Piotr WilkinandClaude Opus 4.8 75e77503b2 vulkan: remove the all-to-all pipeline and fp32 staging from the comm ...

With the ring as the default there is no reason to keep the slower paths:
- Remove ggml_backend_vk_comm_allreduce_pipeline (the O(n^2) all-to-all) and
  GGML_VK_COMM_PIPELINE. The ring is now the unconditional large-tensor path;
  the comm->ring flag and pipe_round are gone, and pipeline_ok (the "has two
  queues" gate the ring needs) is renamed ring_ok.
- Remove fp32 staging and GGML_VK_COMM_FP32. The ring always stages F16 (fp32
  accumulator preserved); its fp32 branch and use_f16 are removed.

Net ~-260 lines. Verified byte-identical greedy output (ring / proxy) and clean
build on 4x A16. The decode single-shot and the meta-backend butterfly fallback
are untouched.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_01ApKCQ32VLqUW4Kus6tUvBL

2026-06-28 21:06:05 +02:00

..

cmake

ggml : Parallelize quant LUT init (#23595)

2026-05-25 10:15:46 +03:00

include

sycl : support --split-mode tensor (#24152)

2026-06-25 08:35:21 +03:00

src

vulkan: remove the all-to-all pipeline and fp32 staging from the comm

2026-06-28 21:06:05 +02:00

.gitignore

vulkan : cmake integration (#8119)

2024-07-13 18:12:39 +02:00

CMakeLists.txt

ggml : bump version to 0.15.3 (ggml/1550)

2026-06-26 15:04:42 +03:00