AMD announces MLPerf Training v6.0 results

Announced MLPerf Training v6.0 results — MI355X within 5% of NVIDIA B200 on Llama 2-70B fine-tuning and within 6% on Llama 3.1-8B pretraining (results submitted by AMD)

Why it matters

Material showing on a standard benchmark that ROCm narrowed the gap to single-digit percentages in 1 year. A real crack in CUDA's monopoly

Source — Primary sources first; secondary reports are labelled “Press”

Officialamd.comhttps://www.amd.com/en/blogs/2026/amd-delivers-breakthrough-mlperf-training-6-0-results.html

Added in Week 35, 2026 · See that week

Same week

John Jumper leaves Google DeepMind for Anthropic after about 9 yearsPressSelective Verification for Budget-Aware Reasoning publishedPaperHassabis says at Stanford GSB that we stand in the foothills of the singularityPressWhat is an AI Agent Harness?Official

Nearby on the same axis

Qualcomm announces Modular acquisition (terms undisclosed, closed 2026-07-29)OfficialModular releases Mojo 1.0, open-sources compiler and tooling under Apache 2.0OfficialResearchers release AInfer-PDPaperScaling Storage for 1 Billion ChatGPT Users (Part I)Official