Powered by the Achronix Speedster 7t1500 FPGA, the VectorPath 815 PCIe accelerator card meets the performance demands of AI and HPC workloads. Speedster FPGAs integrate machine learning processors to deliver a massively parallel architecture, customizable data paths, and efficient processing of sparse and irregular computations.
“The VectorPath 815 card delivers greater than 2000 tokens per second with 10-ms inter-token latency (LLAMA 3.1-8B Instruct) for unmatched generative AI inferencing performance — enabling customers to accelerate bandwidth-intensive, low-latency applications with a greater than 3× total cost of ownership (TCO) advantage vs. competitive GPU solutions,” said Jansher Ashraf, director of AI Solutions Business Development at Achronix.
The Speedster 7t1500 FPGA features 2560 machine learning processors, a 2D network-on-chip, 692k LUTs, and 32 SerDes lanes supporting PCIe Gen5 ×16 and dual 400G Ethernet. The VectorPath 815 card builds on this by integrating 32 GB of GDDR6 memory for 4-Tbps bandwidth, 16 GB of DDR4 memory, dual QSFP-DD ports, and a PCIe Gen5 interface.
VectorPath 815 accelerator cards are now in volume production.
The post PCIe card provides FPGA-based data acceleration appeared first on EDN.
