road-runner Posted December 9, 2007 Posted December 9, 2007 I got it setup with dual SMP Folding. Ubuntu 7.04 Feisty Fawn @ 3920Mhz, less than 10 minutes a frame. According to Fahmon I am going to get 5247 PPD on the Quad core. With the Windows SMP client I was getting 3500 PPD. This is the way to go guys if you have a Quad core and Fold.... Project : 2653 Core : SMP Gromacs Frames : 100 Credit : 1760 -- Q6600#1 -- Min. Time / Frame : 9mn 39s - 2626.32 ppd Avg. Time / Frame : 9mn 41s - 2617.28 ppd Cur. Time / Frame : 9mn 39s - 2626.32 ppd R3F. Time / Frame : 9mn 39s - 2626.32 ppd Eff. Time / Frame : 2h 18mn 20s - 183.21 ppd -- Q6600#2 -- Min. Time / Frame : 9mn 37s - 2635.42 ppd Avg. Time / Frame : 9mn 37s - 2635.42 ppd Cur. Time / Frame : 9mn 37s - 2635.42 ppd R3F. Time / Frame : 9mn 37s - 2635.42 ppd Eff. Time / Frame : 2h 18mn 00s - 183.65 ppd Share this post Link to post Share on other sites More sharing options...
spazmire11 Posted December 9, 2007 Posted December 9, 2007 ?? but isent the smp client native four core? i mean there are four core process running...... Share this post Link to post Share on other sites More sharing options...
road-runner Posted December 9, 2007 Posted December 9, 2007 ?? but isent the smp client native four core? i mean there are four core process running...... Yes it is but it runs just fine on dual cores, so why not use two on a quad core? Share this post Link to post Share on other sites More sharing options...
romeo55 Posted December 9, 2007 Posted December 9, 2007 No, what I thought that the Linux SMP client was designed for 4+ cores already (not in Vmware)? Or does running two instances on two cores seperately result in a better PPD than running one instance using all four cores? Share this post Link to post Share on other sites More sharing options...
kendellrt Posted December 9, 2007 Posted December 9, 2007 Yes it is but it runs just fine on dual cores, so why not use two on a quad core?I don't see the purpose when it is made for quad core chips. Share this post Link to post Share on other sites More sharing options...
road-runner Posted December 9, 2007 Posted December 9, 2007 No, what I thought that the Linux SMP client was designed for 4+ cores already (not in Vmware)? Or does running two instances on two cores seperately result in a better PPD than running one instance using all four cores? No VMWare. I am sure there are 8 folding cores running as I have not set any kind of affinity's and do not know how on Linux either. I don't see the purpose when it is made for quad core chips. The purpose is about 1500 more PPD over Windows SMP... Share this post Link to post Share on other sites More sharing options...
oblivescence Posted December 9, 2007 Posted December 9, 2007 According to their site last I checked, the client is designed for 4 cores. However, from what I heard, the smp client does not quite use 100% so maybe having 2 instances will allow one to glean a few more ppd. heres stuff from stanfords faq that may seem somewhat relevant or just interesting. What about hyperthreaded (HT) CPUs? The SMP client was originally intended for multi-core CPUs, which generally do not support HT. For machines with 2 physical CPUs, we do recommend enabling HT for the SMP client as this presents the operating system with what looks like 4 logical processors (and our SMP client is intended for 4 processors). If you have 4 physical CPUs, we recommend against using HT, as this presents the operating system with 8 logical processors, which will make the SMP client run inefficiently (especially since the logical processors coming from HT run much slower than the normal ones). How well does MPI work? The short answer is pretty well on Linux and OSX and not so well on Windows. MPI was originally delveloped on UNIX, so this is not a surprise (and it's a great feat in many ways for it to even run on Windows). The Windows specific quirks we're seeing are due to MPI-Windows interaction, and we're trying to hunt them down, as well as try out other MPI possibilities. Why lock to four processes? Gromacs in all released versions currently breaks up code to set up calculations and those to run them and the number of SMP processors is decided at setup (Grompp) not running (mdrun). MDRUN is the code running in the FAH core, so it has to have a fixed number of SMP processes. We are investigating possible options to change this. Share this post Link to post Share on other sites More sharing options...
road-runner Posted December 9, 2007 Posted December 9, 2007 Windows would take 13 hours to do 1 WU. Linux is going to do 2 in 15 hrs.... Share this post Link to post Share on other sites More sharing options...
spazmire11 Posted December 9, 2007 Posted December 9, 2007 well now that is quite a diffrence..... now time to figure out why my ubuntu server only gets 500ppd...... Share this post Link to post Share on other sites More sharing options...
sdy284 Posted December 9, 2007 Posted December 9, 2007 holy balls Share this post Link to post Share on other sites More sharing options...
kingdingeling Posted December 9, 2007 Posted December 9, 2007 Now that is fast! I'll have to see what VMware does with two folding clients Share this post Link to post Share on other sites More sharing options...
hardnrg Posted December 10, 2007 Posted December 10, 2007 My current PPD for two VMs (TWO instances of Ubuntu 7.10 inside VMWare, with ONE commandline smp client running in each VM) is 3835 CPU: 8x450=3600 RAM: 450, 4-3-3-10-2T I have to resolve some performance issues in general on this rig... my SPi times suck... but 3835 is more than I used to get in a week not so long ago lol (well, like 2 years ago, but meh) Share this post Link to post Share on other sites More sharing options...
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now