Supermicro this week launched the listing of the industry’s first commercial servers based on Intel’s Data Center GPU Max ‘Ponte Vecchio’ compute GPUs. The machine uses Ponte Vecchio in add-in board and OAM module form factors for high-performance computing and large-scale AI training.
Supermicro is currently Two servers certified with Intel’s Ponte Vecchio computing GPUs: SYS-421GE-TNRT Machines that can accommodate up to 10 Data Center GPU Max 1100 cards with 480 GB of HBM2 memory (48 GB per board) and up to 8 Data Center GPU Max 1550 OAM modules with 1 TB of HBM2 memory It has a capacity SYS-821PV-TNR onboard (128 GB per card) and combines performance of 6.7 BF16/FP16 PetaFLOPS at 4.8 kW.
Both machines are based on Intel’s 2 of 4.th Generation Xeon scalable ‘Sapphire Rapids’ processors can be combined with up to 8TB of DDR5 memory using 32 256GB modules. In terms of storage, both machines have 24x 2.5″ hot-swap bays (8x 2.5″ NVMe hybrid, 8x 2.5″ NVMe dedicated) for U.2/SATA/SAS drives and the SYS-421GE- TNRT also has two bays. M.2 slot for PCIe drive.
For now, Supermicro sells the SYS-421GE-TNRT with Nvidia’s A100 80GB or H100 80GB board, but it looks like Intel’s Data Center GPU Max 1100 AIB can be installed instead if requested. You should install them and not spend too much time installing the required software.
Meanwhile, the SYS-821PV-TNR, which can house up to eight Data Center GPU Max 1550 OAM modules, is listed as “coming soon.”
Supermicro is currently the only supplier of commercial machines for AI and HPC workloads offering systems powered by Intel’s Data Center GPU Max cards and modules. However, it’s reasonable to expect other major server suppliers to start selling similar products soon.
Supermicro offers Intel Ponte Vecchio-based servers, but doesn’t disclose pricing because that’s always the case with individually configured, mass-sold AI and HPC machines.
sauce: super micro (@squash bionic)
