Ggml-org merges Vulkan kernel fusion for Qwen4exp's SCALE‑sigmoid‑SCALE chain
A pull request (29520) was merged into the ggml‑org/llama.cpp repository on Sep 28 2026, adding a fused Vulkan kernel that combines the SCALE‑SIGMOID‑SCALE‑hcpost operations used by the Qwen4exp model. The change was approved by reviewers Jeffbolznv and 0cc4m and automatically tested by the github‑actions Bot. By collapsing the four separate tensor operations into a single GPU kernel, the update…
Key points
- PR #29520 merged on Sep 28 2026 adds a fused Vulkan kernel for Qwen4exp's SCALE‑SIGMOID‑SCALE‑hcpost chain.
- Reviewers Jeffbolznv and 0cc4m approved the change; github‑actions Bot ran automated tests.
- Fusion reduces memory moves and should lower inference latency on Vulkan‑compatible GPUs.
The modification is part of ongoing efforts to optimise ggml’s Vulkan backend, which underpins many open‑source LLM inference projects. Faster execution of the Qwen4exp chain can benefit developers deploying the model on a range of hardware, from desktop GPUs to mobile devices that support Vulkan. The merge does not introduce new model weights or change model behavior, but it streamlines the compute path for existing deployments.
The story so far
2 episodes →- Ggml-org merges Vulkan kernel fusion for Qwen4exp's SCALE‑sigmoid‑SCALE chainthis story
vulkan: fuse qwen4exp's SCALE -> SIGMOID -> SCALE -> hc_post chain by fxgsell · Pull Request #29520 · ggml-org/llama.cpp
github.com · 28 September 2026Loading the full article…
This text was published by github.com. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
Coverage and discussion
1source- Reddit discussionreddit.com
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Hardware & Compute
All →- Applied Materials and Intel expand AI chipmaking collaboration at epic Center · 1 src
- Xeal to deploy 100,000 Nvidia GPUs in EV charging pods across US · 1 src
- Meta’s Muse boosts AMD stock as AI agents shift workload to CPUs · 1 src
- Nokia CEO says data center construction could double without supply constraints · 1 src
- Ghost launches $3,499 AI agent computer with $11M seed · 1 src
Comments
via GitHub Discussions