AMD showcases Instinct accelerator family at Computex 2024

AMD today showcased the AMD Instinct accelerator family at Computex 2024, extending high-performance leadership in data center business.
AMD Instinct MI325X acceleratorMicrosoft Azure is using the accelerators for Azure OpenAI services and the new Azure ND MI300X V5 virtual machines.

Dell Technologies is using MI300X accelerators in the PowerEdge XE9680 for enterprise AI workloads.

Supermicro is providing multiple solutions with AMD Instinct accelerators.

Lenovo is powering Hybrid AI innovation with the ThinkSystem SR685a V3

HPE is using them to accelerate AI workloads in the HPE Cray XD675.

AMD’s Data Center segment revenue rose 80 percent year-over-year and 2 percent quarter-on-quarter to $2.3 billion driven by growth in both AMD Instinct GPUs and 4th Gen AMD EPYC CPUs.

AMD Instinct MI325X

AMD Instinct MI325X accelerators will have 288GB of HBM3E memory that extends AMD generative AI performance leadership.

AMD Instinct MI325X accelerator will be available in Q4 2024. AMD Instinct MI350 series, powered by the new AMD CDNA 4 architecture, is expected to be available in 2025 bringing up to a 35x increase in AI inference performance compared to AMD Instinct MI300 Series with AMD CDNA 3 architecture. AMD Instinct MI400 series, expected to arrive in 2026, is based on the AMD CDNA Next architecture.

“The AMD Instinct MI300X accelerators continue their strong adoption from numerous partners and customers including Microsoft Azure, Meta, Dell Technologies, HPE, Lenovo and others, a direct result of the AMD Instinct MI300X accelerator exceptional performance and value proposition,” said Brad McCredie, corporate vice president, Data Center Accelerated Compute, AMD.

AMD Instinct MI300X accelerators, powered by AMD ROCm 6 open software stack, enables customers to get 1.3x better inference performance and token generation, on a server using eight AMD Instinct MI300X accelerators and ROCm 6 running Meta Llama-3 70B.

On a single AMD Instinct MI300X accelerator with ROCm 6, customers can get more inference performance and token generation throughput by 1.2x on Mistral-7B3.

AMD also highlighted that Hugging Face, the largest and most popular repository for AI models, is now testing 700,000 of their most popular models nightly to ensure they work out of box on AMD Instinct MI300X accelerators. In addition, AMD is continuing its upstream work into popular AI frameworks like PyTorch, TensorFlow and JAX.

The new AMD Instinct MI325X accelerator, which will bring 288GB of HBM3E memory and 6 terabytes per second of memory bandwidth, use the same industry standard Universal Baseboard server design used by the AMD Instinct MI300 series, and be generally available in Q4 2024. The accelerator will have 2x and 1.3x better in terms of memory capacity and bandwidth, respectively, and 1.3x better compute performance than competition.

AMD Instinct MI350X accelerator, based on AMD CDNA 4 architecture and is expected to be available in 2025, will have up to 288 GB of HBM3E memory.

Baburajan Kizhakedath

Related News

Latest News

Latest News