I'm Asking for Help From the Best Guys Out There - Severe Problem w/ 955XE

[Sneaky]

O.K. guys, here's whats up:

I've had a golden 955XE chip that was just insane, but voltage NEVER, went over 1.55v (ran at 1.5125v stable daily)

But about a month ago it had begun to lag and chop in windows, which I thought was a HDD problem, but it persisted even after 3 reformats and even reformatting with the drives taken out of RAID - then later on it began to actually hard lock in windows (with no BSOD), and then a week after that started, it refused to boot into windows period


I attributed it to my PSU, so I tried a friends and it atleast booted into windows a few more times, but then again refused to load windows again... ordered a Zippy 700w PSU, still no windows boot


since then, I have gone through the following to try and troubleshoot everything:

-Asus P5WD2-Premium 955X w/ 0606 or 0704 BIOS
-Intel D975XBX 975X w/ most recent BIOS
-2x Hitachi T7K250 250gb SATA II (currently out for RMA as another attempt to try and make stuff work)
-Maxtor MaxLineIII 300gb SATA II
-Sapphire x1900XTX
-Sapphire x1900XT
-PCP&C 510ASL 510w
-OCZ Powerstream 520w
-Zippy PSL-6701P-SATA 700w
-3 Windows XP Pro install disks (all with SP2, dont have anything else)
-Windows XP Home install disk
-15-20 SATA/SATA II HDD cables
-4 IDE cables


trying everything above, switching everything out, trying it out of case and in case, etc., etc. - still nothing has made it work again

the 955XE will do anything you want it to in the BIOS, it will POST fine, and even pass a full 6 hours of memtest with ZERO errors, and it boots with a knoppix linux CD without any problems either

it will format a HDD for windows with both FAT & NTFS, will copy all setup files, but when when it reboots, it will lock up and/or restart when the GUI install begins (where you can use the mouse, enter your name and CD key, etc.), or if I install a HDD with windows already installed on it, it will get to the boot screen, load 2 or 3 bars, and then lock up and/or restart and repeat the attempted boot cycle

I've played with every voltage setting, tried all kinds of timings and who knows what else, re-flashed the BIOS atleast 20 times on the P5WD2-P and ordered the new D975XBX because I thought the SATA controller on the P5WD2 was toasted

I've played with the Hitachi drives in Hitachi Feature Tool, setting them to every which way, both SATA 150 and SATA 300, same stuff happens - have tried both the ICH7R SATA ports & the Silicon Image SATA ports


There is no way this chip could, or even should be dead if it does everything that I've listed above, but JUST NOT BOOT INTO WINDOWS :stick:


I am calling ALL people of expertise from every corner of the world to try and help me figure this out - I've stumped myself to the point of ripping my hair out, boggled atleast 5 of my friends, including mikeguava and Reinvent3d about this problem

so far the only somewhat feasable suggestion as to why this problem has occured is the fact that my 955XE may have a junked L2 cache (suggested by mikeguava), but he has NEVER, EVER seen that happen on an intel, or the way that I've explained it to him - the only time he's ever experienced that is with a coldbugged AMD CPU that would lock up in 3Dmark in the same spot every time


no peice of hardware in my computer has been tampered with, mishandled, or damaged in any shape, way, or form, that I know of, and all of it was in tip-top shape and 100% working order just a few weeks ago, but now I'm SOL and currently on a customer's machine while I wait for the last few parts for it to arrive


I also have a Celeron D chip coming on monday to try and troubleshoot things further, but I don't have any IDE drives that I can test stuff with (I have one with no windows install that I have EVERYTHING backed up on, and refuse to risk it with anything), and don't have an extra set of DDR2 available either, but with it passing 6 hours of memtest with the 955XE, it should be perfectly fine as far as I'm concerned



Please, I am open to suggestions from ANYONE at this point, even if its a half-assed poke at trying to solve my problem, I will do ANYTHING in my powers to make my rig work again


Please post here with any questions, comments, or suggestions, or feel free to contact me via email: [email protected], or on AIM: Sneakyl3astard to ask more questions or just have a 1 on 1 troubleshooting chat to see if we can sort anything out


Thanks in advance for any assistance that you may provide
-Justin Rubner

 

[BigDan3131]

You have done the basic stuff like clearing the CMOS right? Neither of the two PSU's you listed are enough for everything in you sig. Zippy? Is that a PSU ? never heard of them just because it says 700w doesnt mean it can handle it full time. These companies tests are something to be doubted because they use ideal conditions. What temps are you getting from the CPU?

 

[Sneaky]

zippy/emacs =server PSU company, and its one hell of a PSU - ask anyone that owns it and they'll attest to its ability as one of the best PSU's out right now (only 2 reputable sellers carry it currently though)

here's the rail specs for it:

5V - 35a
12V - 45a
-5V - 0.8a
-12V - 1a
3.3V - 30a
+5VSB - 2a


yeah, i've cleared CMOS a number of times and popped out the battery and let it sit overnight - same thing still happens



also - if anyone is wondering, all these problems are occuring at STOCK CLOCKS - no OCing whatsoever (and both the OCZ and PCP&C should have no problem handling the system at stock clocks... i've seen a PC P&C run a FX-57 two 7800GTX's without thinking twice, i think it can handle a 955XE and x1900XTX pretty easily as well...)



it was on water cooling w/ my storm G5 - temps have never been above 45ºC under load with this processor


ambient temp is never above 70-72ºF in my house

 

[Vrykyl]

What ram are you running? have you considered that as a culprit? I spent ages trying to stablise my old amd rig and it turned out to be my ballistix ram....
Does the cpu play nice if you down clock it (even if just through fsb)?

 

[Stratcat]

[...] Zippy? Is that a PSU ? never heard of them just because it says 700w doesnt mean it can handle it full time. [..]

Hi Dan -

You may never heard of Zippy 'cuz they're generally commercial server level PSU's, and not generally seen in the consumer PC markets, except by those in the know (like us OCC'ers)!

Believe me, the first time I heard the name Zippy, I was like, wtf, no way!

But yeah, funky name, but very high quality unit, as Sneaky stated.

Strat

 

[crotale]

I have had the exact same problem... My CPU was quite tortured before that,
which I believed to be the reason, but you don't seem to have done
anything special to it.

I replaced it with an other 955XE an everything is working just as it should.
I consider my chip to be dead, but as you explain, it posts as it should.

 

[Sneaky]

What ram are you running? have you considered that as a culprit? I spent ages trying to stablise my old amd rig and it turned out to be my ballistix ram....
Does the cpu play nice if you down clock it (even if just through fsb)?

i've even tried dropping the CPU to 12x 200 or 14x200 down from the 13x266 stock and still no fix...

idk if the ram could be the culprit, as it does pass 6 hours of memtest x86, but on monday we'll be able to either confirm or eliminate the ram as the suspect when i get the celeron D to help troubleshoot this issue

 

[Sneaky]

I have had the exact same problem... My CPU was quite tortured before that,
which I believed to be the reason, but you don't seem to have done
anything special to it.

I replaced it with an other 955XE an everything is working just as it should.
I consider my chip to be dead, but as you explain, it posts as it should.

so it posts but doesnt boot windows... will it work with knoppix, etc?

 

[Stratcat]

Sneaky -

Ooops...Vrykyl beat me to it! But, yeah, I'd try clocking down for the hell of it, though not sure what that would accomplish, but at least see if the proc will work at all. Loading the OS & boot apps (from your good HDD) is actually probably one of the most strenuous process there is. So maybe, for some reason, the chip is actually defectively marginal.

I think you're definitely doing the right thing w/the Cellie. If that POSTS & Loads the OS, then you've narrowed it down to your CPU.

Have you inspected & cleaned the contacts on the CPU? I'm sure you have, but hey, figured anything's worth a shot.

If the cellie runs, then the only thing I could think of, might be an ESD caused problem, tho that should not occur while the proc is inside ur machine.

Oh yeah, have you tried running w/just a single stick of RAM?

Oh...triple check you've got the right BIOS for the cellie!

<edit>

One last thing: Any chance the proc could be doing a catastrophic thermal cut-out (not sure how you'd test this)?

Strat

 

[Placid]

Try it with just one stick of memory if that fails try it with the other stick and use different memory slots if possible if one stick fails.

Disable anything in the bios not needed like sound,com-usb ports,etc.
Remove any cards not needed like sound,ect.
Disconnect any drives not needed to install winxp,from the motherboard and psu.

 

[crotale]

so it posts but doesnt boot windows... will it work with knoppix, etc?

When underclocked, 266x12, I got it into Windows a couple of times, but usually it just looped the Windows logo-screen.
It all started with the same things as you describe, lagging and chopping in windows, and then got worse.

 

[Sneaky]

When underclocked, 266x12, I got it into Windows a couple of times, but usually it just looped the Windows logo-screen.
It all started with the same things as you describe, lagging and chopping in windows, and then got worse.

MOTHERF*****... i do indeed have a browned CPU then



already tried the barebones installs, single sticks, diff. ram slots, removing extra cards, etc.



damnit... thats all i can really say right now... just damnit


the thing that ****es me off the most is that i did absolutely nothing to cause this... and of course the CPU being an ES, i can't do diddly squat about it besides buy a new CPU...

 

[hainer36]

now im not sure or anything, but i am just throwing this out there

do you have SP2 installed into windows? as when i first got my new P5WD2-P and 650 it wouldnt boot unless i had SP2 on it (i think the pci-e card had something to do with it)

just a random guess, im up at 4:30am, i think its ok if its way off

 

[Stratcat]

MOTHERF*****... i do indeed have a browned CPU then

Ohhh man, buddy, that hurts, if true!

But now this begs the question: How high is the "safe" OC Vcore on the 65 nM process?

Hmmm... in the Intel Datasheet [56K .pdf WARNING] table 2.3 states a normal operating max of 1.3375V, while table 2.2 states 1.55V as an "absolute" max.

So it would seem that you were safely running w/i the max operating parameters, & let's face it, this is OC'ing, which by it's very nature means running out-of-spec.

Table 2.2 does have footnote 2 which states: "Excessive overshoot or undershoot of any signal will likely result in permanent damage to the processor", tho "excessive" is an ambigous term, and this sounds like a "boiler plate" type of statement, IMHO.

Strat

 

[Npvk_x]

ooo... crap... hate to see the 65nm cores dying, i also have problems with my rig. it just suddenly reboot... cant find anny logical reason... but the fight continue...

too bad about the xe

 

[proth]

It seems like you swapped everything but the ram. Everything points to a bad cpu, but somehow I don't believe it cause you really never pushed it. I've been running this 955 @ 1.55v / 5+G for a while, but only 4-6 hrs/day and pushed suicides to 1.6. and no problems. The problem I caught myself making with other chips is taking/removing the CPU out too many times and/or moving the CPU from one mobo to another. Everytine the chip is reseated, the pins on the socket either make a new dimple or hog out the existing dimple on the cpu pad. Even putting the CPU back in the same socket doesn't guarantee the exact contact point on the pads. Under a microscoop, or loup, the pads look like someone hit it with a 12ga shotgun. By the way, this is the best way to tell if a chip was "used only once", or "brand new", but not boxed. I don't know if this is your problem, but something to be aware of on LGA packages.

You can clean the contacts with one of those soft white rubber erasers. It takes off tarnish but is pretty gentle to the gold. You can use a more traditional pink type eraser, but it's more abrasive. Just clean the contacts off with some alcohol or board cleaner to rempve all the bits.

I doubt this is it because usually a problem like this drives you crazy and then you do something like reseat the block which shifts the CPU in the socket and it goes away. But as a rule, I would recommend that people leave their CPUs in the sockets and only remove it or pop the bail on the lid when absolutely necessary.

 

[speed bump]

I had a thought that maybe might be weird but its worth a shot.

Crotale, your 955xe was an ES right? and Sneaky yours was to?

I think that maybe intel had some booboo with there ES chips or something Becuase the only chips I have seen experiance this are these ES chips.

That may be off the mark but who knows. Maybe they just have SPDS to which I hop isn't the case.

 

[Sneaky]

already tried cleaning the contacts with alcohol two times - nothing


edit: and when i got it, there were only 2-3 dimples on the landings at most

 

[nikhsub1]

OK, I have not read the whole thread, so excuse me I had a similar problem with a stock 3.2 Pressy on an Asus board. Thought it was memory at first, nope. Turned out to be the MOBO. Have you tried a different mobo? That is my only thought on this unless it is indeed the cpu but somehow I doubt it.

 

[Sneaky]

tried both my P5WD2-Premium 955X and ordered a Intel D975XBX 975X board because i thought my P5WD2's SATA controller was dead

same problem on both boards