Titaniumtown 4257f47019 sycl: fix thread/block count in quantized cpy kernel launches (llama/27160) ...

Adjusts the thread/block count to be proportional to the size
of the quant, reducing under/over subscription.

Largest perf improvement is the q4_0 -> f32 path, with, on
a Arc 70, throughput goes from 20.21 GB/s to 158.19 GB/s

The rest of the quants are flat in performance uplift.

2026-08-18 15:33:21 +03:00

..

cmake

cmake : add config version support (ggml/1582)

2026-08-14 22:16:06 +03:00

include

ggml : recurrent state rollback for ggml_ssm_scan (llama/26623)

2026-08-14 22:16:06 +03:00

src

sycl: fix thread/block count in quantized cpy kernel launches (llama/27160)

2026-08-18 15:33:21 +03:00

.gitignore

…

CMakeLists.txt

ggml : bump version to 0.20.0 (ggml/1584)

2026-08-14 22:16:06 +03:00