Jeff Bolz
|
8423d01931
vulkan: Optimize SSM_SCAN (#16645)
|
3 сар өмнө |
Jeff Bolz
|
e56abd2098
vulkan: Implement topk_moe fused shader, ported from CUDA (#16641)
|
3 сар өмнө |
Giuseppe Scrivano
|
3d4e86bbeb
vulkan: Add State Space Model (SSM) Operations Support (#16463)
|
3 сар өмнө |
Jeff Bolz
|
4258e0cfe7
vulkan: Support FA with K/V in F32 (#16543)
|
3 сар өмнө |
Jeff Bolz
|
2aaf0a2a20
vulkan: Replace uses of maxMemoryAllocationSize and VK_WHOLE_SIZE (#16354)
|
3 сар өмнө |
Jeff Bolz
|
e308efda8e
vulkan: in flash attention, bounds check against nem1 (don't rely on GGML_KQ_MASK_PAD) (#16316)
|
3 сар өмнө |
Eve
|
132d673554
vulkan: make ggml_vk_default_dispatcher support older vulkan headers (#16345)
|
4 сар өмнө |
Jeff Bolz
|
d8359f5fde
vulkan: 64-bit im2col (#16135)
|
4 сар өмнө |
Jeff Bolz
|
1384abf8b8
vulkan: handle mat_mul with A matrix > 4GB (#16176)
|
4 сар өмнө |
Acly
|
8656f5de68
vulkan : make the vulkan.hpp dynamic dispatcher instance private (#16224)
|
4 сар өмнө |
Dmytro Minochkin
|
0499b29c6f
vulkan: throw system error instead of SIGABRT during init on older devices (#16156)
|
4 сар өмнө |
Jeff Bolz
|
3f81b4e91c
vulkan: support GET_ROWS for k-quants (#16235)
|
4 сар өмнө |
Sigbjørn Skjæret
|
3ecb2f671a
ggml : implement set_rows with i32 index (#16159)
|
4 сар өмнө |
Shin-myoung-serp
|
96fdca043b
Vulkan: add conv_transpose_2d operation (#16022)
|
4 сар өмнө |
Jeff Bolz
|
a20d810d79
vulkan: add RTE variants of exp shader (#16165)
|
4 сар өмнө |
Giuseppe Scrivano
|
1eeb523c3e
vulkan: optimize UMA buffer operations and fix driver hangs (#16059)
|
4 сар өмнө |
Jeff Bolz
|
5bb4a3edec
vulkan: fix validation error about VK_PIPELINE_CREATE_CAPTURE_STATISTICS_BIT_KHR (#16086)
|
4 сар өмнө |
Jeff Bolz
|
c0b45097c3
rename optimize_graph to graph_optimize (#16082)
|
4 сар өмнө |
Eve
|
cb5bb6cc05
vulkan: automatically remove unsupported devices (#15976)
|
4 сар өмнө |
Ruben Ortlam
|
261e6a20ff
Vulkan: Clean up mul_mm shader (#15987)
|
4 сар өмнө |
Jeff Bolz
|
b9c9c9f789
vulkan: initialize vulkan-hpp to allow using extension function pointers (#15705)
|
4 сар өмнө |
Ruben Ortlam
|
304ac5693d
Vulkan iGPU device selection overhaul and PCI ID API support (#15947)
|
4 сар өмнө |
Mathieu Baudier
|
6c88ad8fa7
vulkan: Make device memory check more portable (#15939)
|
4 сар өмнө |
Diego Devesa
|
360d6533db
ggml-backend : add GGML_BACKEND_DEVICE_TYPE_IGPU device type (#15797)
|
4 сар өмнө |
Ruben Ortlam
|
ae355f6f71
vulkan: throw the oom error instead of no memory type found (#15905)
|
4 сар өмнө |
Jeff Bolz
|
4f63cd705c
vulkan: Fix OOB accesses in soft_max_back (#15861)
|
4 сар өмнө |
lksj92hs
|
ed54e32558
Workaround for subgroup arithmetic failing on MoltenVK with AMD GPUs (issue 15846) (#15886)
|
4 сар өмнө |
Jeff Bolz
|
e68aa10d8f
vulkan: sort graph to allow more parallel execution (#15850)
|
4 сар өмнө |
Xuan-Son Nguyen
|
9fcb29f22f
ggml: allow casting between f32 and i32 (#15783)
|
4 сар өмнө |
Jeff Bolz
|
3976dfbe00
vulkan: support im2col_3d (#15795)
|
4 сар өмнө |