[CLSTR]
< Back to situations

Monitor this situation.

[SITUATION] · [QUIET] · [TECHNOLOGY]

2 clusters · 5 sources · 19 days · First seen · Last updated

AMD and Azure AI performance and software developments

Overview

Technology providers Azure and AMD have reported advancements in AI hardware and software performance. Initial reports from MLPerf Inference 6.1 benchmarks showed Azure achieving high interactive scores and throughput on NVIDIA GB300 and GB200 platforms. Concurrently, AMD demonstrated significant throughput improvements using its ROCm software on MI355X hardware, outperforming certain NVIDIA submissions on specific models.

Following these performance results, AMD released ROCm 10 and made ROCm.AI generally available. This update focuses on easing the migration of AI applications from NVIDIA’s CUDA platform to AMD’s ecosystem. The release includes tools like HIPIFY for code translation, a version of PyTorch that maintains the torch.cuda interface, and various optimization and monitoring tools such as Primus and Hyperloom.

Entities

Nvidia · AMD · DeepSeek · Microsoft Azure · MLCommons

Timeline

  1. [TECHNOLOGY] 2 sources
    AMD launches ROCm 10 to ease AI software migration from CUDA

    AMD has launched ROCm 10 and ROCm.AI, offering new tools to simplify migrating AI workloads and code from NVIDIA’s CUDA platform to AMD’s open software ecosystem.

  2. [TECHNOLOGY] 3 sources
    Azure and AMD report new AI performance results in MLPerf Inference 6.1

    Azure and AMD have released MLPerf Inference 6.1 results, showcasing high-scale AI performance for NVIDIA-based Azure platforms and significant software-driven throughput gains for AMD’s ROCm on MI355X hardware

Sources

365community.online · bug.hr · futurumresearch.com · pcekspert.com · quantumzeitgeist.com