AMD Launches MI300X AI Chips and Ryzen 8040 CPUs

AMD has launched its highly anticipated MI300X AI chips and new Ryzen 8040 CPUs.
AMD has partnered with Microsoft and Meta to use the MI300X chips in their offerings.
The MI300X is said to have 1.4 times the performance of Llama2 and is already shipping to OEMs.
The MI300X series chips are designed for generative AI workloads and offer industry-leading memory bandwidth.

Advanced Micro Devices (AMD) has announced the general availability of its highly anticipated MI300X AI chips and new Ryzen 8040 CPUs. The MI300X series chips are designed for generative AI workloads and offer industry-leading memory bandwidth. They are being launched alongside the Instinct MI300A accelerated processing unit, which combines AMD CDNA 3 architecture with Zen 4 CPUs for higher performance in AI and high-performance computing.

The MI300X is said to have 1.4 times the performance of Llama2 and is already shipping to OEMs. AMD compared the MI300X to Nvidia's H100, stating that it is on par with the H100 for training and beats it by 10-20% for inference. The MI300A, designed for HPC, is said to have 10-20% better performance than the H100.

The MI300A APU offers higher-performance computing, faster model training, and a 30 times energy efficiency improvement. AMD has partnered with Microsoft to put the MI300X in its Azure virtual machines, and Meta will deploy MI300 processors in its data centers. AMD also introduced the MI300X Platform, combining eight GPUs on a single board, offering the same training performance as Nvidia's H100 HGX.

Despite Nvidia's early investment in AI, AMD is now entering the ring with partnerships with Meta and Microsoft. The MI300-series chips utilize advanced packaging and chiplet technology, with 12 I/O and GPU dies combined into a single package. AMD's new products have already been adopted by Microsoft, Oracle, and the US Department of Energy's Lawrence Livermore National Labs.


Confidence

100%

No Doubts Found At Time Of Publication

Sources

96%

  • Unique Points
    • The MI300X is said to have 1.4 times the performance of Llama2.
    • The article concludes by expressing skepticism about AMD's claim of having the world's fastest AI platform and noting that software optimization will take time.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (100%)
    • The article is straightforward and factual, with no apparent deception.
  • Fallacies (100%)
    None Found At Time Of Publication
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication

96%

  • Unique Points
    • The MI300X series chips are designed for generative AI workloads and offer industry-leading memory bandwidth.
    • The company's partners, including Microsoft, Oracle, and Meta Platforms, showcased their plans to use the MI300X chips in their offerings.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (100%)
    None Found At Time Of Publication
  • Fallacies (100%)
    None Found At Time Of Publication
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication

96%

  • Unique Points
    • The MI300A APU offers higher-performance computing, faster model training, and a 30 times energy efficiency improvement.
    • AMD has partnered with Microsoft to put the MI300X in its Azure virtual machines, and Meta will deploy MI300 processors in its data centers.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (100%)
    None Found At Time Of Publication
  • Fallacies (100%)
    None Found At Time Of Publication
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication

95%

  • Unique Points
    • The MI300-series chips utilise advanced packaging and chiplet technology, with 12 I/O and GPU dies combined into a single package.
    • AMD's new products have already been adopted by Microsoft, Oracle, and the US Department of Energy's Lawrence Livermore National Labs.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (100%)
    • The article is straightforward and factual, with no apparent deception.
  • Fallacies (100%)
    None Found At Time Of Publication
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication

96%

  • Unique Points
    • AMD also introduced the MI300X Platform, combining eight GPUs on a single board, offering the same training performance as Nvidia's H100 HGX.
    • Despite Nvidia's early investment in AI, AMD is now entering the ring with partnerships with Meta and Microsoft.
  • Accuracy
    No Contradictions at Time Of Publication
  • Deception (100%)
    • The article is straightforward and factual, with no apparent deception.
  • Fallacies (100%)
    None Found At Time Of Publication
  • Bias (100%)
    None Found At Time Of Publication
  • Site Conflicts Of Interest (100%)
    None Found At Time Of Publication
  • Author Conflicts Of Interest (100%)
    None Found At Time Of Publication