3 ms·
tl;dr RISC-V with C extensions is a bit denser than x86-64 with vector operations disabled, and can execute common operations in slightly less cycles due to ma
by kstenerud 10y ago
tl;dr
RISC-V with C extensions is a bit denser than x86-64 with vector operations disabled, and can execute common operations in slightly less cycles due to macro ops (especially in compare-branch idioms).
So until they add vector extensions to RISC-V, x86-64, though bloated, is still king.
- _chris_ 10y agoI compiled the x86 code using gcc 5.3 using mtune/march=native and -O3. It just turns out that the gcc compiler generates entirely scalar code for SPECInt!
- astrange 10y agoThat totally seems wrong. I'm certain it should fire on SPECint, since all compiler optimizations are pretty much invented just to improve the SPECint score - gcc's autovectorizer was made to do this for IBM's POWER processors, which are of course not x86. Maybe it didn't detect '-march=native' properly?