Get the latest tech news
Interesting PEZY-SC4s
Continuing a supercomputing tradition with power efficient FP64 compute
PEZY targets highly efficient FP64 compute by running a massively parallel array of execution units at lower clocks and voltages than contemporary GPUs. Even the higher figure would give PEZY’s L3 better latency than AMD RDNA4’s similarly sized Infinity Cache (measured to just over 130 ns using scalar accesses). At a higher level, efforts like PEZY-SC4s and Fujitsu’s A64FX show a curious pattern where Japan maintains domestic hardware architecture development capabilities.
Or read this on Hacker News