Commit Graph
5 Commits
Author SHA1 Message Date
Titaniumtown 6c8dcaa7ae sycl: parallelize the non-contiguous concat kernel (#25852)
* sycl: parallelize the non-contiguous concat kernel

Launch geometry only: the non-contiguous concat kernel launched a single-lane
work-group (1, 1, 1), now it will launch a (1, 1, SYCL_CONCAT_BLOCK_SIZE) one.

SYCL_CONCAT_BLOCK_SIZE is defined in `ggml/src/ggml-sycl/presets.hpp`.

llama-bench (Arc Pro B70, Qwen3.6-27B-UD-Q4_K_XL, -fa on, q8_0 KV),
on top of upstream master: pp2048 920 -> 1006 t/s (+9.4%)

* sycl: cap non-contiguous concat block at ne0

* sycl: make non-contiguous concat block width env-tunable (GGML_SYCL_CONCAT_BLOCK_SIZE)

* Revert "sycl: make non-contiguous concat block width env-tunable (GGML_SYCL_CONCAT_BLOCK_SIZE)"

This reverts commit 2709909e790d298cc0746950caff16ad8ecf9d6f.
2026-08-04 10:08:05 +08:00
Titaniumtown 1553725965 sycl: fuse RMS_NORM + MUL (#26015) 2026-07-31 09:17:53 +03:00
Titaniumtown 11b068d066 sycl: contiguous fast path + 32-bit index math for unary elementwise ops (#25946)
* sycl: contiguous fast path + 32-bit index math for unary elementwise ops

* sycl: use fastdiv for elementwise index math
2026-07-29 15:16:57 +03:00
Titaniumtown 8e8681e0e2 sycl(build): parallelize ocloc invocations (#25903) 2026-07-27 15:33:11 +03:00
Titaniumtown c1063ac9d7 sycl: set fattn_vec_nthreads to 256 for Battlemage (#25205)
Currently detects lunarlake + battlemage / xe2 and
sets the value to 256.

Keeps default at 128, Intel's ARC Alchemist's prefered value.
2026-07-14 12:00:00 +03:00