- Radeon AI Professional R9700 targets native AI workloads and multi-GPU setups
- The brand new workstation-class GPU shares its title with a 20 yr outdated ATI card
- New GPU options 128 AI accelerators and 32GB GDDR6 RAM
At Computex 2025, AMD introduced the Radeon AI Professional R9700, a workstation GPU aimed toward native AI duties and multi-GPU compute environments.
For these acquainted with the historical past of graphics playing cards, the title may ring a bell. Over 20 years in the past, the unique Radeon 9700 Professional marked a turning level for ATI. It was one of many first GPUs to beat Nvidia convincingly in each efficiency and supply, and its launch again in 2002 helped shift market dynamics.
Quick ahead to right now, and AMD, which acquired ATI for $5.4 billion in 2006, is reusing the 9700 title for a really totally different card. The AI Professional R9700 just isn’t for players, however for builders and professionals working with large-scale AI fashions.
Tuned for AI
The Radeon AI Professional R9700 options 128 devoted AI accelerators, 32GB of GDDR6 reminiscence, and a PCIe Gen 5 interface. Energy draw is rated at 300W.
AMD says it could hit 96 teraflops of FP16 efficiency and ship 1531 TOPS for AI inference.
In contrast to GPUs constructed for rendering or gaming, this one is tuned for native inference and coaching. AMD claims it could run fashions with as much as 32 billion parameters with out cloud offload.
In a system with 4 playing cards, that scales as much as 123 billion. The AI Professional R9700 is optimized for multi-GPU configurations and workloads like LLM coaching, simulation, and AI-accelerated rendering.
It ships with ROCm assist on Linux, with Home windows assist anticipated later. Availability is about for July 2025.
Whereas the AI Professional R9700 was AMD’s headline launch for skilled AI workloads at Computex, the Ryzen Threadripper 9000 Collection and RX 9060 XT GPU rounded out the line-up with choices aimed toward creators, fanatics, and players.
You might also like
{content material}
Supply: {feed_title}
