Would it be advisable to run a 3930K at 1.5V 24/7? I've managed to get it to pass Intel Burn Test at 4.9 GHz @ 1.5 V on the 'Very High' setting (5 passes).
Printable View
Would it be advisable to run a 3930K at 1.5V 24/7? I've managed to get it to pass Intel Burn Test at 4.9 GHz @ 1.5 V on the 'Very High' setting (5 passes).
Cooling?
No with any type of cooling. Sandy Bridge is in my experience is silicon that will degrade easily. Even though you can do that speed with that voltage atm you will lose that ability if you run it like that 24/7 in a matter of weeks.Quote:
Originally Posted by alpha754293
Corsair H80.Quote:
Originally Posted by felix_w
At 4.8 GHz @ 1.44 V (idle, 1.432 V @ load) - I've topped it out at 74 C using Intel Burn Test. Running Folding@Home right now, it's only topping out at 61 C.
H80 is a low end "water" cooling for SB-E. I couldn't get a 3930K or 3960X over 4.5 stable with a H100 which is a little better than H80. And by stability I mean at least 8 hours of Prime95 27.7. I was already too close to temperature throttling limit. With a golden chip, better results could be achieved, but 4.9 @ 1.5 V doesn't sound too promising.
Btw wrong subforum. :)
It overlaps Intel/Overclocking/Motherboards and a few others.Quote:
Originally Posted by xarot
So unless you want to start creating subforums based on keywords, else; you can move it to wherever you like. Also keep in mind that 5 GHz does NOT appear to be a common OC that a 3930K can hit.
I'm currently running at 4.8 GHz @ 1.44 V wih an H80 at ~ 410 W and it has no problems running LS-DYNA NCAC benchmark models for 30 hours. And that's WITHOUT using *CONTROL_MPP_DECOMPOSITION (which typically runs even faster).
Well it depends how you see it, but on XS, maybe not common a OC but neither much extreme yet in that overclock. The 4.5 & 5 GHz LinX stable thread is in the Intel section as well with people posting 5 GHz LinX 20 loops stable results.
But back to your original question, 1,4 V is generally considered the upper "safe" limit in many threads I have seen. I would not go much higher than that with a H80, maybe max 1,42 V. There was a guy that had used 1,57 V (if I recall right) by using AUTO settings for some time on air cooling and his processor's pin pads burned and later his mobo died. So be careful. Obviously the program you run doesn't tell me anything, but if it runs fine, it probably is the only thing that matters. I've used LinX to compary GFlops on SB-E overclocks since in many cases lower OC can result in higher GFlops than an unstable OC. Good luck.
I ended up backing off the OC back to 4.5 GHz. 4.8 GHz - I was about to do short runs with, but it's been having a REALLY hard time with a 30-hour run and it's been requiring quite a high voltage bump so I've decided to just back it off to 4.5 GHz and stay there while I finish the last of the 30-hour runs.Quote:
Originally Posted by xarot
It doesn't appear to be a thermal issue that's limiting/preventing it from being stable at that OC, but I was already hitting 1.45 V (according to CPU-Z at idle) that I figured the extra 0.3 GHz wasn't worth the risk or the power. Bummer. But oh well. 30 hour it is then. It's still three TIMES faster than my old quad dual-core Opteron system (so now a 92-hour run finishes in 30-hours).
+1 for pretty much everything above.
~1.4 is the max you can push into your Sandy chip 24/7 without having to worry 1 single second. Passed this point, its all up to you... Ive seen people running 3930k at over 1.5v 24/7 but i advised them they could degrade the chip way faster than normal at that Vcore.
In the end, IMO, your better to find a good, stable, fast enough OC for your 24/7 use that stay within safe limit for the chip. I run relax OC as for my 24/7, currently a little 4.2ghz on my 3820, but i could run this chip 4.5-4.6ghz without pushing to much volts. 4.2 is enough for my gaming use, so why running faster, louder, hotter ? ... My opinion !
I agree with what you're saying. But at 4.5 GHz - as it stands right now, my crash sims take 30-31 hours to run. And that's just with a standard benchark model. If I were to be running my own model, it will likely be more detailed than that and therefore; would take even longer to run.Quote:
Originally Posted by Boulard83
And then for my current combustion CFD models, the last one ran in 21 hours. But again, that'll increase as the models get more and more complicated/complex, so the difference between stock and OC may not really be much in the grand scheme of things, but being able to finish the runs just that much faster is still better overall. I dunno. I might actually have to build a SWaP plot. *sigh*
get yourself a dual socket, more cores, more threads, more power !
I'm not totally sure I agree with anyone here. Ive ran my 3930k at 1.44 volts @ 4.8ghz for like........ 6 or 7 months daily gaming use with the occasional 1.5-1.52v overclocking runs. My 3930k has not degraded one bit. On the contrary my chip has been hitting higher and higher clocks over time to my current top clock of 5.2ghz I used to make this 3dmark 11 run http://hwbot.org/submission/2341412_...80_13820_marks
When I first got this chip 4.8ghz was the VERY maximum, and over time 4.9 became available, then 5ghz possible, then 5.1, and recently Ive been able to make 5.2ghz runs. Im sure its not entirely stable but we are talking about damage to silicone.
Well, the quad dual socket mini blade-in-a-2U (Supermicro Twin2) the chassis alone is $3500. Each of the Xeon E5-2670 that I was looking at cost the sum of my 3930K that I've now OC'd. Granted, that system will ultimately be more POWER efficient, but I'm not sure if it would necessarily be faster. (More cores = more intercore communication and moving data around doesn't actually do any useful, computational work).Quote:
Originally Posted by Boulard83
I still plan on getting that (total for that build for the first blade is around $6500, and by the time all's said and done, it'll run around $22k). But in the interim, I'm able to cut my simulation time down by 66%. (It was taking me too to save up for that).
Were you running it 24/7 HARD though? I think that's the difference. I managed to run Intel Burn Test at 4.8 GHz, but the moment that I actually try to run my simulations on it; it failed. Repeatedly. And it just kept needing higher and higher voltages to attempt to stabilize it. I think that I stopped trying to push it to 4.8 GHz at 1.48 V and it still failed to run the simulation. That and I think that the power required to push it to 4.8 GHz - the extra power isn't work the 300 MHz and the time that it would save on account of it. (After 30 hours, what's a few more minutes?).Quote:
Originally Posted by Ace123
It's no 3930k but I've ran my SB i7-2600k @ 5.0GHz, 24/7 stable at 1.5v for a little under a year (Fired up the system in February). No matter how hard I have tried, 5.1GHz is not stable with any BIOS so I am stuck with a chip limitation. It will boot to Windows at 1.45v but is prone to crashing during heavy loads.
My computer is used mostly for gaming.
sorry to bump this old thread, but I am hoping 1 year later people might have more experience with these chips. I recently got around to upgrading my old x58 to x79 and now my 3930k is @ 5ghz with 1.5v and I am curious if this is doable 24/7.
ive had my 3930k running 1.45v 4.7GHz since it was bought(dec 2011 if im not mistaken), hasnt degraded at allQuote:
Originally Posted by b0bd0le
hope that helps