Core2Duo RMA - Advice Requested

[n1tro]

Hey guys. Over xmas, I performed my first full build - a c2d machine that I've been really happy with - until recently.

e6600 - stock setting atm
asus p5w-dh deluxe
zalman 9500
crucial ballistix DDR2 running at 1069mhz, 5-5-5-15 @ 2.2v
antec P180b
thermaltake toughpower 700w
XFX8800gts
150gig Raptor
Xi-Fi Xtreme Gamer
Lite-On SATA Combo Drive
running winxp sp2

Lately, I've been having a recurring problem, which I believe I've traced to a damaged/faulty processor. Once completely shut down and cooled off, the machine refuses to power up on the first try (sometimes even the second try, third try, et al.). The fans spin, the mobo has power, the hd is working, etc., but my display (NEC 1970gx) reads "no input detected" and remains blank. After a few manual shutdowns/startups, it'll finally 'take', so to speak, and it'll run normally. Normally, that is, until I try running Prime95 -- it always reports an error immediately, regardless of the speed/voltage, and even at stock settings. I've tried removing/reseating RAM and changing the slots I've been using, and that doesn't seem to be the issue.

I've done some poking around on google, and it sounds like the symptoms of a failing processor. Now, I've overclocked the thing, but I've been cautious with it. I've never made huge jumps in frequency or voltage, and at no point were the volts over 1.4 or the cpu temps (at load) over 55c. The max overclock I've taken it to was 3.2ghz, at which point it was passing prime95 for hours on end.

As I'm a broke grad student, I'd really like to RMA the thing back to intel. My finals are over after next wednesday, and I could (probably) go computer free until a replacement arrives. However, having overclocked the thing, I'm wondering what my chances are of even getting a successful RMA in the first place. That's where you guys come in. I'd really appreciate any advice or pointers to smooth through this process as quickly as possible. Additionally, if I've no shot in hell of getting an RMA, I'd like to hear about that too. All suggestions are more than welcome.

Oh, and if I do replace the proc, am I going to need to reinstall windows as well?

Thanks.

 

[Riddlinkidstoner]

1) When you first got the chip, was it acting funny already?
2) Whats the max voltage you pushed it up to? Try 1.45 @ stock speeds. If it still doesn't help then try upping the voltage on your ram, about 2.1v should do it and see if that helps.
3) When you RMA to Intel be sure to tell them that you did up the voltage and frequency to see if it would stabilize so they know you already messed with the chip. If they deny your RMA its probably because you voided the warranty overclocking but usually Intel isn't too bitchy about it so I wouldn't worry.
4) Check your temps and be sure that you didn't fry your chip by improperly seating your hs/f

 

[RamonGTP]

A fualty processor would be the last thing I would check as it is usually the last piece of hardware that goes bad, especailly when it isn't overclocked. There is a much better chance that your problem is related to the power supply, or motherboard, or ram. Since you are unable to power it up at times, I would look into either the PSU or mobo before RMAing the CPU as I think that would be just a waste of time and shipping expense.

 

[Dan_D]

A fualty processor would be the last thing I would check as it is usually the last piece of hardware that goes bad, especailly when it isn't overclocked. There is a much better chance that your problem is related to the power supply, or motherboard, or ram. Since you are unable to power it up at times, I would look into either the PSU or mobo before RMAing the CPU as I think that would be just a waste of time and shipping expense.

I agree completely. I've been a service technician for years and I've rarely seen CPU failures. Even more rare is for a CPU to die and no other component be at fault. Usually if a CPU is bad, the motherboard is usually also bad.

 

[LstBrunnenG]

If possible, I would put the RAM in another computer and run memtest.

 

[n1tro]

Thanks for the replies so far.

1.) When I first got the chip, it was fine. I could run prime95, memtest, whatever for hours on end.
2.) I tried 1.45v @ stock cpu speed. Same result. Prime95 reports an error within seconds.
3.) Regarding temps, I've always got asusprobe on. The worst temp I ever saw was around 55c under load for the cpu and the mobo.

A little update. I had to move some stuff around in my office, so the computer was shut down for about a half hour. When I rebooted, it was the typical non-response - blank screen, cd drive working, hd spinning, fans working. I walked away and let it warm up for about ten minutes and when I came back and hit the reset button, it came right on. That doesn't really tell me too much, but I'm hoping it'll be more helpful for one of you.

 

[LstBrunnenG]

To be honest, that sounds like PSU...usually RAM doesn't get better when you "warm it up"...

 

[MontyAC]

i also tend to put the problem as a faulty power supply. Try using a spare one and see if the problem goes away.

 

[n1tro]

To be honest, that sounds like PSU...usually RAM doesn't get better when you "warm it up"...

Right, but the memory isn't a concern atm. I thought the CPU was the problem area.

I don't have a separate power supply I can play with right now, but I'll order one and see how it goes.

 

[unclewebb]

Regardless of what Crucial says your memory is good for, you are pushing it to the edge at that speed with 2.2 volts. It's not uncommon for memory modules to degrade over time and lose their ability to run at speeds that they were previously OK at.

MemTest86+ should be run for an extended period of time. If it finds any errors then try backing it off to somewhere around DDR2-800.

If you can't find a new power supply to test with then try to borrow a digital multimeter and monitor voltages under load. Make sure the 12.0, 5.0 and 3.3 volt rails all check out while at full load running Orthos. Don't trust bios or software voltage readings because they are never as accurate as a meter. Any significant voltage droop on these rails is a good sign that your power supply needs replacing. I had an Antec Neo power supply fail on me last year with similar behavior. Sometimes it had enough balls ( amps ) to start up and sometimes it didn't.

I'm not sure what the official failure rate for Intel Core 2 Duo processors is but if I had to pick one hardware item that was least likely to fail, the cpu would definitely be it.

 

[n1tro]

Update:

After checking around online, I went back and tried to run prime95 again with just small ffts. Surprisingly, it worked just fine. I'm guessing this rules out cpu failure, which would be a huge relief.

After that, I went back to memtest. It failed immediately @ 1067mhz and also @ 800mhz with the ram @ 5-5-5-15. Edit: ram is failing memtest @ a 1:1 divider as well. Strange thing is, I can run games for hours on end without crashing at all. As far as running @ 2.2v goes, that's the voltage listed at newegg. If that's too high, well, I'll have to plead ignorance on that one. I assumed that the best setting to use would be the one suggested at the place I bought the stuff from.

Good news is, I bought two more gigs of ram on sale over the weekend, so I can swap it all out soon and try again.

So what's group consensus on this one? Are we still thinking power supply, or should I hold off and try swapping in the new ram first? And if so, what's a safer voltage to run? 2.0? 2.1?

 

[RamonGTP]

If you've already ordered the ram and don't have a spare PSU around to test, then I'd suggest waiting until you get the memory, testing that, and then move onto the PSU if you need to.

 

[Dan_D]

Update:

After checking around online, I went back and tried to run prime95 again with just small ffts. Surprisingly, it worked just fine. I'm guessing this rules out cpu failure, which would be a huge relief.

After that, I went back to memtest. It failed immediately @ 1067mhz and also @ 800mhz with the ram @ 5-5-5-15. Edit: ram is failing memtest @ a 1:1 divider as well. Strange thing is, I can run games for hours on end without crashing at all. As far as running @ 2.2v goes, that's the voltage listed at newegg. If that's too high, well, I'll have to plead ignorance on that one. I assumed that the best setting to use would be the one suggested at the place I bought the stuff from.

Good news is, I bought two more gigs of ram on sale over the weekend, so I can swap it all out soon and try again.

So what's group consensus on this one? Are we still thinking power supply, or should I hold off and try swapping in the new ram first? And if so, what's a safer voltage to run? 2.0? 2.1?

You want to use as little voltage as possible. What ever the minimum voltage to keep the system stable is what you want to use. For some modules that is around 1.8v, and for others it's 2.1v. I'd typically set it for what the manufacturer recommends for stock speeds and then move the voltage up while overclocking if and when the system starts to show signs of instability.

Generally in situations like yours, motherboard and or PSU are more likely canidates for your problems than the CPU or memory would be.

 

[ziddey]

definitely sounds like ram problem. hopefully your new ram will rule that out. also, you know you can rma your old set. if 2.2v was specced, then you're by no means voiding warranty. also, it'd be highly advisable to tie a 80mm fan or something blowing on ram, or get the corsair cooler

 

[unclewebb]

n1tro: I think you misunderstood my comments. It's not that running your memory at 2.2 volts is way too high. The problem is that there are not a lot of memory ICs in the market that can run DDR2-1066 speeds with only 2.2 volts.

The DDR2 spec, which most memory chips including the Micron ICs that Crucial use, is 1.8 volts.

High performance memory manufacturers take the good chips out of the bin, feed them more voltage and if they're stable they throw them on a PC2-6400, PC2-8000 or PC2-8500 module.

Team Memory uses Micron chips exclusively for their Xtreem series.
To achieve DDR2-1066 speed at CL5 they have to rate their memory at 2.3 volts to 2.4 volts.
http://www.teamgroup.com.tw/xtreem/overclocking/team-xtreem-ddr2-1066mhz-cl-5%115515/

The same DDR2-1066 memory that Team rates at CL4 needs even more voltage.
http://www.teamgroup.com.tw/xtreem/overclocking/team-xtreem-ddr2-1066mhz-cl-44412/

After having just finished the RMA process with Team ( great experience by the way ), my conclusion is that some companies are getting a little carried away chasing after big performance numbers at the expense of long term stability and reliability. The Micron D9GMH chips that are likely on your Crucial module and that Team uses were originally rated at DDR2-667 at CL5 with 1.8 volts. Keep that in mind when trying to get over DDR2-1000 speeds out of them.

The main reason high performance memory is so expensive is to help cover the RMA of defective modules that don't live up to the hype.

Edit: Here's a great article on what state of the art ( hand picked ) D9GMH memory is capable of. The real world advantages of running memory on the edge isn't that great, especially if you're sacrificing long term reliability:
http://www.xbitlabs.com/articles/memory/display/pc2-10000.html

and here's the data sheet that shows what Micron is designing D9GMH for:
http://www.micron.com/products/partdetail?part=MT47H128M4B6-3

 

[n1tro]

Well, got the new RAM today. Out with the old, in with the new, and presto -- it started right up. I've been running prime95 for the past twenty minutes or so, and it seems good to go. I'll let it keep going for a few hours while I finish up some work, but it seems like flaky ram was the culprit, which is a huge relief.

Thanks everyone for your help, and thanks to unclewebb for the heads-up on expensive memory. Good information to know, even if it left me feeling more than a little ripped off for dropping so much cash on ram. Kind of wishing I went with cheaper memory and the 8800gtx instead, but I guess that's just a lesson for the next build.