]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
overview / pkg / ggml / sources / llama.cpp / commit
summary | shortlog | log | commit | commitdiff | tree
(parent: b41b4ca)
opencl : support k-quants (#1836)
author0cc4m <redacted>
Fri, 16 Jun 2023 18:59:49 +0000 (20:59 +0200)
committerGitHub <redacted>
Fri, 16 Jun 2023 18:59:49 +0000 (21:59 +0300)
commitd411968e990c37f51328849c96a743dd78f3c3dd
tree96dca9b0f2fa6b1d78c698a3c93e0d9846fd9744tree
parentb41b4cad6f956b5f501db0711dd7007c32b5eee5commit | diff
opencl : support k-quants (#1836)

* Porting q2_k kernel to OpenCL

* Set global and local sizes for kernel calls for dequantizing k-quants

* Added q6_k kernel

* Fix q4_k opencl struct order

* Replace uchar with uint8_t

* Finish dequant kernels

* Added OpenCL DMMV kernels

* Fix q2_k, improve code

* Fix q3_k

* Shorten switch statements

* Improve code formatting

---------

Co-authored-by: Concedo <redacted>
ggml-opencl.cpp diff | blob | history
Packaging of ggml-org/llama.cpp