• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Early Ends on dual core Intel

marty9876

[H]ard|DCer of the Month - February 2006
Joined
Jun 11, 2003
Messages
4,906
Huh, well this kinda bites.

Brand new Intel MB/CPU combo

Intel 945 GTP Motherboard
Intel 930 3.0 Gzh x2 2MB Cache CPU
Cheap Crosair Value Select 2x512 MB DDR2 667
Antec 480 True Blue PSU thingy

System just sitting on a bench, no case. Voltage rails look stable, temp ~52-54C CPU

Core 1: Grabbed a QMD, finished it and grabbed another

Core 2: Grabbed a QMD, early end on step ~49. Grabbed a 56 point Gromac and finished it, grabbed a QMD early end step 9 and now grabbed another QMD.


I guess I'll have to test stuff, have not run Prime 95. If memory was bad, I'd guess both core would poop out. If PSU was bad, I'd guess both cores would poop out. Same goes for temps, but maybe not.

Any ideas, ever run across this before? I know, Prime/Memtest blah blah :)
 
That sucks. When you run P95, make sure to run two copies. Maybe try running one P95 and one CPU Burn and see if that makes it crash. If the machine's at stock, that's crazy.

 
How do you run two copies?

I just started stopped all folding. Starting Prime on 2nd core, then started folding on core 1 and prime died right away.

Now Prime running on core 2, nothing on core 1.
 
Just crapped out just running P95 on second core by itself.

Bad core?
 
Make two folders, put prime95 in each and start each up. 1 will core to 1 core and the other will go to the second core. Then you can go in and manually set the affinity in the process manager.

 
I want a dual core - one that works. Marty will you buy me one? :D

Good luck getting everything to work. Have you tested your mem sticks?

 
marty9876 said:
Just crapped out just running P95 on second core by itself.

Bad core?
Or a bad board or, the memory itself as Scout said.
 
intel mobo's are pretty damn stable. run memtest86 on your memory and see what happens.
http://www.memtest86.com/


bad memory is the root of all things evil. I couldn't install windows on one machine because of it :(
 
anyone know a mem test which can be run in windows?

working on this box via remote desktop... ;) :rolleyes:
 
Prime 95 has a blend test option in it, thats basicly testing th ram the most. Thats All I can think of
 
I found a stupid "memtest" windows based deal, testing now. Not sure if it's any good.
 
Correct, just makes a boot floppy.

If it's failing memory tests there's something really wrong. Have another chip to test in the machine, or another machine to try the memory in? That would help narrow the memory/CPU/board problem down a bit.

 
That link is for making a boot disk. I don't think memtest can be run from with in windows.

Well, just tried another memory test, more failures. I've got some DDR2 533 I can plug in there, also can switch boards and try a different (way smaller) PSU.

Kinda buggers. I would have figured FaH would dies on both cores form bad RAM vs. just on one.

Kinda odd testing memory in windows... some can't be tested since windows has it... :p

I always test from the boot disks (UBCD) just not there right now. I crashed the box a few times, good thing it reboots!
 
Is the computer not at your place? I'd think it would be easyest to test if you have it physically there and can run a boot cd with Memtest etc. If you don't have it you should sit back, read a book, watch a movie, etc, and wait until you can get access to the box to run a boot disk or boot cd.



 
MN Scout said:
Is the computer not at your place? I'd think it would be easyest to test if you have it physically there and can run a boot cd with Memtest etc. If you don't have it you should sit back, read a book, watch a movie, etc, and wait until you can get access to the box to run a boot disk or boot cd.




I've been playing Strawberry Shortcake and My Little Pony... serious.

I just killed the box, can't RD into it and won't return pings. Guess I'll have to fix it Monday, it's at work. Left it running over the weekend and wanted to keep and eye on it. Guess this is why.

Switch the RAM, switch the board, switch the PSU. Kinda looking like bad RAM.

First time I tried OZC- sticks would not with with the boards onboard video
First time I try Corsair- this

Mayeb I should stick with Kingston :rolleyes:
 
I was under the impression that you shouldn't run two instances with -advmethods / large packets. I am not sure if the consensus has changed, but running two QMD's will pretty much take up all your memory bandwidth. If you past memtest, try taking off advmethods/large packets for the second core.

 
dr_dirtnap said:
I was under the impression that you shouldn't run two instances with -advmethods / large packets. I am not sure if the consensus has changed, but running two QMD's will pretty much take up all your memory bandwidth. If you past memtest, try taking off advmethods/large packets for the second core.



On HT chips, you really should not. On dual core chips, I assume your good to go running what ever you want. On the Intel chips, I assume you would run out of bandwidth thou anyways due to shared memory bus.

Odd thing 1: System was doing 317 PPD on one QMD, and 634 PPD on 2 QMD. Huh

Odd thing 2: System stable on 1 QMD/1 Gromac, not stable on 2 QMD.
 
Even on Dual core's you really shouldent run 2 QMD's

1 QMD by itself will just about saturate your memory bus
2 QMD's and it is brought to its knees.

I remember a discussion on the official folding forums, that
for dual core Intels' you actually get a better overal PPD
going 1 QMD/1 Something else, than you do 2 QMD's because
if you got 2 QMD's rolling they end up having to fight for
memory.

 
marty9876 said:
On HT chips, you really should not. On dual core chips, I assume your good to go running what ever you want. On the Intel chips, I assume you would run out of bandwidth thou anyways due to shared memory bus.

Odd thing 1: System was doing 317 PPD on one QMD, and 634 PPD on 2 QMD. Huh

Odd thing 2: System stable on 1 QMD/1 Gromac, not stable on 2 QMD.

Check out OC-Amd almost leaves FAH

I used to run big packets/advmethods on dual xeon 2.8's, but two qmd's choked the life out of the memory bandwidth.
 
yea, I wanted to see how bad it choked. from the PPD, looks like not too bad. these were short runs, so the numbers are not all that valid
 
Yeah, I've always heard that 2 QMDs will bring a machine to it's knees. I run 1 QMD/1 regular WU on my dual xeon 3.0, and it still responds fine(w/ 2GB ram).

As far as your problem, it almost has to be the memory or motherboard(or maybe PSU). Have you checked the processor heatsink to make sure it is mounted good? I doubt it would be the problem, as the cores should just scale back.

I haven't ever had problems with Corsair ram. I'm pretty much a Corsair whore, and run it in just about all my machines. Most of them are work boxen where I use value select, without any problems. If you have already tried 2 different types of ram though, I'd guess it points more toward the motherboard. Do you have a spare board you can move everything to, or at least a different processor to try in the same board?
 
2 QMDs seem fine to me on my dual xeon, but 4 is a big no way! Got 4 once and it was a nightmare, to even shut the machine down.

Marty, I hope you can get everything working monday. Another dualcore processor would be a nice addition to the team.

 
MN Scout said:
2 QMDs seem fine to me on my dual xeon, but 4 is a big no way! Got 4 once and it was a nightmare, to even shut the machine down.

Marty, I hope you can get everything working monday. Another dualcore processor would be a nice addition to the team.


It's gonna be working one way or another.

Kinda funny, I've read/posted a million times you should not run 2x QMD's. I think generally running QMD's on the HT part of the chip is where you get into trouble.

I've got a dual Xeon coming in tomorrow, I'll mess around with it too.
 
LOL marty9876,

I think stanford was afraid... very afraid, I heard from an inside source that you are the real reason that they are taking QMD's off line!
 
dr_dirtnap said:
LOL marty9876,

I think stanford was afraid... very afraid, I heard from an inside source that you are the real reason that they are taking QMD's off line!

it's kinda funny timing. I don't prime every box I get so if I had not been messing around with QMD's I never would have found this problem. The box does normal's -ok- so far.
 
Update: Looks like an iffy motherboard or stuff was not seated all that well. Moved everything (CPU/RAM/PSU) over to a different board, good to go.

I've had problems running Intel boards near the top of their supported CPU listings... :rolleyes:
 
Back
Top