mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2026-09-04 02:37:27 +02:00
Todd Malsbary
8efbf65dbd
* Add DMMV Q4_K and Q6_K ESIMD kernels Configure cmake build with -DGGML_SYCL_ESIMD=ON to enable. Signed-off-by: Todd Malsbary <[email protected]> * Refactor ESIMD kernels to share common code Signed-off-by: Todd Malsbary <[email protected]> * Move control of ESIMD from compile to runtime Signed-off-by: Todd Malsbary <[email protected]> * Use ESIMD by default when available Signed-off-by: Todd Malsbary <[email protected]> * Fix possible error when using ESIMD by default While not an issue in the current version, this will become an issue when additional QK ESIMD kernels are added (such as Q2_K). Signed-off-by: Todd Malsbary <[email protected]> * Add explicit unroll to ESIMD kernels Signed-off-by: Todd Malsbary <[email protected]> * Tidy up ESIMD kernels a bit Signed-off-by: Todd Malsbary <[email protected]> * Add DMMV Q3_K ESIMD kernel Signed-off-by: Todd Malsbary <[email protected]> --------- Signed-off-by: Todd Malsbary <[email protected]>