Intel confirmed the release date of its Sapphire Rapids HBM and Ponte Vecchio processors. The high-performance CPUs and GPUs are expected to be available in January 2023. From then on, the processors will be used by supercomputers in the United States and Japan.

The Sapphire Rapids HBM series is not to be confused with Sapphire Rapids, Intel’s next-gen processor series for servers. Unlike Sapphire Rapids, the Sapphire Rapids HBM series is designed for supercomputers. Intel is launching the series in conjunction with Ponte Vecchio, a series of GPUs for supercomputers.

Together, the models form the Intel Max Series, a new product line for high-performance computing and artificial intelligence. The Sapphire Rapids HBM series consists of CPUs while the Ponte Vecchio series includes GPUs. The powerful processors are extremely expensive. Intel designed the models for supercomputers.

Delays

The release date and first customers were recently announced. Intel expects to deliver the models starting in January 2023. The deadline is likely to be met, as Intel can no longer afford delays.

The Sapphire Rapids series of server processors was repeatedly delayed in recent years. Analysts expect competitor AMD to gain a significant market share as a result. Intel is unlikely to make the same mistake with its supercomputer series.

Supercomputers

Intel announced that the upcoming models have been pre-ordered by several customers. For example, the Ponte Vecchio GPUs will be used in the Aurora supercomputer at Argonne National Laboratory in the United States, while the Sapphire Rapids HBM CPUs are being shipped to Kyoto University in Japan.

Researchers will use the models as engines for future supercomputers and other high-performance systems. A single Sapphire Rapids HBM CPU offers 56 cores. Each core provides 1GB of high-bandwidth memory capacity. “Enough for most common high-performance workloads”, Intel said.

A single Ponte Vecchio GPU offers 128 cores and an L2 cache of 408MB — the highest in the industry, according to Intel. The model is the first high-performance GPU with ray tracing, a modern visualization technology. Ray tracing is used in gaming GPUs to render lifelike graphics. When applied to high-performance GPUs, the technology opens up new possibilities for scientific visualizations and animations.

Tip: Fujitsu launches (super) Computing as a Service