This seems more generalized than a GPU though just from the interconnect along: GPUs are optimized for maximum throughput whereas the individual vector processors here appear to be more autonomous than, say, NVIDIA's CUDA cores or AMD's NCUs.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
3 Comments
Back to Article
evancox10 - Tuesday, August 21, 2018 - link
So, it's a computer oriented GPU with a NoC (network-on-chip) for data movement. "massively multicore vector ISA" == gpuevancox10 - Tuesday, August 21, 2018 - link
computer -> computekfishy - Monday, August 27, 2018 - link
This seems more generalized than a GPU though just from the interconnect along: GPUs are optimized for maximum throughput whereas the individual vector processors here appear to be more autonomous than, say, NVIDIA's CUDA cores or AMD's NCUs.