Results 1 to 3 of 3

Thread: PrimeGrid vs SoB

  1. #1

    PrimeGrid vs SoB

    Is there a significant performance difference between running SoB work using the SoB client versus the PrimeGrid client? I'm running an AMD 6-core and want to make sure I'm being as efficient/helpful as possible. If you don't know an answer offhand, can you point me to a way to benchmark or compare the two? Thanks!

  2. #2
    Quote Originally Posted by Bananaman42 View Post
    Is there a significant performance difference between running SoB work using the SoB client versus the PrimeGrid client? I'm running an AMD 6-core and want to make sure I'm being as efficient/helpful as possible. If you don't know an answer offhand, can you point me to a way to benchmark or compare the two? Thanks!
    The general consensus back when PG first started doing SoB tasks was that the speeds were pretty much the same. I don't know if there's been much comparison between the two since then, but there haven't been any major developments in the testing softwares, so it should still be applicable.

    The main difference between the two (besides the fact that they use two different softwares for testing - PG uses LLR inside of Boinc, SoB uses prime95/mprime) is that the testing ranges are different, PG is currently testing around n=21M while SoB is currently testing around n=24M, so the SoB's will naturally take a while longer per test. Also, PG does double checking up-front by running each number twice, while SoB saves all double checking efforts until some unknown future date. This double-checking makes PG take twice as long to go through a range, but basically once a range is done, it's done, unlike with SoB. (Given SoB's history with finding primes during double-check efforts, I think the real-time double-checks, or at least keeping the double-checks closer behind the first-checks than they have been, is the way to go, but that's another topic.)

    Personally, I run SoB with Prime95, but I've been a set-it-and-forget-it cruncher for the past few years. PG is kind of fun in that you can participate in the challenges, collect badges, etc. if you are so inclined.

    So, I guess I'd tell you to choose whichever method you prefer, and performance will be the same. With a 6-core, you might compare varying the amount of threads/workers to find an optimum setting. For me, I've landed on 2 workers with 2 threads each on my i7 965 quad-core (w/ hyperthreading disabled.)

    If you want to benchmark for yourself, you can manually run a test with Prime95/mprime to compare times. (It's been too long since I've used the PG client, but if you can't run a manual test with PG then I would just run whichever test PG hands out and run that same test manually with Prime95/mprime)

  3. #3
    Personally I like Prime95 because I can see what k and n are being tested.
    I also have PrimeGrid tasks running it too for BOINC credit but can only really see what number it was testing after its done AND validated and you have to go clicking all over and logging into stuff etc to find it..
    I have been testing SoB numbers on and off for 9 years or whatever and these numbers become familiar and almost like family. So I like to know which numbers I am testing.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •