My CPU is degrading

Amaroque

Platinum Member
Jan 2, 2005
2,178
0
0
This is interesting...

I have an A64 3400+ (single core) in one system. It didn't ever OC much, so I never bothered with OCing that machine.

It started about two weeks ago, when I had to set the Vcore to 1.55v, and 2.4 GHz. This CPU is supposed to run at 2.4 GHz, and 1.5v. Recently, I've had to clock it at 2.2 GHz, and 1.6v to be stable.

I've run the CPU at 100% load 24/7, and stock Vcore since I bought it. It's also running very cool with an oversized HSF.

I've never had a CPU degrade this fast. The last CPU that I had do this was an AXP 2000+, but that took about four years of 24/7 operation.
 

CTho9305

Elite Member
Jul 26, 2000
9,214
1
81
Clean the dust out of the heatsink and/or case fans?
(You didn't provide actual temp numbers, so for now I'm assuming that you just assume they're good because they were in the past)
 

dflynchimp

Senior member
Apr 11, 2007
468
0
71
I'm having trouble deciphering your problem here. Is it that you're getting system shut offs or blue screen mem-dumps + auto restarts?
My last system was a Pentium D 805 OC'd to 3.8GHz and that started giving me problems after an year or so (might've been my boot-leg copy of XP also...good thing I've gone vista now). cpus degrade exponentially faster when OC'd, although this is tempered by good cooling.

Also, if you're running an A64 3400+ it must be a socket 754 (3400+ was 754 exclusive if I remember right). I can't be sure but I think those don't OC as well as the socket 939 variants.
 

darkfalz

Member
Jul 29, 2007
181
0
76
Your CPU is made by AMD, that is the problem. AMD chips are essentially overclocked out of the box.

YGPM

Anandtech Moderator - KP
 

Amaroque

Platinum Member
Jan 2, 2005
2,178
0
0
Originally posted by: CTho9305
Clean the dust out of the heatsink and/or case fans?
(You didn't provide actual temp numbers, so for now I'm assuming that you just assume they're good because they were in the past)

Blew out every fan in the system (including the PSU). The CPU is idleing at at 30c rite now, as I had to take it off DC.

Originally posted by: dflynchimp
I'm having trouble deciphering your problem here. Is it that you're getting system shut offs or blue screen mem-dumps + auto restarts?
My last system was a Pentium D 805 OC'd to 3.8GHz and that started giving me problems after an year or so (might've been my boot-leg copy of XP also...good thing I've gone vista now). cpus degrade exponentially faster when OC'd, although this is tempered by good cooling.

Also, if you're running an A64 3400+ it must be a socket 754 (3400+ was 754 exclusive if I remember right). I can't be sure but I think those don't OC as well as the socket 939 variants.

I'm getting shutoffs, blue screens, and mem dumps. Your are correct the system is a S754 A64. Tried different RAM, and a different Vid card too. It might be the PSU. I'll try a spare 550w TP2 tommrow.

That's the only S754 board that I have, so bench testing the CPU with a different board isn't an option rite now.
 

Quiksilver

Diamond Member
Jul 3, 2005
4,725
0
71
Yarr... Edited for reasons.

I kind of does sound like the cpu is beginning to fade, but it could also be the motherboard.
 

dflynchimp

Senior member
Apr 11, 2007
468
0
71
I would slap you for fanboi-ism, but that would be a waste of energy.

Overclocking and Underclocking only apply to aftermarket tweaking by the consumer.

Technically speaking there is no such thing as "overclocking" or "underclocking" before a processer hits the market. There's only "maximum clock threshold" for a particular revision of a cpu architecture and/or core. CPU manufacturers basically take the same chip and brand them differently (usually sorted by quality) with different base clock speeds and labels, but otherwise a A64 3200+ and a 3500+ share the same Venice core, and are thus physically identical; the only thing that's changed is the CPU-ID.

If you claim AMD cpus to be OC'd right out of the box than the same could be said for the old Pentium 4's and D's, as they were hitting the "maximum clock threshold" of around 4GHz on air. Die shrinks only helped that a teensy bit. In fact, you could say that all Extreme Edition or FX chips are pre-overclocked cpus, since they inevitably share the same cores with lower numbered parts.

mmkay, back to topic:

Amaroque, blue screens and mem dumps are usually due to undervolting. shut-offs are due to overheating. If you're getting both then you probably need to back down the clockspeed. That or move on to better cooling if you want to keep the clockspeed up and just increase the voltage for stability.
 

Idontcare

Elite Member
Oct 10, 1999
21,110
59
91
Originally posted by: CTho9305
Clean the dust out of the heatsink and/or case fans?
(You didn't provide actual temp numbers, so for now I'm assuming that you just assume they're good because they were in the past)

Good points, to which I'll add that it may be time to clean off your thermal paste and redo it.
 

darkfalz

Member
Jul 29, 2007
181
0
76
His temps are fine, his CPU really is "degrading" - the effects of long term overclocking (even though he's not overclocking, like I said, his AMD chip is essentially factory overclocked). I had a P3 700 MHz that ran overclocked at 933 MHz for 5 years, but then just started acting dodgy - dropping it down to 700 and it was fine again, but it would no longer hold the clock it had held for 5 years.

It happens.
 

dflynchimp

Senior member
Apr 11, 2007
468
0
71
I've said it before and I'm saying it again, it's not factory overclocked. There is no such thing.
 

The Sly Syl

Senior member
Jun 3, 2005
277
0
0
I've been using an actual overclocked graphics card (350/1.0 is standard, mines been over 410/1.10) for almost 3 years straight now. It's still entirely stable.

I had an AMD 64 3000+ running at 2.4ish ghz for 1.5 years, and it was also entirely stable. It never degraded. (Just got fried when my motherboard did)

I think this guy just needs to clean out his case. The processer shouldn't be on the kaputz yet. I still wouldn't try to *start* overclocking a processor this late in the game however.
 

dflynchimp

Senior member
Apr 11, 2007
468
0
71
that last part is true, OCing a proc after it's settled into a clock-speed for some time is less desirable. I'd also venture that s754 A64's were less able overclockers, as they were the first of the A64 line. s939 was much more refined (plus supported dual channel memory)
 

nyker96

Diamond Member
Apr 19, 2005
5,630
2
81
check to see if the HSF is dirty I had similar issue with an older machine while back. When you clean the HSF and reinstall the problem might go away like mine cause over time HSF gets dirty you no longer got the prestine coolings when it's new. Thus you had to clock you CPU down becuase of this.
 

GuitarDaddy

Lifer
Nov 9, 2004
11,465
1
0
I doubt very much that the CPU is your problem. I would suspect two areas:
1. Power supply is dying (check stability of the rails with a multimeter)
2. Motherboard is dying (check for bulging or leaking caps)
 

Amaroque

Platinum Member
Jan 2, 2005
2,178
0
0
Like I've previously stated, temps are good at ~30c idle.

I think I figured out what the problem is. Somehow I overlooked this... The Vcore is jumping all over the place from 1.3v to 1.7v. All of the other voltages are rock stable however.

So this now leads me to believe that it's either the PSU, or MB.
 

nick1985

Lifer
Dec 29, 2002
27,153
6
81
Originally posted by: darkfalz
Your CPU is made by AMD, that is the problem. AMD chips are essentially overclocked out of the box.

YGPM

Anandtech Moderator - KP


Fanboys like you suck.
 

heymrdj

Diamond Member
May 28, 2007
3,999
63
91
Could be the PSU, but when this happened to me it ended up being the VRM's on the motherboard, so that ended up being the part replaced. Check your motherboard for swelled or blown out capacitors.
 

yelo333

Senior member
Dec 13, 2003
990
0
71
Originally posted by: Amaroque
It started about two weeks ago, when I had to set the Vcore to 1.55v, and 2.4 GHz. This CPU is supposed to run at 2.4 GHz, and 1.5v. Recently, I've had to clock it at 2.2 GHz, and 1.6v to be stable.

Maybe it's something simpler...Something similar to this was happening to me with me my (admittedly still current, don't game though) AXP 2000+.

My story (can be skipped :)): I noticed that turning down the clockspeed & upping the vcore was helping stability. OC to 1.9 ghz/1.65v and the thing would lock up a few times a day. Never failed p95/orthos/superPI/memtest/etc. Drop it to 1.8/1.575v and it would run for a couple days at a time. Eventually, I gave up and ran it at the stock 1.67ghz/1.6V, even overvolting to ~1.65v, but it would still lock up once a week or so. I thought for sure it must be the mobo/CPU just getting old & tired. But,...

It turns out disabling fast writes in the BIOS solved the problem for me.

So, I'd recommend that you try disabling any options you can it the bios, remove unnecessary PCI cards, etc. As GuitarDaddy said, I'd bet it's something besides the CPU.
 

nyker96

Diamond Member
Apr 19, 2005
5,630
2
81
I think this does sound like PSU/mb is going down. try another PSU if you got one, see if that can fix hopping vcore, if so your mb is probably dying.
 

stevty2889

Diamond Member
Dec 13, 2003
7,036
8
81
It's more likely the motherboard than the PSU, since the CPU voltage is regulated at the motherboard(although it's still possible for an unstable PSU to cause it as well).
 

soonerproud

Golden Member
Jun 30, 2007
1,874
0
0
Originally posted by: GuitarDaddy
I doubt very much that the CPU is your problem. I would suspect two areas:
1. Power supply is dying (check stability of the rails with a multimeter)
2. Motherboard is dying (check for bulging or leaking caps)

This has been my experience in cases like yours. Check the power supply first and if that does not do it, then it is probably the motherboard. In that case take the PC to a local shop that can properly test the motherboard to see if it is dying.
 

Amaroque

Platinum Member
Jan 2, 2005
2,178
0
0
Originally posted by: soonerproud
Originally posted by: GuitarDaddy
I doubt very much that the CPU is your problem. I would suspect two areas:
1. Power supply is dying (check stability of the rails with a multimeter)
2. Motherboard is dying (check for bulging or leaking caps)

This has been my experience in cases like yours. Check the power supply first and if that does not do it, then it is probably the motherboard. In that case take the PC to a local shop that can properly test the motherboard to see if it is dying.

Err, I worked in a local computer shop for a few years. 'nuff said.

I can just shoot myself, as I sold a whole box of PSU's about three months ago (I was downgrading my fleet). Looked over the MB with a fine tooth comb, no bad caps that I can see. Although that doesn't rull out the MB.

I'll have to borrow a known good PSU from one of my other systems to rule out the PSU.

I'd like to avoid rebuilding that system, as it's my wifes computer, and she pretty much only surfs. No reason to put a AX2, C2D, or C2Q in that system. ;)
 

Amaroque

Platinum Member
Jan 2, 2005
2,178
0
0
Update: I pulled the 850W TP Quattro 850 out of my Q6600 and put it in the aforementioned machine, and I still get Vcore probs. So I Guess It's The MB.

Thanks all for your input. (except the fanboys) ;)