Hi,
Last Friday, on the IRC channel, we've noticed some machines are able to run
certain operations on Hurd as KVM guest up to 10x faster than others. As
antrik correctly guessed, this is the effect of Intel's Extended Page
Tables, which allow the guest operating system to deal with it's own page
faults. I suppose AMD's Rapid Virtualization Indexing can have a similar
effect, but I don't have the hardware to test it.
Since most people are running Hurd in a virtualized environment, I think
it's worth pointing this info somewhere in the wiki.
Just as an example of this difference:
Running without EPT ("modprobe kvm-intel ept=0"):
root@debian:~# dd if=/dev/zero of=/dev/null bs=256k count=1000
1000+0 records in
1000+0 records out
262144000 bytes (262 MB) copied, 2.08 s, 126 MB/s
And with EPT:
root@debian:~# dd if=/dev/zero of=/dev/null bs=256k count=1000
1000+0 records in
1000+0 records out
262144000 bytes (262 MB) copied, 0.23 s, 1.1 GB/s
This also applies for some other tests, like the one suggested by Thomas
here (
http://www.gnu.org/software/hurd/open_issues/performance/io_system/binutils_ld_64ksec.html
):
root@debian:~/ldtest/test# \time ./ld-new.stripped -o dump dump?.o dump??.o
0.00user 0.00system 0:07.72elapsed 0%CPU (0avgtext+0avgdata 0maxresident)k
0inputs+0outputs (0major+0minor)pagefaults 0swaps
