It's not just about a given microbenchmark - that platform just sucks tbh. Hopefully something better appears.
Re:Scaleway launches RISC-V servers
by MuffinFlavored on Sunday March 03, 2024 @07:45PM
(#39585930)
Can you expand on why it sucks?
Real world performance reported: glibc build on Unmatched ~7:30, on Pioneer ~16:20 hours.
Re:Scaleway launches RISC-V servers
by brucehoult on Sunday March 03, 2024 @08:22PM
(#39586145)
I don't have a Pioneer (I wish!) but I just yesterday compiled GCC head (14.0.1) on my 16GB LicheePi 4A (same cores as the Pioneer, but only 4 instead of 64) and it took just under 7:03 in your notation:
real 422m41.367s
user 1430m56.638s
sys 70m17.994s
It's hard to believe the Pioneer with 16x as many cores and a slightly higher clock speed and 128GB RAM and NVMe SSD vs eMMC would be slower.My expectation is that it would beat my i9-13900HX laptop (24 cores) running riscv64/ubuntu in docker (QEMU) which took 1:42.
real 101m44.472s
user 1695m26.395s
sys 24m36.671s
Note: newlib build, not glibc. But still!The glibc times were reported on the RISC-V IRC channel by davidlt. Here's the backlog for that day.
https://libera.irclog.whitequark.org/riscv/2024-02-02
The stress-ng benchmark was suggested the previous day.
Re:Scaleway launches RISC-V servers
by brucehoult on Sunday March 03, 2024 @09:17PM
(#39586435)
Who knows what he did? There are opposite experiences cited in that same log:
19:03 <xypron> Concerning Pioneerbox vs Unmatched: My experience with building Ubuntu'g glibc was 7 h on Pioneer box vs 3.5 days on Unmatched.
stress-ng --verbose --metrics --aggressive --atomic 64 --timeout 600
If you try this, adjust the number (64) to match the number of cores on your system.