AiHPC - MX

Powerful solution for VU37P FPGA with HBM to accelerate application scenarios that are subject to storage latency

For computing acceleration, is a PCle board, using Xilinx VU37P FPGA, integrated HBM2 memory. HBM2's size and speed (8G/16GB, up to 512GB/s) can accelerate applications that are limited by memory. The board's 100G QSFP28s is ideal for clustering, and the OCuLink connector allows for expansion.

Back

Case:

High performance computing

It is used to accelerate application requirements that are constrained by memory read and write speed

Product parameters:

Xilinx VU37P FPGA

8G/16G HBM2 Up to 512GB/S storage throughput

Up to 4x100GbE, support RP Slim SAS (SFF-8654) *2,

Link to external FPGA through 16xGTY

Up to 256GBytesDDR4 SDRAM

2.8 million LCs