Intel Max Series Brings Breakthrough Memory Bandwidth and Performance to HPC and AI
The new product family offers a processor with high-bandwidth memory, 4.8 times faster than the competition, and Intel’s highest-density graphics processor, to solve the world’s toughest challenges.
What’s New: Ahead of Supercomputing ’22 in Dallas, Intel Corporation introduced the Intel Max Series product family with two cutting-edge products for high-performance computing (HPC) and artificial intelligence (AI): Intel® Xeon® CPU Max Series (codenamed Sapphire Rapids HBM) and Intel® Data Center GPU Max Series (codenamed Ponte Vecchio). The new products will power Argonne National Laboratory’s upcoming Aurora supercomputer, with updates on its deployment shared today.
The Xeon Max processor is the first and only x86-based processor with high-bandwidth memory, accelerating many HPC workloads without the need to modify code. The Max Series GPU is Intel’s highest-density processor, packing more than 100 billion transistors into a 47-tile package with up to 128 gigabytes (GB) of high-bandwidth memory. The oneAPI open software ecosystem provides a single programming environment for the two new processors. Intel’s oneAPI and AI 2023 tools will provide functionality to enable advanced features of Intel Max Series products.
“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. The Intel Max Series product family brings high-bandwidth memory to a wider market, along with a single API, making it easier to share code between CPUs and GPUs and solve the world’s biggest challenges faster .
–Jeff McVeigh, vice president and general manager, Super Compute Group at Intel
Why is this important: High-performance computing (HPC) represents the forefront of technology, using the most advanced innovations at scale to solve the greatest challenges of science and society, from mitigating the impacts of climate change to curing the world’s deadliest diseases.
Max Series products meet the needs of this community with scalable and balanced processors and GPUs, incorporating breakthroughs in memory bandwidth and united by oneAPI, an open, standards-based, cross-architecture programming framework . Researchers and businesses will solve problems faster and more sustainably using Max Series products.
When he arrives : Max-series products are expected to launch in January 2023. In line with its commitments to customers, Intel is shipping blades with Max-series GPUs to Argonne National Laboratory to power the Aurora supercomputer and will supply Xeon Max processors to the Los Alamos National Laboratory, Kyoto University and other supercomputing sites.
What the Intel Xeon Max processor offers: The Xeon Max processor offers up to 56 performance cores made up of four tiles and connected using Intel’s multi-die interconnect bridge (EMIB) technology, in a 350-watt envelope. The Xeon Max processors pack 64GB of high-bandwidth onboard memory, along with PCI Express 5.0 and CXL1.1 I/O. Xeon Max processors will provide over 1 GB of High Bandwidth Memory (HBM) capacity per core, enough to accommodate the most common HPC workloads. The Max Series processor delivers up to 4.8 times better performance than the competition on real HPC workloads.1
- 68% less power consumption than an AMD Milan-X cluster for the same HPCG performance.
- AMX extensions improve AI performance and provide 8x peak throughput over AVX-512 for INT8 with INT32 accumulation operations.2
- Provides the flexibility to operate in different HBM and DDR memory configurations.
- Workload benchmarks:
- Climate modeling: 2.4 times faster than AMD Milan-X on MPAS-A using only HBM.
- Molecular Dynamics: On DeepPMD, 2.8x performance improvement over competing products with DDR memory.
What the Intel Max Series GPU offers: Max-series GPUs offer up to 128Xe-HPC cores, the new foundational architecture for the most demanding computing workloads. Additionally, the Max Series GPU features:
- 408 MB of L2 cache – the highest in the industry – and 64 MB of L1 cache to increase throughput and performance.
- The only HPC/AI GPU with native ray-tracing acceleration, designed to accelerate scientific visualization and animation.
- Workload benchmarks:
- Finance: 2.4x performance gain compared to the NVIDIA A100 on the pricing of the Riskfuel credit option.
- Physics: 1.5x improvement over A100 for NekRS virtual reactor simulations.
Max-series GPUs will be available in multiple form factors to meet different customer needs:
- Max Series 1100 GPU: A double-wide 300-watt PCIe card with 56 Xe cores and 48 GB of HBM2e memory. Multiple boards can be connected through Intel Xe Link bridges.
- GPU Max Series 1350: a 450 watt OAM module with 112 Xe cores and 96 GB of HBM.
- GPU Max Series 1550: 600 Watt OAM module at Intel’s maximum performance with 128 Xe cores and 128 GB of HBM.
Beyond individual boards and modules, Intel will offer the Intel Data Center GPU Max Series subsystem with OAM GPU x4 carrier board and Intel Xe Link to enable high-performance multi-GPU communication within the subsystem.
What Max Series products allow: In 2023, the Aurora supercomputer, currently under construction at Argonne National Laboratory, is expected to become the first supercomputer to exceed 2 exaflops of peak computing performance in double precision3. Aurora will also be the first to showcase the power of combining Max-series GPUs and CPUs in a single system, with more than 10,000 blades, each containing six Max-series GPUs and two Xeon Max CPUs.
Ahead of SC22, 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.
Max Series products will power several other HPC systems critical to national security and basic research, including Crossroads at Los Alamos National Laboratory, CTS-2 systems at Lawrence Livermore National Laboratory and Sandia National Laboratory, and Camphor3 at Kyoto University.
And after: At Supercomputing ’22, Intel and its customers will showcase more than 40 upcoming system designs from 12 original equipment manufacturers using Max Series products. Attendees can explore demos showcasing the performance and capabilities of Max Series products for a range of AI and HPC applications, as well as hear from Intel architects, customers, and end users about the power of the solutions Platform Manager at the Intel booth, #2428. More information on Intel’s activities at SC22 is available.
The Intel Data Center Max Series GPU, named Rialto Bridge, is the successor to the Max Series GPU and is expected to arrive in 2024 with improved performance and a seamless upgrade path. Intel then plans to release the next major architectural innovation to enable the future of HPC. The company’s next XPU, named Falcon Shores, will combine Xe and x86 cores in a single package. This revolutionary new architecture will also have the flexibility to integrate new IP addresses from Intel and customers, fabricated using our IDM 2.0 model.
“end of press release”
#Intel #introduces #MAXseries #processors #GPUs #Sapphire #RapidsHBM #Ponte #Vecchio #VideoCardz.com