Go Back   Hardware Canucks > HARDWARE > CPU's and Motherboards

    
Reply
 
LinkBack Thread Tools Display Modes
  #31 (permalink)  
Old August 1, 2007, 05:27 PM
3.0charlie's Avatar
3.0 "I kill SR2's" Charlie
F@H
 
Join Date: May 2007
Location: Laval, QC
Posts: 9,628

My System Specs

Default

I know that it's a huge drop, but that's all I have for the moment. And I'm running Win2k.
Reply With Quote
  #32 (permalink)  
Old August 1, 2007, 05:31 PM
Somed00d's Avatar
Top Prospect
 
Join Date: May 2007
Location: Oshawa, Ontario
Posts: 69
Default

yeah there is a 20->24 adapter. $3 us from monoprice Only for $2.10 each when QTY 50+ purchased - ATX Power cable 20M/24F - 6 inch | Internal Power Cable=

Quote:
Originally Posted by Wankerfx View Post
I think you can use a 20 pin with a 24 pin slot, I'd have to check up on it.



EEEEEEK!
__________________
Q6600 G0 @ 3.2ghz (965 chipset holding me back)
DFI Infinity P965-S
Gigabyte Geforce 8500GT
Buffalo Firestix FSX800D2C-K2G 2X1GB PC2-6400 CL5
OCZ XTC Active Memory Cooler
Seagate Barracuda 250GB SATA2
MCE Wireless Keyboard / Remote
Logitech X-530 Speakers
Swiftech H20-220 Water kit
Swiftech MCW30 NB Cooling
OCZ PowerStream 600W
Reply With Quote
  #33 (permalink)  
Old August 1, 2007, 05:35 PM
3.0charlie's Avatar
3.0 "I kill SR2's" Charlie
F@H
 
Join Date: May 2007
Location: Laval, QC
Posts: 9,628

My System Specs

Default

Thanks for the info; but since that PSU is too short on power reqmts, I'll have to go for something bigger on the short term.

John
Reply With Quote
  #34 (permalink)  
Old August 1, 2007, 08:04 PM
Fallen Folder
F@H
 
Join Date: Mar 2007
Location: folding in GTA
Posts: 375
Default

Quote:
Originally Posted by 3.0charlie View Post
I think I spoke too fast. E2140 @ 3.0 I started folding yesterday morning and got EUEs in less than 1 hour. So I went back to Orthos - and after 21 hours it failed.... Memtest is fine for that single stick.

So I'm back to square 1 testing with Orthos @ 2.9, stock 5-5-5-15 1T timings and 1.37V. Still have that major Vdroop of .05V and a 12V of 10.2! Temps are beautiful though: 24C idle, load max of 45C using Speedfan.

I think I need a new PSU to replace that FSP 350W. It's just asking too much. I do have an Antec 380W (with more amps on the 12V+), but with 20 pins, not 24 pins.

John
Othos/p95 stability does not ensure folding stability. I only run orthos or quad p95 for an hour or 2 at most. After that i run a test smp wu. If it can complete 30-40% without an eue it is likely gonna be stable.

When you first start folding smp after doing the orthos stable thing for a while, make a copy of the first smp wu you get as soon as it downloads. Just copy the work dir and queue.dat file to a backup directory. Then stop the client, disable the lan connection, restart the client and let it fold.

With the lan disabled it can't report an eue. If you get an eue, just stop the client. And if mpiexec is still running in task mgr, kill it since it will have file handles into the work dir files. Then delete the work dir and queue.dat file from the folding directory and copy the backed up one back in.

before you start the client again adjust your oc ... either more volts to cpu, mem etc, or drop the fsb of memory speed, etc etc.

Repeat above until the rig can complete the wu and you can be pretty sure your rig is stable. Just make sure you have the lan disabled/disconnected before you start the client, and don't disable it while the client is running - smp beta doesn't always handle it clean.

If the wu crunches through to completion then stop the client, re-enable the lan and restart the client to upload the results and get the next wu. You can be pretty sure your rig is stable at this point.

Keep the backed up wu to use again for testing. Just change advanced setting in the client to ignore deadlines, and you can run a test wu later that has expired, been returned.

If you are careful, the bonus of this approach is you don't waste days running orthos only to find you are still not folding stable. Also if you are using a live wu (vs an expired/completed one), when you are done you get the points for a completed wu ... as long as you get is done before the 3-4 day deadline.

It is better for the science to use an expired wu to test with though since it doesn't delay the project. Each wu is part of a sequence ... the next wu in the sequence cannot begin until the one you are running is completed and uploaded since results from one are needed to start the next.

If you turn in an EUE, it will get reassigned to several other rigs to check if it was a 'genuine' eue due to the simulation parameters, or one caused by hardware instability. So 'unnecessary' hardware caused eue's slow down the science and waste cycles. The genuine ones actually give the researchers useful results about their simulation parameters.

The best wu to use atm for win smp testing is a 2610. 2604 and 2605 give better points especially on the small cache cpu's, but do not stress it hard enough. Since you only have one smp rig, and lost your first smp wu when it eue'd, save the next one as above for your next smp test, and for future testing. If you get a 2608/09 you will find it very slow on your rig ... they fold best with lots of cache.

Last edited by pscout; August 1, 2007 at 08:06 PM.
Reply With Quote
  #35 (permalink)  
Old August 1, 2007, 08:11 PM
3.0charlie's Avatar
3.0 "I kill SR2's" Charlie
F@H
 
Join Date: May 2007
Location: Laval, QC
Posts: 9,628

My System Specs

Default

Gotcha. Will do so. I will bring the little rig that could offline and start testing.

Thanks,

John
Reply With Quote
  #36 (permalink)  
Old August 1, 2007, 08:34 PM
Fallen Folder
F@H
 
Join Date: Mar 2007
Location: folding in GTA
Posts: 375
Default

It is also the best way to tune/tweak your memory. Folding perf is usually dependant on memory bandwidth. Sandra is a useful measure and quick, but in my final tuning i will adjust mem speed/timings then run 1 frame of the wu to get a timing - tpf or time per frame. You will see pretty quickly which setting helps folding performance the most. It can get tedious restoring the backed wu after eue's and killing the mpiexec which often hangs around after the client has an error.

In general for smp folding on c2d's, sacrificing tight timings for higher memory speed gives better folding tpf's. There is alway a sweet spot.

Also, running too close to the edge may result in eue's later ... usually due to changes in ambient temps. This is normally a result of cpu temps rather than memory isues and dropping the fsb by 5 or 10 will fix it and give you a safety margin to eliminate any more eue's.

Last edited by pscout; August 1, 2007 at 08:37 PM.
Reply With Quote
  #37 (permalink)  
Old August 2, 2007, 02:28 AM
sswilson's Avatar
Moderator
F@H
 
Join Date: Dec 2006
Location: Moncton NB
Posts: 14,490

My System Specs

Default

Quote:
Also, running too close to the edge may result in eue's later ... usually due to changes in ambient temps. This is normally a result of cpu temps rather than memory isues and dropping the fsb by 5 or 10 will fix it and give you a safety margin to eliminate any more eue's.
I ran into this last week when summer finally hit here in HFX. Both my rigs started to fail randomly on different WUs.... backing off the mem OC on one and dropping 2 FSB on the other seems to have solved the problems.
__________________
MSI Z87I Gaming AC / i5 4670K / 2X 4G Gskill 1866 DDR3 / XFX XTR 750 / EVGA GTX 680 SC+ 2GB / Intel DC S3700 200G / random 160G Sata HDD
Inwin 904 / Swiftech MCP655-b / Alphacool NexXxos XT45 120 Rad / 2X Scythe GT AP-15 / EK Supreme HF / Dell UltraSharp U2412M

Asrock AM1H-ITX / AM1 Athlon 5350 / 2X4G Gskill PC3-14900 / Intel 6235 Wi-Fi / 90W Targus Power Brick / 320G Seagate Momentus / Mini-Box M350 / 1X 22" Dell IPS / 1X 22" HP
Reply With Quote
  #38 (permalink)  
Old August 2, 2007, 07:20 AM
3.0charlie's Avatar
3.0 "I kill SR2's" Charlie
F@H
 
Join Date: May 2007
Location: Laval, QC
Posts: 9,628

My System Specs

Default

Temps were not a problem - highest was 45C. The Swiftech kit was very efficient and the rig being located in the basement, it was very cool....

And I use the past tense since the little rig that could died this morning. No sparks, no flames, no smoke - but a voltage of 1.2V in CPU-Z (with a BIOS selection of 1.45...) and a 12V of 9.4V in Speedfan....

I need a new PSU.

John
Reply With Quote
  #39 (permalink)  
Old August 6, 2007, 04:35 PM
3.0charlie's Avatar
3.0 "I kill SR2's" Charlie
F@H
 
Join Date: May 2007
Location: Laval, QC
Posts: 9,628

My System Specs

Thumbs up

So I got a new PSU - an Antec Trio 550W. And then some Ram - 2x1Gb Mushkin EM2-6400.

Other than the fact that using the XFX 650i, I cannot have both sticks installed, it is running quick.

Folding SMP ATM, at 3.0GHz (Bus Speed of 375MHz), on 1.45V VCore. The Ram is running @ 500MHz, 5-6-6-18 2T Trc 31, 2.1V (default is 5-5-5-18 2T Trc 23, 1.9V).

Temps are 24 idle, 41 load folding 100%. Not bad for a 88% OC.

Life is good.

John
Reply With Quote
  #40 (permalink)  
Old August 6, 2007, 07:33 PM
Wankerfx's Avatar
Allstar
 
Join Date: Mar 2007
Location: Ottawa, Ontario
Posts: 583
Default

Awesome, I'm glad that new PSU helped your overclock.
__________________
[ER]
Reply With Quote
Reply


Thread Tools
Display Modes

Similar Threads
Thread Thread Starter Forum Replies Last Post
Overclocking help tzetsin Overclocking, Tweaking and Benchmarking 41 January 27, 2008 11:05 PM
Pentium Lives!! Pentium E2160 and Pentium E2140 Babrbarossa CPU's and Motherboards 4 June 4, 2007 04:25 AM
DirectCanada is amazing... Arcygenical Off Topic 11 May 11, 2007 05:43 PM
Amazing printer technology from Memjet Babrbarossa Press Releases & Tech News 5 March 23, 2007 12:11 PM