Long-term stability when overclocking

[bebop06]

Hey all, I just have a quick question about long-term stability when overclocking. If you kept the core voltage close to stock and settled for a moderate overclock, would the overclock be pretty much as stable over two or so years as a processor running at stock speeds? I'm about to buy either an Opterton 165 (and OC it) or an Athlon 4400+ (and run it at stock); I've read a bunch of threads about choosing between the two, and obviously it came down to whether you wanted to OC or not.

The price is nearly the same: the Opteron 165 is $325 plus an XP or SI-120 heatsink ($55) plus 2x1GB G.SKILL DDR500 ($192) = $572, vs. the 4400+ at $460, stock heatsink, and 2x1GB G.SKILL DDR400 ($143) = $603. The question is what sort of a really stable, long-term overclock I could expect to run with the opteron (and I know, as with all OCing, it depends on the chip). I'm going to be writing my thesis on this computer, and I don't want it crashing randomly or corrupting the HDD or anything. And I'd like to be able to run my computer 24/7. With a moderate overclock (that's 24 hours+ dual Prime stable), could I expect the opteron to be essentially as safe and stable as the 4400+ at stock? And could I expect such a moderate overclock to be >2.2 GHz? If a moderately oc'ed opteron wouldn't be faster than the 4400+, I would rather spend the extra $30 for the x2.

I've searched the forum (and on Google) for information about this kind of long-term stability, but I couldn't find anything. Thanks for the help.

 

[bebop06]

Does anyone who overclocks moderately have any information on how stable their system is (specifically if they can run it 24/7 without problems)? I have to make my decision tonight so that I can build my rig over break. If I could probably moderately overclock the 4400 close to where I could moderately oc the opteron (2.4/2.5?), then the 4400 seems the better choice, since I could run it at stock for at least a year or so without being held back by the cpu in games (I'm getting an X1900XT video card). Any help much appreciated!

 

[fixxxer0]

I have my Winchester 3500 OC'd for a little over two years now at the maximum my chip could take without failing prime. (2500mhz - not a lot compared to others getting 2.6-7 on air). The vCore was only bumped up .025v throughout the whole duration of this two years 24/7 on OC.

Just recently, a week ago actually, I started getting some random crashes - i think temperature related due to my HSF being gunked with dust but the temps werent crazy (46C underload). I cleaned it off - reseated the HSF, and temps went back to 39C underload after I cut voltage back to stock. I had to lower the clock however about 60mhz for it to be stable again. And its been running fine all week now w/ Prime95 on for at least 48hrs (non-continuous).

 

[Proletariat]

I've had my Athlon XP-M oc'ed from 1.8 Ghz to 2.2 Ghz for about 2 years.

I've had zero processor related problems.

Stable as a rock in my book. Haven't even done a reformat yet.

 

[furballi]

Keep Vcore under 1.55. Also limit max CPU temp to 52C. Test the system to failure, then back-off about 100MHz CPU core speed.

I would recommend the use of Cool n Quiet. With Cool n Quiet, Vcore is approximately 1.2V most of the time (when CPU is idling or working at moderate load).

 

[F1shF4t]

Only way to make sure is to run it for a few months, i'm sure that after a few years u will have to lower the clock a bit as the chip does deteriate over time, but otherwise it should be stable for quite a long time as long as u keep vcore (more voltage means the chip deteriorates faster) down and the clock 50-100mhz bellow the absolute max ur chip can do.

 

[TrevorRC]

There is a max TDP rating/etc on the CPU... it'll tell you the leakage of your CPU and the max t case (Max Thermal Rating, per AMD).
So long as you don't exceed that temperature, you should be fine..
--Trevor

 

[bebop06]

Thanks so much for all your replies. You've convinced me to go with the Opteron 165. Just put in an order at TankGuys (stepping CCBWE 0551 WPMW--anyone got this one? I saw two people on an earlier thread got this stepping up to 2.5 at close to stock volts). Everything should come early next week, and I'll let you guys know how everything goes.

 

[3chordcharlie]

If you keep the chip under the highest rated speed that was released 'stock' for that core revision, and don't exceed the voltage for that higher speed rating, then if it comes up prime-stable, etc, you shouldn't need to worry about long-term stability anymore than you would buying the faster chip to start with.

 

[robertk2012]

Originally posted by: 3chordcharlie
If you keep the chip under the highest rated speed that was released 'stock' for that core revision, and don't exceed the voltage for that higher speed rating, then if it comes up prime-stable, etc, you shouldn't need to worry about long-term stability anymore than you would buying the faster chip to start with.

not true. The cpus are binned by performance. Some cpus can only perform at there rating or just a little above. Most have headroom but its kind of luck of the draw.

 

[3chordcharlie]

Originally posted by: robertk2012

Originally posted by: 3chordcharlie
If you keep the chip under the highest rated speed that was released 'stock' for that core revision, and don't exceed the voltage for that higher speed rating, then if it comes up prime-stable, etc, you shouldn't need to worry about long-term stability anymore than you would buying the faster chip to start with.

not true. The cpus are binned by performance. Some cpus can only perform at there rating or just a little above. Most have headroom but its kind of luck of the draw.

I didn't say it's guaranteed to reach that level, but if it does, and it tests stable, you shouldn't have to worry about degraded performance over time @stock volts. With clockspeed, or especially voltage above 'stock' for the fastest processor based on the revision you have, you are essentially 'creating' a part specification that the OEM was unwilling to sell under warranty.

 

[robertk2012]

Originally posted by: 3chordcharlie

Originally posted by: robertk2012

Originally posted by: 3chordcharlie
If you keep the chip under the highest rated speed that was released 'stock' for that core revision, and don't exceed the voltage for that higher speed rating, then if it comes up prime-stable, etc, you shouldn't need to worry about long-term stability anymore than you would buying the faster chip to start with.

not true. The cpus are binned by performance. Some cpus can only perform at there rating or just a little above. Most have headroom but its kind of luck of the draw.

I didn't say it's guaranteed to reach that level, but if it does, and it tests stable, you shouldn't have to worry about degraded performance over time @stock volts. With clockspeed, or especially voltage above 'stock' for the fastest processor based on the revision you have, you are essentially 'creating' a part specification that the OEM was unwilling to sell under warranty.

no more than overclocking above the highest stock limit.

 

[3chordcharlie]

Originally posted by: robertk2012

Originally posted by: 3chordcharlie

Originally posted by: robertk2012

Originally posted by: 3chordcharlie
If you keep the chip under the highest rated speed that was released 'stock' for that core revision, and don't exceed the voltage for that higher speed rating, then if it comes up prime-stable, etc, you shouldn't need to worry about long-term stability anymore than you would buying the faster chip to start with.

not true. The cpus are binned by performance. Some cpus can only perform at there rating or just a little above. Most have headroom but its kind of luck of the draw.

I didn't say it's guaranteed to reach that level, but if it does, and it tests stable, you shouldn't have to worry about degraded performance over time @stock volts. With clockspeed, or especially voltage above 'stock' for the fastest processor based on the revision you have, you are essentially 'creating' a part specification that the OEM was unwilling to sell under warranty.

no more than overclocking above the highest stock limit.

The question reads as 'is there a way to improve the chances that a stable OC on a new chip will remain stable'.

You can almost certainly go farther than I put here and still be pretty safe, but it is one highly restrictive set of guidelines for doing improving your odds.