 Sigbjørn Skjæret
|
229bf68628
cuda : fix argsort with 64k+ rows (#16849)
|
hace 2 meses |
 Jeff Bolz
|
b9ce940177
vulkan: Fuse rope+set_rows (#16769)
|
hace 2 meses |
 Acly
|
10640e31aa
ggml : fix interpolate with align-corners and ne=1 (#16700)
|
hace 2 meses |
 Aman Gupta
|
75cbdd3fce
test-backend-ops: print failed tests at the end (#16785)
|
hace 2 meses |
|
Aman Gupta
|
75d33b9302
CUDA: support for weight clamp in top-k norm (#16702)
|
hace 2 meses |
 leejet
|
bbac6a26b2
ggml: fix cuda kernel launch configuration for k_compute_batched_ptrs to support large batch (#16744)
|
hace 2 meses |
|
Aman Gupta
|
f77c13b91f
CUDA: General GEMV fusion (#16715)
|
hace 2 meses |
|
leejet
|
55945d2ef5
ggml: fix CUDA grid launch condition for large block_nums.y in binbcast (#16742)
|
hace 2 meses |
|
Aman Gupta
|
03792ad936
CUDA: topk-moe: add optional parameter for gpt-oss (#16649)
|
hace 2 meses |
 safranowith
|
2330de7b84
SYCL: Add support for FLOOR,CEIL,ROUND and TRUNC unary operators (#16613)
|
hace 3 meses |
 Ilia Ilmer
|
9ad4f1931e
metal : add `CONV_TRANSPOSE_2D` (#16542)
|
hace 3 meses |
 lhez
|
0cb7a0683b
opencl: add q8_0 mm support (#16469)
|
hace 3 meses |
 Sam/Samuel
|
f4ce81c45e
metal: optimise `GGML_OP_SUM` (#16559)
|
hace 3 meses |
|
Aman Gupta
|
48e2fa9fb7
CUDA: add fp kernel for larger batch size MoE (#16512)
|
hace 3 meses |
 Georgi Gerganov
|
e60f241eac
metal : FA support F32 K and V and head size = 32 (#16531)
|
hace 3 meses |
|
Georgi Gerganov
|
0a319bb75e
metal : add support for non-padded FA KV (#16148)
|
hace 3 meses |
|
Georgi Gerganov
|
1d6092fc72
tests : add -INF blocks to the KQ mask in the FA tests (#16380)
|
hace 3 meses |
 Reese Levine
|
ef07a40906
ggml webgpu: add support for soft_max, optimize rms_norm (#16357)
|
hace 3 meses |
|
Reese Levine
|
8d78cd2613
ggml webgpu: support for rope,div,sub,glu,scale,cont operators (#16187)
|
hace 3 meses |
|
Jeff Bolz
|
a74a0d69f3
tests: override test_set_rows::max_nmse_err to allow for occasional rounding differences (#16295)
|
hace 3 meses |
|
Sigbjørn Skjæret
|
adc76347d7
ggml : check cuda and metal argsort limits and add test (#16323)
|
hace 3 meses |
|
Sigbjørn Skjæret
|
b887d2f341
ggml : fix GGML_F32_VEC_FMA argument order in ggml_vec_mad1_f32 (#16307)
|
hace 3 meses |
|
Jeff Bolz
|
d8359f5fde
vulkan: 64-bit im2col (#16135)
|
hace 3 meses |
|
Georgi Gerganov
|
6a2c6145a0
metal : extend mat-mat multiplication support (#16225)
|
hace 3 meses |
|
Jeff Bolz
|
1384abf8b8
vulkan: handle mat_mul with A matrix > 4GB (#16176)
|
hace 3 meses |
|
Aman Gupta
|
c0bfc57af4
CUDA: mul_mat_id for mmf for bs <= 64 for f16 and bs <= 32 for f32 (#16277)
|
hace 3 meses |
|
Aman Gupta
|
077c94d0ca
CUDA: add a fused top-K MoE kernel (#16130)
|
hace 3 meses |
|
Georgi Gerganov
|
dfcd53f7ec
metal : fuse NORM + MUL + ADD, support non-multiples of 4 (#16220)
|
hace 3 meses |
|
Sigbjørn Skjæret
|
3ecb2f671a
ggml : implement set_rows with i32 index (#16159)
|
hace 3 meses |
 Shin-myoung-serp
|
96fdca043b
Vulkan: add conv_transpose_2d operation (#16022)
|
hace 3 meses |
