Prime95 fails within seconds (4400+, Toledo)

[krille]

Hey!

I'm having serious stability issues here with my sig rig. When dual priming my 4400+ (I even tried 4x200=800Mhz) core 0 keeps failing within seconds. It doesn't even pass the first iteration. It just says like this all the time:



I've tried raising vcore, lowering vcore, lowering multiple, raising and lowering fsb, loosening ram timings, raising and lowering chipset voltage, etc. It's always core 0 failing. However, if I run P95 starting with core 1 and then changing in the task manager to core 0, it doesn't seem to crash? Btw, I just removed the IHS from my toledo. But it wasn't really prime stable before either. I've always had issues with this system crashing. Core 1 on the otherhand seems to be able to go on forever, and temps (while priming on core 1) are below 40C (38C currently). I'm really without a clue here, any help would be greatly appreciated!

Core info:
ADA4400DAA6CD
CCBWE 0522TPMW
1265168F50124
There is an S on the IHS, and a T on the core.

Thanks!

[Lood007]

Try putting the ram on an insanely low divider?
Run it at like 100 MHz and re-run the test.

[krille]

Try putting the ram on an insanely low divider?
Run it at like 100 MHz and re-run the test.

I will try this. Yet, if it is the RAM, why would only core 0 fail and not core 1? Anyway, thanks for the input, will return soon with results.

[dfx]

Try Dual Prime Small FTT
If u are doing a Large FTT or Blend FTT u will most likely fail
Due to the ram tweaking value in the bios

A guide for u..
Use SP2004 to prime
It's more user friendly
Mine can do dual prime 2.6GHz @ 1.48v for 24hrs on stock hsf

Over volted at the point of time when i trying to run it in
Now doing 1.48v dual prime stable

[krille]

I tried lowering Memory divider to 100 (2fsb/1memory), didn't help the slightest.

dfx, I've never tried SP2004, but I will try running first dual Large FFT and then dual Small FFT and see if there's a difference. If not, I shall go on and try SP2004.

edit: I used Dual Blend before. I used 800MB RAM / instance (since I have 2GB total), the rest I left at default values. Now I'm trying Large FFT, hasn't failed so far, but I've only run for a minute or so.

[krille]

Ok, it doesn't seem to fail when using dual Big FFT (lasted a good 7 minutes error-free before I ended testing, dual blend didn't last a minute), nor dual Small FFT.

But now, question still remains, why does it fail during Blend?

[dfx]

Try tweaks the rams value

[krille]

Very interesting. We're on to something here dfx, the lower I set the RAM to on the core 0 instance of P95 (while still having the core 1 instance of P95 running in the background at all times), the longer it took for it to fail. Then I tried running only core 1, but with 1600 MB (with 800MB it was no problem) but the error occured when I set it to 1600 MB - this time with core 1 too. So obviously something's up here. Any clues?

[Bar81]

dfi mobo probably. My X2 4400+ and Opty 170 have no problems with any of the tests at full ram speeds and oced.

[dfx]

I experience what u had gone thru before
I'm having headache running large FTT or Blend on my 2 x 512MB BH5 (No problem small FTT @ 2.65GHz on Air and my 2 x 512MB BH5 can do 265MHz 2225 memtest #5 error free for hundreds of loops)
But the moment i fire up large FTT or Blend (Even @ 2.6GHz 1:1), one of the core would failed definitely
All this problem solved when i tried loosen the ram timing in the bios
Same goes to my 2 x 1GB UCCC as well
I need to set the ram timing value @ all auto in order to run L/B FTT
So i highly suspect its the timing of the ram that caused the instability

[dfx]

dfi mobo probably. My X2 4400+ and Opty 170 have no problems with any of the tests at full ram speeds and oced.

Care to share your experience different between this 2 dual cores?
I'm in dilemma whether should i go for the 170 or not
Thx..

[krille]

Ok, I'll try to set all BIOS values to Auto and try again. Do you know if this can just as well be done in A64 tweaker? Or is it better done in the BIOS, that is the standard way?

[dfx]

I done it thru the bios

[krille]

Set everything to Auto (everything!), but no, still same problem. (Raised fsb/memory ratio to 1:1 or 200mhz at the same time though.) Could it be that one of my memory sticks are bad or something? Should I try running with only 1x1GB?

[dfx]

Try this setting below (All auto except the following)
Catch up with u again
It's 3.15am now
Need some sleep

Idle Cycle @ 256
Dynamic Counter @ Disable
RW Queue @ 16x
Bypass Max @ 7x
32 Byte Granu @ Disable

[Lood007]

try each stick individually

[krille]

Nope, still doesn't work with those timings dfx.

Will have to try each stick individually, I guess.

[krille]

"First" stick used to be in slot "DDR 4" in the manual (the orange closest to the cpu), didn't even boot up when I took out the "second" stick in slot "DDR 2" (the orange slot farthest from the cpu). Then I took out the first stick and pulled the second stick back in the fourth slot. Now it not only booted, but was P95 stable for well over 15 minutes before I cancelled the test. I will try to boot up the first stick alone in slot "DDR4", if it still doesn't work, the first stick seems to be faulty. If it does boot up, it's probably my mobo that's faulty (or the 4th DDR slot, orange closest to CPU to be exact).

Update 1
Now I've booted up with the first set of RAM, but this time in slot DDR4 (orange one farthest away from CPU). Gonna try some prime, but both sticks do boot into windows, at least in the fourth slot.

Update 2
First stick fails P95 within seconds using the same settings as second stick. Second stick passed 15 minutes at least, first dies within seconds. It's probably the RAM then? Will try the second stick again in the same slot just to make sure.

Update 3Yup. Second stick passes dual P95 for several minutes (than I manually cancel), first stick fails dual P95 within seconds. Both in same slot and using the same settings. Clearly something's up with the first. I'd still appreciate more input though on what to do next. Thanks!

[ixtapalapaquetl]

Man, thank you so much for starting this thread. I was beating my head against a wall trying to get my Opteron 175 past 2.97 - exact same problem regardless of voltage. You convinced me to loosen up past what I would normally ever consider. Switched to 3-4-4-10 2T and am sailing through Prime95 blend as we speak! Time to ratchet things up! Bet you get me 50-100 MHz more clock! Thanks again.

[krille]

Man, thank you so much for starting this thread. I was beating my head against a wall trying to get my Opteron 175 past 2.97 - exact same problem regardless of voltage. You convinced me to loosen up past what I would normally ever consider. Switched to 3-4-4-10 2T and am sailing through Prime95 blend as we speak! Time to ratchet things up! Bet you get me 50-100 MHz more clock! Thanks again.

np
But, I think your thanks should rather be directed to dfx's fine responses.

For me, though, loosening the timings didn't seem to work. One of my sticks seems very faulty to me, I should probably RMA, right?

[dfx]

Had u tried different ram slots?
Mine better in the orange slots
Tried memtest?

[krille]

Had u tried different ram slots?
Mine better in the orange slots
Tried memtest?

I had tried faulty ram stick in both orange slots. In the one closest to the cpu socket, it didn't boot for some reason. In the other one, it did boot, but failed prime95 within a few seconds (even with a 2fsb:1ram divider). I only tried the functioning ram stick in the orange ram slot farthest away from the cpu socket (in which the faulty booted) and it seems to run p95 just fine under identical conditions.

Nope, I haven't tried mem test, but if the ram module fails P95, even when underclocked, it ought to be faulty, right? Or should I try mem test too?

[dfx]

Tried memtest test 1st before u can confirm that your rams is faulty
P95 results are mixed
So cant really confirm where could had went wrong

[lancediamond]

If you're running single sticks, you want to use only the farthest from the CPU slots *I think*... Seems like I read this someplace in the past few days.

If you can replicate the problem with one stick but not with the other stick in the same slot, I'd RMA that memory...

Edit: Also, if you suspect bad memory, you might try this memtest - it found problems within minutes that memtest86 from BIOS didn't find in hours:

http://hcidesign.com/memtest/

[dfx]

Yep a very good suggestion from lancediamond
Win Memtest are more efficent