TR Forums

ok I have an E8400 (currently OC'ed gently at 3.4GHz) 2 Gigs of ram win xp pro and an 8800GTS 512MB OC edition

I'm Trying to think of the differnt configurations I can use with my equipment to maximise my PPD.

At the moment I'm running this:
Folding at home console client on core 0 of E8400 with -forceasm in service mode.
FAHGFX on Nvidia GPU. assigned to core 1 (trying to get it not to interfeer with the other CPU FAH process)

I was wondering Would it be possible for me to run an extra FAH CPU client on Core 1 without any noticiable slowing down of PPD? would it improve my PPD?
or would a better choice be to run the SMP client on my E8400 ( I know its really meant for quads but i have heard it can run on dual cores)

Note I am aslo running fah on my media server and family desktop (climbing to the top of the ladder !)

EDIT: My 8800GTS 512 is currently running driver version 180.48 WHQL - If i upgrade to the newest version (181.20 WHQL) do you think it would effect the eficiency of FAH?

First check your current CPU use for your GPU client - it should be very low, otherwise install the latest Nvidia drivers.
Your CPU is more than adequate to run the SMP client effectively. It would replace your existing CPU client. There are two SMP clients that you can run, the Windows one and a Linux one under VMware.
Why not just the Windows SMP client? Unfortunately it has the honors of being the worst folding client at this time with many people have difficulty getting it to run properly. Also the much better Liunux SMP client is faster (and gets more points) even in VMware. The general FAQ with Windows instructions is at http://folding.stanford.edu/English/FAQ-SMP - to see more information about the vmware/linux option, check the first post for the thread link in my signature (if you go the VMware route, double check to make sure when downloading the latest version as it might have been updated since the last time I've updated the post). Another option with VMware is to use notFred's client.
Good luck!

If you decide to run the Linux SMP client (whether on a native Linux system or under VMware), make sure you do something to deal with the a2 core hangups unless you will be checking on the system multiple times a day. More info here.

just brew it! wrote:

If you decide to run the Linux SMP client (whether on a native Linux system or under VMware), make sure you do something to deal with the a2 core hangups unless you will be checking on the system multiple times a day. More info here.

I would second this suggestion, my friend had faced the same problem last time we tried ubuntu.

Jigar2speed5095 wrote:

just brew it! wrote:

If you decide to run the Linux SMP client (whether on a native Linux system or under VMware), make sure you do something to deal with the a2 core hangups unless you will be checking on the system multiple times a day. More info here.

I would second this suggestion, my friend had faced the same problem last time we tried ubuntu.

It's not just an Ubuntu problem, it affects Fedora (and presumably other distros) as well. The SMP client simply doesn't consistently shut down cleanly at the end of a WU. But best PPD (for a PC-based, non-GPU client) is still the Linux SMP one; so if you're really looking to maximize your PPD, you need to deal with it. The scripts notfred and I have worked out automate the recovery procedure, so you don't need to deal with it manually.

I concur that your best PPD will be found by:

1) running the SMP client in a linux VMware environment
2) running the GPU2 client on your 8800GTS with the shaders OC'd

Glad to see another folder joining our ranks, especially when they have the kind of power your rig has!

Thanks to everyone who helped me out, I'm still in the middle of setting up my little fleet of folding machines

So far I have:
Main Desktop: -
E8400 (@3.4GHz) Currently running x2 Standard FAH clients 1 per core. (Working on switch to SMP Client within VMware)
2GB DDR2 RAM(840MHz)
8800 GTS 512mb OCed (Shader :1880Mhz, Memory: 1080 Mhz, Core: 780 Mhz) Running GPU2 FAH client

Home Server:
P4 Northwood (@3.15 GHz) - 1x FAH standard Client
1 GB DDR Ram

Family Desktop:
AMD Athlon 4000+ (@2.65GHz) - 1xFAH Standard client
2GB DDR Ram

Laptops:
Intel celeron 1.6Ghz - 1x FAH standard Client

New "Dominator" Build (inprogress, still buying parts):
Intel Core 2 Quad Q8200 2.33GHz Socket 775 1333FSB 4MB Cache Retail Boxed Processor
Gigabyte GA-EP35-DS3L iP35 Socket 775 8 channel audio ATX Motherboard
ASUS HD 4870 512MB GDDR5 Dual DVI HDCP HDTV out PCI-E Graphics Card
Crucial 2GB Kit (2x1GB) DDR2 1066MHz/PC2-8500 Ballistix Memory Non-ECC Unbuffered CL5 x2

MickyT wrote:

setting up my little fleet of folding machines

That is known as a "farm" around here.

MickyT wrote:

Main Desktop: -
E8400 (@3.4GHz) Currently running x2 Standard FAH clients 1 per core. (Working on switch to SMP Client within VMware)
2GB DDR2 RAM(840MHz)
8800 GTS 512mb OCed (Shader :1880Mhz, Memory: 1080 Mhz, Core: 780 Mhz) Running GPU2 FAH client

Do check the CPU utilization of the GPU client. You don't want SMP folding to make the GPU client slow to a crawl as it is still the better ppd performer. Use a monitoring tool like FahMon to see if the dropoff of the GPU client is worth it. If it is ok, then you should still set the GPU client's process priority to be higher than than the VM (suggestion: VM = idle, GPU client = lower).

Welcome to Team TR MickeyT!! You should be able to get a good total output with the folding farm you are putting together.

Glad to see another get bitten by the Folding Bug!

With the latest Nvidia drivers, an 8800GTS should be able to share the CPU fine with an Linux SMP client.

- JP

I've started my own folding team #158474 . My mom has alzheimers.

I have two systems up and running the folding @ Home Nvid GPU CLIENT and a third that is almost ready.

All Three are modified Dell Inspiron 530's with adequate power supplies, running Windows Vista Premium 32 bit:

This is what I've downloaded and am running from the folding @ home site:

Windows XP/2003/Vista GPU System tray client with special viewer for NVIDIA GPU's (installer msi) 6.23

Primary:
Intel Core 2 Quad Q9300 @ 2.53ghz
4 Gigs DDR2 @ 800mhz
GTX 260 (BFG Max Core Oc'd)


Secondary
Pentium Dual core E 5200 @ 2.5ghz
2 gigs Ram 800
8800 GT (zotac amp!) THIS RAN IT's First 24 hours and scored 2,600 Points is this the norm?

THird system Not yet unpacked and want your advice on third vid card:
Celeron Processor 440 @ 2 ghz *(My earlier posting revealed that today's celeron's are adequate when running the gpu client).
2 gigs @ 800 mhz
8800 or 9800 NVID or AMD HD 4830 or 4850.... Please suggest the best one as this will be folding ONLY.
Do I need a monitor for this? Can I set it up and just run it after disconnecting the monitor (probably yes but just double checking).

questions.
1.) How do I optimize the folding clients (I enabled 10MB size files already) to maximize my PPD on each one of these systems. I have simply downloaded and run the Folding @ Home Nvidia Screen Saver GPU Client. Is this the best one for Max PPD? I've also downloaded the latest nvid drivers.
2.) AMD OR NVID?
3.) Have you found any benchmarks for Folding PPD broken down by GPU?
4.) Is there a way to fold on the gpu and also have each cpu core fold?
5.) How many Points Per day should each system get. Do the points vary daily. Do they depend on the work units?

Well my evga 8800gt sc card makes 3-5kppd.. I've oc'd it harxd with an aftermarket cooler and get 4-6Kppd.
My bfg 260oc maxcore gets 5-8kppd stock

My quad using notfreds vm-applience gets ~2-3000ppd

I've been averaging 11-12K ppd this way.

Right now Nvidia has an advantage with their design and fahs current work units.
Dual cores are hard o reach max numbers as you're gonna steal power from either gpu or cpu core.
I've seen a gpu list with ppd and power useage over at extreme overclocking...

Nvidia is superior right now if you are just looking at Folding performance.

You can run both an SMP client and GPU client on the Quad and Dual Core boxes.

- JP