#12611)

2025-03-30 10:59:38 +02:00

concat.cuh

…

conv2d-dw.cu

CUDA: add conv_2d_dw (#14265)

2025-06-20 09:50:24 +08:00

conv2d-dw.cuh

CUDA: add conv_2d_dw (#14265)

2025-06-20 09:50:24 +08:00

conv2d-transpose.cu

CUDA: add conv_2d_transpose (#14287)

2025-06-20 22:48:24 +08:00

conv2d-transpose.cuh

CUDA: add conv_2d_transpose (#14287)

2025-06-20 22:48:24 +08:00

conv-transpose-1d.cu

musa: fix all warnings, re-enable -DLLAMA_FATAL_WARNINGS=ON in ci and update doc (#12611)

2025-03-30 10:59:38 +02:00

conv-transpose-1d.cuh

…

convert.cu

CUDA: add bf16 and f32 support to cublas_mul_mat_batched (#14361)

2025-06-29 01:30:53 +08:00

convert.cuh

CUDA: add bf16 and f32 support to cublas_mul_mat_batched (#14361)

2025-06-29 01:30:53 +08:00

count-equal.cu

…

count-equal.cuh

…

cp-async.cuh

CUDA: FA support for Deepseek (Ampere or newer) (#13306)

2025-05-09 13:34:58 +02:00

cpy.cu

musa: Upgrade MUSA SDK version to rc4.0.1 and use mudnn::Unary::IDENTITY op to accelerate D2D memory copy (#13647)

2025-05-21 09:58:49 +08:00

cpy.cuh

ggml: Re-enable CUDA graphs in presence of CONT and DUP nodes (#12970)

2025-04-17 15:19:42 +02:00

cross-entropy-loss.cu

…

cross-entropy-loss.cuh

…

dequantize.cuh

…

diagmask.cu

llama : reorganize source code + improve CMake (#8006)

2024-06-26 18:33:02 +03:00

diagmask.cuh

…

fattn-common.cuh

CUDA: broadcasting for FlashAttention mask (#14500)

2025-07-02 15:48:33 +03:00

fattn-mma-f16.cuh

CUDA: broadcasting for FlashAttention mask (#14500)

2025-07-02 15:48:33 +03:00

fattn-tile-f16.cu

CUDA: broadcasting for FlashAttention mask (#14500)

2025-07-02 15:48:33 +03:00

fattn-tile-f16.cuh

…

fattn-tile-f32.cu

CUDA: broadcasting for FlashAttention mask (#14500)

2025-07-02 15:48:33 +03:00

fattn-tile-f32.cuh

…

fattn-vec-f16.cuh

CUDA: broadcasting for FlashAttention mask (#14500)

2025-07-02 15:48:33 +03:00

fattn-vec-f32.cuh

CUDA: broadcasting for FlashAttention mask (#14500)

2025-07-02 15:48:33 +03:00

fattn-wmma-f16.cu

CUDA: broadcasting for FlashAttention mask (#14500)

2025-07-02 15:48:33 +03:00

fattn-wmma-f16.cuh

…

fattn.cu

CUDA: faster Deepseek FA, add Turing support (#13435)

2025-05-14 16:08:20 +02:00

fattn.cuh

…

getrows.cu

CUDA: fix crash on large batch size for MoE models (#13384)

2025-05-09 12:14:04 +02:00

getrows.cuh

CUDA: batched+noncont MMQ, refactor bs>1 MoE code (#13199)

2025-04-30 23:12:59 +02:00

ggml-cuda.cu

CUDA: add softmax broadcast (#14475)

2025-07-02 15:48:33 +03:00

gla.cu

…

gla.cuh

…

im2col.cu

…

im2col.cuh

…

mean.cu

CUDA: add mean operation (#14313)

2025-06-22 12:39:54 +08:00

mean.cuh

CUDA: add mean operation (#14313)

2025-06-22 12:39:54 +08:00

mma.cuh

musa: fix all warnings, re-enable -DLLAMA_FATAL_WARNINGS=ON in ci and update doc (#12611)

2025-03-30 10:59:38 +02:00

mmq.cu

CUDA: fix crash on large batch size for quant. MoE (#13537)

2025-05-14 16:41:02 +02:00

mmq.cuh

cuda : remove nrows_x in mul_mat_q_process_tile (#13325)

2025-05-07 09:48:23 +02:00

mmv.cu

CUDA/HIP: optimize mmv paths taken for HIP devices (#14324)

2025-06-24 01:12:56 +02:00

mmv.cuh

CUDA: mul_mat_v support for batch sizes > 1 (#14262)

2025-06-23 13:11:31 +02:00

mmvq.cu

CUDA: fix crash with partial offloading of MoE (#13439)

2025-05-11 16:09:33 +02:00

mmvq.cuh

CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (#13014)

2025-04-22 21:27:40 +02:00

norm.cu

llama: Add support for RWKV v7 architecture (#12412)

2025-03-18 07:27:50 +08:00

norm.cuh

llama: Add support for RWKV v7 architecture (#12412)

2025-03-18 07:27:50 +08:00

opt-step-adamw.cu

…

opt-step-adamw.cuh

…

out-prod.cu

…

out-prod.cuh

…

pad.cu

musa: fix all warnings, re-enable -DLLAMA_FATAL_WARNINGS=ON in ci and update doc (#12611)

2025-03-30 10:59:38 +02:00

pad.cuh

…

pool2d.cu

…

pool2d.cuh

llama : reorganize source code + improve CMake (#8006)

2024-06-26 18:33:02 +03:00

quantize.cu

CUDA: fix crash on large batch size for quant. MoE (#13537)

2025-05-14 16:41:02 +02:00

quantize.cuh

CUDA: batched+noncont MMQ, refactor bs>1 MoE code (#13199)

2025-04-30 23:12:59 +02:00

rope.cu

…

rope.cuh

…

scale.cu

…

scale.cuh

…

softmax.cu

CUDA: add softmax broadcast (#14475)

2025-07-02 15:48:33 +03:00

softmax.cuh

…

ssm-conv.cu

fix MUSA compiler warning (#12704)

2025-04-03 09:32:55 +02:00

ssm-conv.cuh

ggml : faster ssm scan (#10558)

2025-03-31 18:05:13 +02:00

ssm-scan.cu

CUDA/HIP: fix ssm_scan on devices where warp size is not 32 (#14196)

2025-06-15 17:30:13 +02:00

ssm-scan.cuh

ggml : faster ssm scan (#10558)

2025-03-31 18:05:13 +02:00

sum.cu

llama/ggml: add LLM training support (#10544)

2025-05-12 14:44:49 +02:00

sum.cuh

…

sumrows.cu

CUDA: add mean operation (#14313)

2025-06-22 12:39:54 +08:00

sumrows.cuh

CUDA: add mean operation (#14313)

2025-06-22 12:39:54 +08:00

tsembd.cu

…

tsembd.cuh

…

unary.cu

ggml : implement REGLU/GEGLU/SWIGLU ops (#14158)

2025-06-29 11:04:10 +02:00

unary.cuh

ggml : implement REGLU/GEGLU/SWIGLU ops (#14158)

2025-06-29 11:04:10 +02:00

upscale.cu

musa: fix all warnings, re-enable -DLLAMA_FATAL_WARNINGS=ON in ci and update doc (#12611)

2025-03-30 10:59:38 +02:00

upscale.cuh

…

vecdotq.cuh

CUDA: noncont MMVQ + batched bs1 MUL_MAT_ID (#13014)

2025-04-22 21:27:40 +02:00

wkv.cu

llama: Add support for RWKV v7 architecture (#12412)

2025-03-18 07:27:50 +08:00

wkv.cuh

llama: Add support for RWKV v7 architecture (#12412)

2025-03-18 07:27:50 +08:00