AMD outlined its AI GPU roadmap and moved to an annual cadence as it aims to compete with Nvidia.
A day after Nvidia outlined its GPU roadmap at Computex, AMD CEO Lisa Su outlined an annual cadence. The company also outlined new EPYC CPUs along with processors for AI PCs.
Su said the new AMD Instinct MI325X accelerator will be available in the fourth quarter with up to 288GB of HBM3E memory and 6 terabytes per second of memory bandwidth with new AMD Instinct MI350 accelerators based on AMD's CDNA 4 architecture available in 2025.
- AMD sees AI inference, training workloads increasing in enterprise | AMD Q1 delivers data center, AI sales surge of 80%
According to AMD, the AMD Instinct MI325X accelerator will bring a 35x increase in AI inference performance compared to the AMD Instinct MI300.
In 2026, the AMD Instinct MI400 series will be based on AMD CDNA "Next" architecture.
AMD's Instinct MI300X accelerators are being rolled out by Microsoft Azure, Meta and major server makers such as Dell Technologies, HPE and Lenovo.
Su also touted AMD's software ecosystem, which is playing catchup to Nvidia. AMD said its AMD ROCm 6 open software stack is developing its connections to popular AI frameworks.
Other AMD items of note at Computex include:
- AMD previewed 5th Gen AMD EPYC server processors to launch in the second half of 2024. These processors, codenamed Turin, are designed to continue AMD's server momentum with CPUs.
- AMD outlined the AMD Ryzen AI 300 series, the third generation of AMD AI-enabled mobile processors. These chips will be used in AI PCs and Copilot+ PCs, which are now powered by Qualcomm.
- AMD highlighted its AMD Ryzen 9000 Series processors for laptops and desktops.
- The company also highlighted its next-gen Zen 5 CPU core and AMD XDNA 2 NPU core.
- AMD also announced the AMD Radeon PRO W7900 Dual Slot workstation graphics card.
More on genAI:
- The real reason Windows AI PCs will be interesting
- Copilot, genAI agent implementations are about to get complicated
- Generative AI spending will move beyond the IT budget
- Enterprises Must Now Cultivate a Capable and Diverse AI Model Garden
- Secrets to a Successful AI Strategy
- Return on Transformation Investments (RTI)
- Financial services firms see genAI use cases leading to efficiency boom
- Foundation model debate: Choices, small vs. large, commoditization