Benchmark app for your CPU

[dangi12012]

^ Are you running Windows? In Linux this works. I use this in C++ for threading:

Code: Select all

  g_cpus = std::clamp(nth, 1, static_cast<int>(std::thread::hardware_concurrency()));

collatzzz tries to grab all cores possible. Use "-cores 3" to limit cores. As for open source code. Will see. That's just stress...

I added some scaling pyplot on my 16 CPU / 4800U: https://github.com/SamuraiDangyo/collatzzz

I generator 1 million Collatz conjecture steps here: https://github.com/SamuraiDangyo/collat ... -steps.txt
Are you finding errors?

Collatz sum 0 -> 1B in 22.9 seconds

Code: Select all

> ./collatzzz -sum 0 1000000000

... Sum ...
==============================
Collatz:    0 -> 1,000,000,000
Sum(steps): 203,234,783,374
CPU(s):     16
NPS:        8,867,910,000
Time(ms):   22,918

Yes you borked the windows performance (also please add .exe to the release filename)
because you are using POSIX and its not build with MSVC - or CLANG on Windows.

Ryzen 9 5950X:

Code: Select all

collatzzz-x86-0.3-windows.exe -sum 0 1000000000
collatzzz 0.3 by Toni Helminen
... Sum ...
==============================
Collatz:    0 -> 1,000,000,000
Sum(steps): 203,234,783,374
CPU(s):     32
Hash(MB):   0
SPS:        5,016,161,106
Time(ms):   40,516

I cant say much more without sourcecode. I would be interested to see CLANG vs GCC vs MSVC on a basic problem like this.
No worries - just push them to the git repo in a src folder.