AMD has announced the availability of its latest generation of processors and graphics for professional work machines, Ryzen Threadripper 9000 and Radeon AI Pro r9700. Will be available in the international market As of July 23.
Ryzen Threadripper 9000
The latest AMD processing platform was presented at the Computex 2025 with the goal of Motorizar High performance desktop computers (HEDT) and professional work stations. Developed under the code name ‘Shimada Peak’, the platform stands out for the premiere of Zen5 architecture in the most powerful personal computers group on the market, since we talk about those destined for 3D rendering, video edition or scientific simulations, as well as the most advanced consumption client machines and mixed use, enthusiastic degree.
AMD has developed it as a variant of the MCM EPYC ‘Turin’. It is designed for the TR5 socket And it offers up to 128 PCIE Gen 5 lines, but its I/O has been slightly modified on server processors, lacking the CXL capacity in these lines and with the memory configured in 8 -channel DDR5 (16 sub -channels), compared to the 12 DDR5 channels of the “Turin” (24 sub -esteems).
The Californian company sells the series in Threadripper 9000 versions and Threadripper 9000 Prothe most powerful, with variants of 96, 64, 32 and 16 cores, all with the same consumption: 350W TDP. The range of range is the Ryzen Threadripper Pro 9995Wx. It includes a 96 -core and 192 wire settings. It is composed of 12 CCD, each with eight nuclei and 32 MB of third level cache L3. It has an I/O PCIE 5 of 128 lines and an 8 -channel DDR5 memory interface.
They will be available from July 23 on the specialized channel and in work stations of AMD OEM partners such as Lenovo, Dell, HP, Supermicro and others.
Radeon AI PRO R9700
Another of AMD’s great novelties in Computex was a professional graph that the firm focuses as the ideal complement to the CPUS Ryzen Threadripper 9000 and to cover a wide range of uses, from professional visualization to acceleration of AI in the Edge, through the local inference of AI, the fine adjustment of complex creative work models and charges.
The card is an improved version of RADEON RX 9070 XT of consumption. Based on architecture RDNA 4take full advantage of the “Navi 48” silicon of 4 Nm, allowing 64 computer units for a total of 4,096 shaders. It has 32 GBYTES OF DEDICATED MEMORY GDDR6 at 20 Gbps (twice the variant for consumption) in an 256 -bit chip bits interface, which provides a bandwidth of 640 GB/s. The card has a total power of 300 W.
Has 128 AI accelerators second generationwhich offer up to twice as much performance as the previous generation and total compatibility with the Rocm software stack in Linux and next support of its equivalent in Windows. With these specifications, AMD ensures that Radeon AI PRO R9700 reaches up to 96 TFLOP of performance FP16 and up to 1531 tops of AI (int4 dispersed).
The Radeon AI Pro r9700 also scale effectively in Multi GPU settingsexpanding memory and computing capacity for large models and parallel tasks. This flexibility makes it ideal for advanced work stations that manage complex simulations, rendered in real time or simultaneous work loads. It will also be available from July 23.
More information:
