The Aurora supercomputer is expected to go live in 2023. Image: Argonne National Laboratory
Intel officially introduced the Intel Max Series product family for high-performance computing (HPC) and artificial intelligence (AI) on Wednesday. Consisting of the Max series of Intel Xeon processors (codenamed Sapphire Rapids HBM) and the Max series of data center GPUs (codenamed Ponte Vecchio), these new products will power the future Aurora supercomputer at Argonne National Laboratory .
When the Aurora supercomputer comes online in 2023, it is expected to be the first supercomputer to surpass two exaflops of double-precision computing performance. It will also be the first system to demonstrate the power of Max-series GPUs and CPUs together. The supercomputer will house more than 10,000 blades, each containing six Max-series GPUs and two Xeon Max CPUs.
Although the Max series makes a sensational debut at the Argonne National Laboratory in the United States, the product family is truly designed to bring supercomputing and artificial intelligence to a broader market, Intel said.
“We’re trying to help solve the world’s biggest challenges faster and to do so in a sustainable way,” Jeff McVeigh, Intel vice president and general manager of the Super Compute Group, told reporters ahead of the announcement. .
Accelerate problem solving
Problems like climate change and global health crises can be solved faster with the help of supercomputing and AI. There is a “great diversity” of workloads in supercomputing and AI, he added. Some are compute-heavy, while others have high memory bandwidth constraints.
“Our view is that all of these things are important and we need to have the right solutions to deal with them,” he said.
CPUs and GPUs are united by oneAPI, Intel’s open and standards-based multi-architecture programming framework. Researchers and businesses will solve problems faster and more sustainably with Max Series products.
Xeon Max processor
The main advantages of the Xeon Max processor, the first and only x86 processor with high bandwidth memory, are:
- that it offers up to 56 performance cores made up of four tiles;
- contains 64GB of high-bandwidth onboard memory;
- provides over 1 GB of high-bandwidth memory (HBM) capacity per core, which is sufficient for most common HPC workloads;
- Delivers up to 4.8 times better performance than the competition on real HPC workloads.
- 68% less power consumption than an AMD Milan-X cluster for the same HCPG performance
Max-series GPUs
The Max-series GPU, meanwhile, is Intel’s highest-density processor, with over 100 billion transistors in a 47-tile package. Among its properties:
- it offers up to 128 gigabytes (GB) of high-bandwidth memory;
- provides up to 128 HPC Xe cores;
- includes 408 MB L2 cache and 64 MB L1 cache for increased throughput and performance;
- offers native ray tracing acceleration, for scientific visualization and animation.
In addition to the Aurora supercomputer, Intel will supply Xeon Max processors to Los Alamos National Laboratory, Kyoto University and other supercomputer sites.
Meanwhile, Argonne and Intel unveiled Sunspot, Aurora’s test development system, comprised of 128 production blades. Aurora Early Science Program researchers will have access to the system beginning in late 2022.
Source: ZDNet.com