Unveiling the Arm-basedGrace CPUNVIDIA also emphasized expanding its collaboration with the Arm architecture to larger-scale computing applications, including cloud applications in collaboration with AWS, and high-performance computing in collaboration with Oak Ridge National Laboratory, Los Alamos National Laboratory, Stony Brook University, Taiwan National High-Speed Network and Computing Center, and the Korea Institute of Science and Technology. NVIDIA also collaborated with Marvell on edge computing applications and with MediaTek to create a PC reference platform design that combines Arm CPUs and NVIDIA GPUs and is compatible with Chromium, Linux, and the NVIDIA SDK.
Among them, through previous cooperation with Arm, Arm CPUs can be compatible with CUDA and CUDA-X libraries, so that Arm architecture CPUs can be used with NVIDIA GPUs. Therefore, the cooperation with AWS will add NVIDIA GPUs to Amazon EC2 servers based on the Graviton2 CPU built on the Arm architecture. At the same time, the Arm architecture CPUs built by Ampere Computing can also be used with NVIDIA GPUs to achieve higher computing performance.
The NVIDIA Arm HPC Developer Kit, designed for scientific computing needs, includes an Ampere Altra CPU with 80 Arm Neoverse architecture cores, paired with two NVIDIA A100 GPUs and two BlueField-2 DPUs, creating an Arm-based computing system accelerated by NVIDIA GPUs. Currently, Oak Ridge National Laboratory, Los Alamos National Laboratory, Stony Brook University, Taiwan National High-Speed Network and Computing Center, and the Korea Advanced Institute of Science and Technology have all adopted this developer kit, which will be officially launched in the third quarter of this year.
In edge computing, NVIDIA is collaborating with Marvell to combine Marvell's OCTEON DPU with NVIDIA GPUs to accelerate cloud-to-edge computing and ensure data security. Their collaboration with MediaTek is bringing NVIDIA GPUs to PCs powered by Arm-based CPUs, compatible with Chromium and Linux operating systems, as well as NVIDIA SDK tools. This allows Arm-based PCs to utilize NVIDIA GPUs for accelerated computing.
Other collaborative applications include cooperation with Fujitsu and SiPearl in the supercomputing field. In addition, it is expected that more Arm-based PCs using NVIDIA GPUs will be launched in the future.
