Don't have Telegram yet? Try it now!
https://hgpu.org/?p=4089
Analyzing throughput of GPGPUs exploiting within-die core-to-core frequency variation