ACAP is the brainchild of Xilinx CEO Victor Peng (pictured) and, says Peng, “accelerates a broad range of workloads with dynamically adaptable silicon which is 10-100 times faster than CPU, with more use cases than GPUs and ASICs and delivers more than 10x performance and performance per Watt.”
ACAP can adapt to meet the requirements of different workloads so reducing the electricity bill for data centres.
Peng, a GPU designer, says ““GPUs are OK for heavy lifting because they don’t care about latency – that’s why they can sell top-end GPUs using 300W,” says Peng, “but burning several hundred Watts is not a good thing for a datacentre.”
ACAP was developed in a project, called Everest, which cost a billion dollars over four years. It is made on a 7nm process.