llama.cpp/ggml
Jeff Bolz 878b9685de vulkan: Optimize GGML_OP_CUMSUM
There are two paths: The preexisting one that does a whole row per workgroup
in a single shader, and one that splits each row into multiple blocks and does
two passes. The first pass computes partials within a block, the second adds
the block partials to compute the final result. The multipass shader is used
when there are a small number of large rows.

In the whole-row shader, handle multiple elements per invocation.
2026-01-01 14:14:19 -06:00
..
cmake ggml: Skip backend library linking code when GGML_BACKEND_DL=ON (#15094) 2025-08-07 13:45:41 +02:00
include vulkan: extend topk_moe to handle sigmoid w/exp_probs_b for nemotron (#18295) 2026-01-01 08:58:27 +01:00
src vulkan: Optimize GGML_OP_CUMSUM 2026-01-01 14:14:19 -06:00
.gitignore vulkan : cmake integration (#8119) 2024-07-13 18:12:39 +02:00
CMakeLists.txt ggml : bump version to 0.9.5 (ggml/1410) 2025-12-31 18:54:43 +02:00