Nathan Maurice

Results 1 comments of Nathan Maurice

Zen 4 based CPUs perform poorly because AMD's implementation of `compressstoreu` is extremely inefficient ([~142 cycles latency](https://www.mersenneforum.org/showthread.php?p=614191) / [throughput of *54-72* according to uops.info](https://uops.info/table.html?search=vcompress%20(M512&cb_lat=on&cb_tp=on&cb_uops=on&cb_ports=on&cb_ZEN4=on&cb_measurements=on&cb_doc=on&cb_base=on&cb_avx512=on&cb_bmi=on)). It is even worse than an...