- Radeon AI Pro R9700 targets local workloads and multi-GPU configurations
- The new Workstation class GPU shares its name with a 20 -year -old Ati card
- The new GPU includes 128 AI and 32 GB GDDR6 RAM accelerators
At Computex 2025, AMD announced the Radeon AI Pro R9700, a workstation GPU intended for local AI tasks and multi-GPU calculation environments.
For those who know the history of graphics cards, the name could sound a bell. Over 20 years ago, the original Radeon 9700 Pro marked a turning point for Ati. He was one of the first GPUs to beat Nvidia convincingly both in performance and delivery, and its launch in 2002 helped change market dynamics.
Quick advance until today, and AMD, which acquired at $ 5.4 billion in 2006, reuses the name 9700 for a very different card. AI Pro R9700 is not for players, but for developers and professionals working with large -scale AI models.
Listening for AI
The Radeon AI Pro R9700 has 128 ACA dedicated accelerators, 32 GB of GDDR6 memory and a PCIe Gen 5 interface. Power print is evaluated at 300 W.
AMD says it can reach 96 FP16 performance terraflops and deliver 1531 peaks for AI inference.
Unlike GPUs built for rendering or play, it is set for local inference and training. AMD says it can run models with up to 32 billion parameters without cloud unloading.
In a system with four cards, which evolves up to 123 billion. AI Pro R9700 is optimized for multi-GPU configurations and workloads such as LLM training, simulation and rendering accelerated by AI.
It is shipped with ROCM support on Linux, with the Windows support expected later. Availability is set for July 2025.
While AI Pro R9700 was the release of the title of AMD for the professional workloads of the AI at Computex, the Ryzen Threadripper 9000 series and the GPU RX 9060 XT completed the range with options for creators, enthusiasts and players.
