In Japan, the latest professional GPU launched by AMD has been listed, the Instinct MI210graphic whose official price has never been announced until now, it was seen listed in a store at a price of 2,087,800 yen, which in exchange is approximately €15,300 / $16,000.
For that money, the AMD instinct MI210 offers 6656 stream processors under a CDNA2 @ 7nm architecture at a maximum operating frequency of 1700MHz promising peak performance of 22.6 TFLOPsjust under half of Instinct MI250X (47.9 TFLOPs).
Memory has also been halved, providing 64 GB of HBM2e ECC memory at 1.60 GHz which, with a 4096-bit memory interface, achieves a bandwidth of up to 1638.4 GB/s, exactly half of the higher models. All this results in maximum consumption of 300W requiring the use of an 8-pin PCI-Express connector for power. The rest of the information is complemented by peak performance of up to 45.3 TFLOPs (FP64/32 Matrix), up to 181 TFLOPs (FP16/bf16) and up to 181 TOPS (INT8).
AMD today announced the availability of the AMD Instinct ecosystem with expanded system support from partners such as Asus, Dell Technologies, Gigabyte, HPE, Lenovo and Supermicro, the new AMD Instinct MI210 accelerator and robust ROCm software capabilities 5. AMD Instinct and the ROCm ecosystem delivers exascale-class technology to a broad base of HPC and AI customers, meeting the growing demand for compute-accelerated data center workloads and reducing analysis time and of discovery.
“With twice as many platforms available compared to our previous generation of accelerators, growing customer adoption of HPC and AI applications, and new support for commercial ISVs on key workloads, we continue to drive the adoption of AMD Instinct MI200 accelerators and the ROCm 5 software ecosystem. Now, with the availability of the AMD Instinct MI210 accelerator in the MI200 family, our customers can choose the accelerator that best suits their workloads , whether they need state-of-the-art accelerated computing for large-scale HPC and AI workloads, for example if they want to access exascale-class technology in a commercial format,” said Brad McCredie, vice – President of AMD’s Data Center and Accelerated Computing GPU Division.
“The Lumi supercomputer powered by AMD EPYC processors and AMD Instinct MI200 accelerators will deliver a generational leap in performance for large-scale modeling and simulations, as well as deep learning and AI workloads , to solve some of the biggest research questions,” said Pekka Manninen, director of the LUMI Computing and Leadership Facility, CSC.
“We used AMD Instinct MI210 accelerators to gain hands-on experience with the Instinct MI200 family, preparing our scientists to tackle the many challenging and complex projects they will run once Lumi is fully deployed.”