Sigbjørn Skjæret
|
28657a8229
ggml : implement GEGLU_ERF and GEGLU_QUICK ops (#14445)
|
7 kuukautta sitten |
Jeff Bolz
|
2b72bedec1
vulkan: support mixed/deepseekR1 FA head sizes (#14509)
|
7 kuukautta sitten |
Georgi Gerganov
|
a70c8a0c4b
kv-cache : use ggml_set_rows (#14285)
|
7 kuukautta sitten |
Georgi Gerganov
|
9067487c44
ggml : fix FA mask dim 2 and 3 (#14505)
|
7 kuukautta sitten |
Jeff Bolz
|
8875523eb3
vulkan: support softmax/FA batch and broadcast (#14449)
|
7 kuukautta sitten |
Georgi Gerganov
|
ec68e84c32
ggml : support bcast ggml_soft_max_ext, ggml_flash_attn_ext (#14435)
|
7 kuukautta sitten |
Jeff Bolz
|
6a746cf9c4
vulkan: Split large mul_mat_id to fit in shared memory (#14451)
|
7 kuukautta sitten |
Sigbjørn Skjæret
|
eff5e45443
add GELU_ERF (#14455)
|
7 kuukautta sitten |
Sigbjørn Skjæret
|
a0535ffa0d
ggml : implement REGLU/GEGLU/SWIGLU ops (#14158)
|
7 kuukautta sitten |
Jeff Bolz
|
bd9c981d72
vulkan: Add fusion support for RMS_NORM+MUL (#14366)
|
7 kuukautta sitten |
Jeff Bolz
|
63a7bb3c7e
vulkan: handle noncontig in the final case of ggml_vk_get_cpy_pipeline (#14378)
|
7 kuukautta sitten |
Jeff Bolz
|
00d5282c7f
vulkan: lock accesses of pinned_memory vector (#14333)
|
7 kuukautta sitten |
Markus Tavenrath
|
bb16041cae
Add support for VK_EXT_debug_utils to add labels to Vulkan objects. (#13792)
|
7 kuukautta sitten |
0cc4m
|
10bb545c5b
Vulkan: Set device max size for host memory to avoid OOM warning and fallback to CPU buffer (#14249)
|
7 kuukautta sitten |
Jeff Bolz
|
c89c2d1ab9
vulkan: mutex around vkQueueSubmit (#14127)
|
7 kuukautta sitten |
Jeff Bolz
|
bd248d4dc7
vulkan: Better thread-safety for command pools/buffers (#14116)
|
7 kuukautta sitten |
Jeff Bolz
|
1f7d50b293
vulkan: Track descriptor pools/sets per-context (#14109)
|
7 kuukautta sitten |
0cc4m
|
97340b4c99
Vulkan: Don't default to CPU device (like llvmpipe), even if no other device is available, to allow fallback to CPU backend (#14099)
|
7 kuukautta sitten |
Masato Nakasaka
|
669c13e0f6
vulkan: Enable VK_KHR_cooperative_matrix extension for Intel Xe2 GPUs (#14001)
|
8 kuukautta sitten |
Jeff Bolz
|
5a8ae3053c
vulkan: automatically deduce size of push constants (#13936)
|
8 kuukautta sitten |
Ervin Áron Tasnádi
|
0d3984424f
ggml-vulkan: adds support for op CONV_TRANSPOSE_1D (#13813)
|
8 kuukautta sitten |
Jeff Bolz
|
7e00e60ef8
vulkan: fix warnings in perf logger querypool code (#13937)
|
8 kuukautta sitten |
Kai Pastor
|
108009f5c7
vulkan : Remove unexpected ; (ggml/1253)
|
8 kuukautta sitten |
Jeff Bolz
|
bef8176387
vulkan: use timestamp queries for GGML_VULKAN_PERF (#13817)
|
8 kuukautta sitten |
Jeff Bolz
|
fef693dc6b
vulkan: mark IM2COL as supporting non-contig (#13783)
|
8 kuukautta sitten |
Jeff Bolz
|
1dcd01960c
vulkan: support CPY from any type to itself (#13695)
|
8 kuukautta sitten |
Jeff Bolz
|
c10ed6cbcc
vulkan: Disable coopmat/coopmat2/bfloat extensions if glslc doesn't support it (#13696)
|
8 kuukautta sitten |
Judd
|
a127ff1780
use LOG_WARN to replace `std::cerr` (#13657)
|
8 kuukautta sitten |
Eve
|
fb1cab201c
vulkan: fix warnings (#13626)
|
8 kuukautta sitten |
0cc4m
|
8960efd0a6
Vulkan: Add f32 accumulator support to quantized mul mat to fix GLM4 32B incoherence (#13607)
|
8 kuukautta sitten |