Arm says its AGI CPU delivers nearly 2× performance per rack of AMD EPYC 9755
Arm announced that its AGI CPU, built for agentic AI workloads, shows a modeled 1.99x rack‑level performance advantage over AMD’s EPYC 9755 when evaluated across seven workloads. The claim comes from Arm‑commissioned analysis by Signal65 using a new evaluation methodology developed with Fujitsu, Meta and OpenAI.
Key points
- Arm’s AGI CPU shows a modeled 1.99x rack‑level performance advantage over AMD EPYC 9755 across seven workloads.
- The chip beat AMD in five of seven tests, leading by 9% on Terminal‑Bench, 13% on SWE‑bench and 97% on fastText.
- Arm’s methodology models server power at 950 W (128‑core) and 1,050 W (136‑core) versus 1,700 W for AMD, fitting 23 or 21 Arm servers into a 24 kW rack.
The tests used a 128‑core, 2.8 GHz Arm chip with DDR5‑7200 and a 136‑core, 3.2 GHz chip with DDR5‑8000, compared to a single‑socket 128‑core EPYC 9755 with DDR5‑6400. Arm’s CPU beat AMD in five of the seven tests, leading by 9% on Terminal‑Bench, 13% on SWE‑bench and 97% on fastText, while lagging 12% and 5% on MySQL and Milvus respectively. Across the suite the chip showed a 20% (1.20x) per‑chip advantage and, after accounting for server‑power assumptions of 950 W, 1,050 W and 1,700 W, the model fits 23 or 21 Arm servers versus 13 AMD servers in a 24 kW rack, yielding the 1.99x rack result. Signal65 reproduced the seven results within 0.5% but noted some configuration details could not be independently verified.
Arm says its AGI CPU delivers nearly 2X the performance per rack of AMD EPYC 9755 in new agentic AI CPU tests
Tom's Hardware · 19 October 2026
Loading the full article…
This text was published by Tom's Hardware and written by Etiido Uko. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
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 →- Tenstorrent raises $1.5B Series E at $6.2B valuation · 1 src
- Alibaba launches Qwen Glasses N1 smart eyewear starting at $450 · 1 src
- Apple CEO John Ternus calls memory chip shortage a crisis · 1 src
- China's economic growth picks up to 4.5% as AI boom boosts exports · 2 src
- Nvidia announces RTX Spark laptop superchip and Vera Rubin NVL72 rack system · 2 src
Comments
via GitHub Discussions