Folding Setup Question

Discussion in 'Distributed Computing' started by tsuehpsyde, Mar 7, 2014.

  1. tsuehpsyde

    tsuehpsyde [H]ardness Supreme

    Oct 22, 2004
    Here are my current stats:
    (Note: currently down!)

    So a little background. I folded for a good long time, and at some point, just fell out of it (mostly because I just stopped maintaining instances and didn't run as many computers 24/7). Fast forward a few years, and I work at a place where I get some credit towards cloud servers and I have access to more computers again. Fast forward 2 more years and my grandfather passed away from colon cancer and other complications. So in November of 2012, I started back up again.

    My real goal here is to make sure I am getting the most bang for my compute buck. I assume that's points, but honestly if something provides more to the project and gives me less points, I'd rather do that. But I've just been doing things the way I've always done them, and I want to see if there are new 'best practices' out there based on what I have available to me. I've read around, but I really can't tell.

    So, onto the computers. Everything is running v6.something via CLI on Linux with -smp and advanced units set to yes. Windows too, though I dunno the GPU CLI it's running off hand.

    Gentoo - 15 x Quad-Core AMD Opteron(tm) Processor 2374 HE with 256MB of RAM
    CentOS - Single Core from an AMD Opteron(tm) Processor 4332 HE with 512MB of RAM

    Dedicated stuff:
    Ubuntu - Intel(R) Core(TM) i7-3770 CPU @ 3.40GHz with 16GB of RAM
    Ubuntu - AMD Athlon(tm) II X2 250 Processor @ 3GHz with 8GB of RAM (unused GTX 460)
    Ubuntu - Intel(R) Celeron(R) CPU G530 @ 2.40GHz with 8GB of RAM
    Windows 7 - Intel 2500K @ 4.2GHz with 16GB of RAM and GTX 560 Ti 448 Core (both folding when it's cold)

    But everything here is CLI (including Windows), and I haven't tried setting up the 460 at all.

    I'm around 110K PPD with my setup.

    Some notes:
    -I think(?) a newer version of the client on Windows may bring more points? I see beta/bonus units and so on, but I gloss over trying to figure it out.
    -The first 2 dedicated boxes are 100% free to run, but I don't know how loud the 460 is folding. If it's super loud, I really can't run it but heat isn't a problem at all.
    -The cloudy folders have variable performance, but I do some magic to try and find systems without CPU contention.

    Any thoughts/questions? Or do I have it mostly okay?
  2. Zagen30

    Zagen30 [H]Lite

    Jun 28, 2012
    The new v7 client primarily benefits Kepler on the Nvidia side and 7000 series and newer cards on the AMD side. This is not really due to the improvements in the client (the client itself doesn't do any calculations) but because the new GPU core (core 17) that came out about a year ago only works with v7. Any reference to beta WUs is probably talking about core 17 WUs, though at this point none of the core 17 projects are restricted to beta. I'm not sure how much your 560 Ti will benefit- Fermi cards can run core 17, but they don't see the large PPD gains over the old cores that Kepler cards do.

    Regarding your 460, one of core 17's benefits is that it runs on Nvidia cards in native Linux, while the previous GPU cores did not (AMD Linux GPU support is waiting for better drivers). It is possible to set up the v6 GPU client via WINE, but that isn't the easiest thing to do. I used to fold on a pair of 460s and they weren't particularly loud.
  3. tsuehpsyde

    tsuehpsyde [H]ardness Supreme

    Oct 22, 2004
    So I went ahead and setup folding on the GTX 460 today and it seems pretty quiet. Estimated PPD is 15K, which is certainly higher than whatever the dual core AMD could churn out. =) Thanks!