It’s called helsing, it’s an open source vampire number generator.
This is a project I started for devember, and now it’s up on the phoronix test suite!
https://openbenchmarking.org/test/pts/helsing-1.0.0
They ran it on a 2 x 64-core epyc system and it scaled better than I imagined.
I have been testing it thoroughly for the past months, but It’s still on beta and I haven’t validated all aspects of it.
If you have any CPU you want to benchmark, try it out, installation takes less than a megabyte.
Looks cool, I will download it and run it on my workstation and a few other PC’s
Do you have a zen3 cpu?
Yes I have a laptop with a 5800H and a desktop with a 5600X. I also have a Threadripper Workstation.
I am testing out a new feature that reduces cache usage almost in half.
Would you like to try it out?
Sure, I can test it on whatever you need.
Thank you so much.
Could you measure the runtime of these two branches on the 5600x system.
For each of them please do:
git clone https://github.com/plerros/helsing.git -b NAME_OF_BRANCH
cd ./helsing/helsing
make
time ./helsing -n 14
And send me the runtime.
This should take ~11 minutes for each of them.
You need me to run linux? I don’t have linux on either Zen3 machine, but I suppose I could boot a live CD on the AW box I have sitting here to test it.
You could run it on wsl, but that is still too much work on your side.
Don’t do it, I can find other ways to test it.
I could boot a live cd and run it from there. I just don’t have linux installed on either zen3 machine that I have.
Well, if that’s not too much trouble for you, I’d appreciate the help.
It should work on any linux distro. It even works on bsd, haiku, solaris, and redox.
Will do it in the next days no problem. I have an Ubuntu ISO on my Multiboot drive as well as a few other distros for fixing/testing PC’s.
I’m sorry I have taken so long to do it, I will get it done for you this weekend. I have been slammed with work and other personal commitments.
The only cpu I have with proper BMI2 implementation is my laptop, so I did a pilot experiment with it.
Almost double the performance if the workload is big enough.
The 5600x machine I have had to come apart for some upgrades. When I finish putting it back together I will test for you. On top of that I got sick in this time since I posted. Sorry again for being so useless lol
Aggressive timing for ram at 3466
Is running it on a 5800U any use to you? I have that as a Zen3 system right now.
I haven’t had a chance to touch the 5600x system again since I have been in the process of moving.
Yeah, it is.
I merged the pdep branch, so here are new instructions:
git clone https://github.com/plerros/helsing.git
cd ./helsing/helsing
for the non-pdep:
make
time ./helsing -n 12
time ./helsing -l 1000000000000000 -u 1020000000000000
time ./helsing -n 14
for the pdep version:
sed "s|^#define USE_PDEP .*|#define USE_PDEP true|g" configuration.h > tmp
mv tmp configuration.h
rm tmp
make
time ./helsing -n 12
time ./helsing -l 1000000000000000 -u 1020000000000000
time ./helsing -n 14