Intel finally unveiled its long-awaited Sapphire Rapids update to the Xeon family of processors, along with the company's new data center GPU.
Both are now available as Max Series and will soon be installed on major supercomputers, such as the Aurora at Argonne National Laboratory.
"To ensure that no HPC workload is left behind, we need a solution that maximizes bandwidth, maximizes compute, maximizes developer productivity, and ultimately maximizes impact," said Jeff McVeigh, vice president of Super Intel Computing Group.
Intel Xeon Max series
In a press release (opens in a new tab) on the company's website, Intel explains that these supercomputers play crucial roles in some of the world's biggest scientific and societal challenges, "from mitigating impacts to curing the deadliest diseases in the world.
As such, the latest chips have gotten a big boost across the board. The Max-series GPU is the company's highest-density processor, now offering up to 128 GB of high-bandwidth memory in the form of more than 100 billion transistors in a 000-core, 128-tile package.
The latest GPUs will be available in three forms: Max Series 1100 GPUs, 1350 GPUs, and 1550 GPUs, each with 48GB, 96GB, and 128GB of memory respectively.
In the same vein, the 350W Xeon Max processor packs 64GB of high-bandwidth memory in its quad-tile configuration with up to 56 performance cores based on the same Golden Cove microarchitecture features found in the 12th generation Intel Core processors.
The new hardware will support DDR5, PCIe 5.0, and Compute Express Link (CXL) 1.1, allowing memory to connect directly to the CPU via PCIe 5.0.
McVeigh acknowledged the delays the company's new supercomputing components have faced, noting that "we are always going to push the limits." Sometimes that can keep us from not making it, but we do it to help our developers, to help the ecosystem, to help solve big challenges.”
Intel now plans production of the Xeon CPU and Data Center GPU Max series in early 2023, when it begins deliveries to Argonne, Los Alamos National Laboratories, Kyoto University, and other supercomputing sites.