Diego Devesa 94b87f87b5 cuda : add ampere to the list of default architectures (#11870)

2025-02-14 15:33:52 +01:00

..

template-instances

CUDA: use mma PTX instructions for FlashAttention (#11583)

2025-02-02 19:31:09 +01:00

vendors

CUDA/HIP: add support for selectable warp size to mmv (#11519)

2025-02-02 22:40:09 +01:00

acc.cu

…

acc.cuh

…

arange.cu

…

arange.cuh

…

argmax.cu

cuda : optimize argmax (#10441)

2024-11-21 18:18:50 +01:00

argmax.cuh

…

argsort.cu

…

argsort.cuh

…

binbcast.cu

CPU/CUDA: fix (GQA) mul mat back, add CUDA support (#11380)

2025-01-24 12:38:31 +01:00

binbcast.cuh

…

clamp.cu

…

clamp.cuh

…

CMakeLists.txt

cuda : add ampere to the list of default architectures (#11870)

2025-02-14 15:33:52 +01:00

common.cuh

CUDA: fix CUDART_VERSION checks (#11821)

2025-02-12 13:16:39 +01:00

concat.cu

fix: add missing msg in static_assert (#11143)

2025-01-08 20:03:28 +00:00

concat.cuh

…

conv-transpose-1d.cu

…

conv-transpose-1d.cuh

…

convert.cu

CUDA: use arch list for compatibility check (#11775)

2025-02-11 00:17:22 +01:00

convert.cuh

…

count-equal.cu

…

count-equal.cuh

…

cpy.cu

…

cpy.cuh

…

cross-entropy-loss.cu

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

cross-entropy-loss.cuh

…

dequantize.cuh

…

diagmask.cu

…

diagmask.cuh

…

fattn-common.cuh

HIP: fix flash_attn_stream_k_fixup warning (#11604)

2025-02-02 23:48:29 +01:00

fattn-mma-f16.cuh

CUDA: use mma PTX instructions for FlashAttention (#11583)

2025-02-02 19:31:09 +01:00

fattn-tile-f16.cu

CUDA: use mma PTX instructions for FlashAttention (#11583)

2025-02-02 19:31:09 +01:00

fattn-tile-f16.cuh

…

fattn-tile-f32.cu

CUDA: use mma PTX instructions for FlashAttention (#11583)

2025-02-02 19:31:09 +01:00

fattn-tile-f32.cuh

…

fattn-vec-f16.cuh

CUDA: use mma PTX instructions for FlashAttention (#11583)

2025-02-02 19:31:09 +01:00

fattn-vec-f32.cuh

CUDA: use mma PTX instructions for FlashAttention (#11583)

2025-02-02 19:31:09 +01:00

fattn-wmma-f16.cu

CUDA: fix Volta FlashAttention logic (#11615)

2025-02-03 14:25:56 +02:00

fattn-wmma-f16.cuh

CUDA: use mma PTX instructions for FlashAttention (#11583)

2025-02-02 19:31:09 +01:00

fattn.cu

CUDA: fix Volta FlashAttention logic (#11615)

2025-02-03 14:25:56 +02:00

fattn.cuh

…

getrows.cu

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

getrows.cuh

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

ggml-cuda.cu

musa: bump MUSA SDK version to rc3.1.1 (#11822)

2025-02-13 13:28:18 +01:00

gla.cu

llama: add support for QRWKV6 model architecture (#11001)

2025-01-10 09:58:08 +08:00

gla.cuh

llama: add support for QRWKV6 model architecture (#11001)

2025-01-10 09:58:08 +08:00

im2col.cu

…

im2col.cuh

…

mma.cuh

CUDA: fix min. version for movmatrix (#11751)

2025-02-08 10:46:07 +01:00

mmq.cu

HIP: Remove GCN from list of devices that avoid MMQ (#11831)

2025-02-12 22:25:28 +01:00

mmq.cuh

CUDA: use arch list for compatibility check (#11775)

2025-02-11 00:17:22 +01:00

mmv.cu

CUDA: support for mat. mul. with ne03 != ne13 (#11656)

2025-02-05 08:58:31 +01:00

mmv.cuh

…

mmvq.cu

rocBLAS: Avoid fp32->fp16->fp32 conversion on cdna (#11356)

2025-01-24 17:50:49 +01:00

mmvq.cuh

…

norm.cu

CUDA: non-contiguous (RMS) norm support (#11659)

2025-02-04 22:21:42 +01:00

norm.cuh

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

opt-step-adamw.cu

…

opt-step-adamw.cuh

…

out-prod.cu

CPU/CUDA: fix (GQA) mul mat back, add CUDA support (#11380)

2025-01-24 12:38:31 +01:00

out-prod.cuh

…

pad.cu

…

pad.cuh

…

pool2d.cu

…

pool2d.cuh

…

quantize.cu

cuda : optimize argmax (#10441)

2024-11-21 18:18:50 +01:00

quantize.cuh

…

rope.cu

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

rope.cuh

RoPE: fix back, CUDA support for back + noncont. (#11240)

2025-01-15 12:51:37 +01:00

scale.cu

…

scale.cuh

…

softmax.cu

HIP: fix flash_attn_stream_k_fixup warning (#11604)

2025-02-02 23:48:29 +01:00

softmax.cuh

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

sum.cu

CUDA: fix CUDART_VERSION checks (#11821)

2025-02-12 13:16:39 +01:00

sum.cuh

…

sumrows.cu

…

sumrows.cuh

…

tsembd.cu

…

tsembd.cuh

…

unary.cu

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

unary.cuh

CUDA: backwards pass for misc. ops, add tests (#11257)

2025-01-16 16:43:38 +01:00

upscale.cu

…

upscale.cuh

…

vecdotq.cuh

…

wkv6.cu

llama: add support for QRWKV6 model architecture (#11001)

2025-01-10 09:58:08 +08:00

wkv6.cuh

…