 compilade
|
557410b8f0
llama : greatly reduce output buffer memory usage (#6122)
|
há 1 ano atrás |
 Michael Podvitskiy
|
9fa2627347
ggml : introduce ggml_status (ggml/750)
|
há 1 ano atrás |
 Georgi Gerganov
|
ab336a9d5e
code : normalize enum names (#5697)
|
há 1 ano atrás |
 slaren
|
fbe7dfa53c
ggml : add max buffer sizes to opencl and metal backends (#5181)
|
há 1 ano atrás |
 0cc4m
|
2307523d32
ggml : add Vulkan backend (#2059)
|
há 1 ano atrás |
|
0cc4m
|
a1d6df129b
Add OpenCL add kernel (#5151)
|
há 2 anos atrás |
|
slaren
|
e7e4df031b
llama : ggml-backend integration (#4766)
|
há 2 anos atrás |
 Jared Van Bortel
|
15f5d96037
build : fix build info generation and cleanup Makefile (#3920)
|
há 2 anos atrás |
 shibe2
|
465219b914
CLBlast: Add outer loops over src0 for broadcasting in mulmat
|
há 2 anos atrás |
|
shibe2
|
1117d06607
opencl : fix element-wise multiplication (#3656)
|
há 2 anos atrás |
|
shibe2
|
40e5ce054f
CLBlast: Fix temporary buffer size for f16 conversion (wsize)
|
há 2 anos atrás |
|
shibe2
|
1e0e873c37
CLBlast: Fix matrix-vector multiplication (#3544)
|
há 2 anos atrás |
|
shibe2
|
e2583cbc29
CLBlast: Fix handling of on-device tensor data
|
há 2 anos atrás |
|
shibe2
|
665018c749
CLBlast: Add broadcast support for matrix multiplication (#3402)
|
há 2 anos atrás |
|
shibe2
|
36b904e200
ggml-opencl.cpp: Make private functions static (#3300)
|
há 2 anos atrás |
|
slaren
|
bd33e5ab92
ggml-opencl : store GPU buffer in ggml_tensor::extra (#2994)
|
há 2 anos atrás |
 Wentai Zhang
|
6460f758db
opencl : fix a bug in ggml_cl_pool_malloc() for ggml_cl_mul_mat_f32() (#2955)
|
há 2 anos atrás |
 Howard Su
|
481f793acc
Fix opencl by wrap #if-else-endif with \n (#2086)
|
há 2 anos atrás |
 Govlzkoy
|
14a2cc71f6
[ggml] fix index for ne03 value in ggml_cl_mul_f32 (#2088)
|
há 2 anos atrás |
 LostRuins
|
96a712ca1b
Porting the improved K-Quant CUDA kernels to OpenCL (#1966)
|
há 2 anos atrás |
|
Howard Su
|
3d59ec5935
ggml : fix warnings under MSVC (#1908)
|
há 2 anos atrás |
|
0cc4m
|
d411968e99
opencl : support k-quants (#1836)
|
há 2 anos atrás |
|
Howard Su
|
58970a4c39
Leverage mmap for offloading tensors to GPU (#1597)
|
há 2 anos atrás |
 Robert Sung-wook Shin
|
98ed165574
OpenCL: Add release memory (#1741)
|
há 2 anos atrás |
 Johannes Gäßler
|
17366df842
Multi GPU support, CUDA refactor, CUDA scratch buffer (#1703)
|
há 2 anos atrás |
|
LostRuins
|
d5b111f53d
Clblast fixes + enhancements to save VRAM and offload more layers (#1675)
|
há 2 anos atrás |
|
0cc4m
|
dcb2ed4826
OpenCL: Fix duplication of layers in VRAM and RAM, add GPU mul kernel (#1653)
|
há 2 anos atrás |
|
Howard Su
|
bb051d9723
opencl : no need to allocate cl_mem on heap (#1612)
|
há 2 anos atrás |
|
Howard Su
|
ca74884f66
opencl : use strstr to check if fp16 supported (#1611)
|
há 2 anos atrás |
 Maarten ter Huurne
|
7d873811f3
Fix handling of "invalid property" when creating OpenCL command queue (#1565)
|
há 2 anos atrás |
