Coda: Rewriting Transformer Blocks as GEMM-Epilogue Programs
The paper introduces CODA, a new GPU kernel abstraction designed to optimize Transformer block computations. By reparameterizing these computations as GEMM-plus-epilogue programs, CODA aims to reduce memory-bound bottlenecks in training systems. The results indicate that this approach can enhance both productivity and efficiency in machine learning frameworks.
- ▪CODA addresses the inefficiencies caused by memory-bound operators in Transformer training systems.
- ▪The abstraction allows for the execution of computations while keeping GEMM output tiles on chip.
- ▪Both human- and LLM-authored CODA kernels demonstrate high performance across various Transformer workloads.
Hacker News (Newest) files mainly under programming. We currently carry 5,306 of its stories.
Story provenance
Source · retrieval · rights · ranking — open for full record
inspect →
Story provenance
Attribution is not the same as permission. This drawer separates discovery metadata, excerpts, WeSearch-generated summaries, reuse status, and whether the publisher receives the visit. Nothing here claims a legal grant the publisher has not made.
Record
| Original publisher | arXiv.org |
| Canonical URL | https://arxiv.org/abs/2605.19269 |
| Publication time | Fri, 22 May 2026 04:54:33 +0000 |
| Retrieval time | 2026-05-22T05:02:00.539Z |
| Last seen | 2026-05-22T05:02:00.539Z |
| Headline source | Publisher (no WeSearch rewrite) |
| Excerpt source | publisher body |
| Excerpt method | First ~120 words (~800 chars) of extracted publisher body, fair-use limited. |
| Summary | WeSearch · cerebras-chat (WeSearch summarizer) |
| Summary source text | contentText |
| Citation coverage | Summary is a WeSearch-generated derivative; primary citation is the original publisher URL. |
| Cluster | DAVeN18w1urO |
| Cluster logic | Grouped by semantic title/content similarity across sources within a rolling window. Same-publisher template collisions are excluded from coverage comparison. |
| Ranking reason | Story pages are not engagement-ranked. Hub feeds use recency, with optional source-diversified chronological ordering (cap consecutive stories per source). No personalized ranking. |
| Publisher visit | Yes — open original |
| Substitutes article? | No — link-out required for full text |
Rights status (four layers)
WeSearch handling by dimension
| Indexing | May the item be indexed (stored, ranked, made findable)? | Allowed |
| Snippet | May a short excerpt of the publisher's text be shown? | Allowed |
| AI summary | May WeSearch generate its own short summary of the article? | Limited |
| Retrieval / RAG | May the content be exposed for third-party retrieval-augmented generation? | Not asserted |
| Model training | May the content be used to train AI models? | Not asserted |
| Commercial reuse | May the content be reused commercially? | Not permitted |
Basis: Derived from the published RSS/Atom feed. Contact: [email protected]. Reviewed: 2026-07-24.
Opening excerpt (first ~120 words) tap to expand
Computer Science > Machine Learning arXiv:2605.19269 (cs) [Submitted on 19 May 2026 (v1), last revised 20 May 2026 (this version, v2)] Title:CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs Authors:Han Guo, Jack Zhang, Arjun Menon, Driss Guessous, Vijay Thakkar, Yoon Kim, Tri Dao View a PDF of the paper titled CODA: Rewriting Transformer Blocks as GEMM-Epilogue Programs, by Han Guo and 6 other authors View PDF HTML (experimental) Abstract:Transformer training systems are built around dense linear algebra, yet a nontrivial fraction of end-to-end time is spent on surrounding memory-bound operators.
…
Excerpt limited to ~120 words for fair-use compliance. The full article is at arXiv.org.
Discussion
0 commentsMore from arXiv.org
