Finally broke down & purchased I7 2600k setup

[mrjoltcola]

Actually I like 42 multiplier best, but I'm a nerd.

I'm no pro at OC, and I'm very conservative (I run my business off my PC so can't have junk getting corrupted just for 200-300 mhz). At 4.7Ghz my PC appears to run fine, regular apps act right, but weird things happen that can even emulate software bugs. Visual Studio starts throwing odd compile errors, my Kingston 6Gb/s SATA3 drive starts acting like it doesn't have a partition. All based on overclocking! Bump down to 4.4 and its all sane again. Nothing outright crashing, though. This is what makes me think a _lot_ of folks reporting 4.7+ stable may be cheating.

Anyway, back on topic...

For CPU temp, 80C would seem too high for me (not same as core temps) based on my reading. Note, I didn't say it shouldn't run, I see no problems at that temp, I'm just going by the Intel spec. 72.6C Tcase (measured on the CPU die) is as high as I want to go, though you can be sure, the real number is higher, their "consumer" number is to ensure plenty of safety room. http://ark.intel.com/products/52214/Intel-Core-i7-2600K-Processor-(8M-Cache-3_40-GHz)

Here are my temps today with the H60 cooler. Just testing different frequencies to see what I can run while staying under the official 72.6C. Voltage set on auto.
4.0Ghz / 1.245v Vcore observed
Idle Temp - 33C/91F
Load Temp - 67C/152F - Prime95 large FFT 4 thread - measured at 15 minutes
4.1Ghz / 1.250v Vcore observed
Load Temp - 68C/154F - Prime95 large FFT 4 thread - measured at 15 minutes
4.2Ghz / 1.258v Vcore observed
Load Temp - 69C/156F - Prime95 large FFT 4 thread - measured at 25 minutes
4.3Ghz / forgot to check voltage
Load Temp - 70C/158F - Prime95 large FFT 4 thread - measured at 35 minutes

I didn't go higher today because I ran this in the background while actually doing real work, so didn't want to crash anything.

 

[lol123]

I'm no pro at OC, and I'm very conservative (I run my business off my PC so can't have junk getting corrupted just for 200-300 mhz). At 4.7Ghz my PC appears to run fine, regular apps act right, but weird things happen that can even emulate software bugs. Visual Studio starts throwing odd compile errors, my Kingston 6Gb/s SATA3 drive starts acting like it doesn't have a partition. All based on overclocking! Bump down to 4.4 and its all sane again. Nothing outright crashing, though. This is what makes me think a _lot_ of folks reporting 4.7+ stable may be cheating.

Just out of curiosity, is there any way you can be sure that you are not encountering (possibly silent) data corruption at the overclocks you are running? We had an interesting exchange here a while ago about the effects of overclocking, specifically the higher temperatures, on chip longevity where Idontcare said that an increase of 10°C cuts life time in half (I forgot to thank you for that informative post, Idontcare). Is there any similar, known correlation between temperatures and/or clock frequency and data corruption?

 

[mrjoltcola]

Just out of curiosity, is there any way you can be sure that you are not encountering (possibly silent) data corruption at the overclocks you are running? We had an interesting exchange here a while ago about the effects of overclocking, specifically the higher temperatures, on chip longevity where Idontcare said that an increase of 10°C cuts life time in half (I forgot to thank you for that informative post, Idontcare). Is there any similar, known correlation between temperatures and/or clock frequency and data corruption?

Exactly, you can't be sure on everything, though some applications include data checksums (databases), and I keep important stuff under source control so I know for sure if a bit changes. But then again, this really applies even to stock frequencies. We get a bit error every so often when using non-ECC RAM, and the same would probably apply to OC. So I think you just have to stress test. A rig that stress tests with multiple uilities for 24 hours should be stable for work.

This why I first install everything (OS and software) under stock clock, then full backup, then start tinkering. Once I hit instability, I don't just back down 1 ratio, I back down 3-4.

I approach overclocking like tuning my daily driver car with a new K&N air filter and muffler, not like boring the cylinders and installing a new chip. There is some safe headroom. You can do some light things to it if you improve your cooling, that shouldn't put anything out of spec. The stock coolers are not very good. So I put water cooler on and mine runs cooler at 4.2Ghz than people I see with the stock cooler at stock speed.

But at best, I'm just a software guy with a half-finished Electrical Engineering degree, mouthing off about hardware like an armchair quarterback. I'd be interested to hear what the real hardware engineers say about this topic.

 

[choliscott]

Well I finally got my board & cpu installed. I installed all 16 gigs of ram & noticed the bios would "lockup" after a short period of time (mouse wouldn't move nor would keyboard would respond), so I had to make one adjustment at a time & save it & go back.

When I went to reinstall windows, it would do the inital loading of the files & then when it showed the Windows logo, it would freeze. I tried this twice until I decided to remove sticks of memory. At that point, there was no problems with the bios or installing Windows.

Did a initial stress test for about 1/2 hr or so (just to make sure everything seemed to be working). Worked fine. Next I bumped it up to 4.6ghz at 1.36v, which that also worked on stress test, but got up to 81c after about 5 minutes, so I stopped it & lowered it back to 4.5ghz @ at 1.36v & the highest it hit was 78c. After a couple hours (as I write this & should be asleep), I stopped it so I can lower the voltage down to see if 4.5 ghz still works.

The one question i have is, CPU-Z does not show the correct speed & even when I was running prime, it acted like the highest speed set was 3.6ghz. I had to use Everest for it to show the speed that I have set. Is this normal with CPU-Z ?

Thanks

 

[choliscott]

Well after doing some testing, I have settled for 4.5 Ghz at 1.28v & high LLC, which is what I had my mind set on before I had recieved the chip.

HW Monitor showed it to be in the mid 70's on stress & low 40's on idle

I tried benching it at 4.6ghz with 1.35v, but after about 30 seconds the computer would freeze (I could still see everything) & about a minute later I would get the BSOD.

Thanks everyone!!

 

[Idontcare]

Is there any similar, known correlation between temperatures and/or clock frequency and data corruption?

You can never be certain, obviously, but the way you would go about preparing yourself to be able to speak to the probabilities of silent data corruption is to generate data in the test conditions that forced detectable errors and then project that trend using a Weibull Analysis.

 

[mrjoltcola]

Well after doing some testing, I have settled for 4.5 Ghz at 1.28v & high LLC, which is what I had my mind set on before I had recieved the chip.

HW Monitor showed it to be in the mid 70's on stress & low 40's on idle

I tried benching it at 4.6ghz with 1.35v, but after about 30 seconds the computer would freeze (I could still see everything) & about a minute later I would get the BSOD.

Thanks everyone!!

Congrats.

One more person with higher clocks than me.

I wonder if the fact that I have 4 DRAM chips could be affecting my OC?

I think its time for me to try another chip. Ordering another 2600K and an ASUS M/B with an H80 cooler while I'm at it. If I do better the second time, I'll sell my current combo as 4.3Ghz "cool and stable"

 

[choliscott]

When I had 4 DRAM chips installed, the bios would keep "locking up", also when I tried to install windows, it wouldn't get past the initial Logo screen to show it starting to copy files.

Once I removed 2 sticks, I had no issues.

Doesn't hurt to try removing 2 sticks & see what kind of result that you get.

Congrats.

One more person with higher clocks than me.

I wonder if the fact that I have 4 DRAM chips could be affecting my OC?

I think its time for me to try another chip. Ordering another 2600K and an ASUS M/B with an H80 cooler while I'm at it. If I do better the second time, I'll sell my current combo as 4.3Ghz "cool and stable"

 

[Idontcare]

Congrats.

One more person with higher clocks than me.

I wonder if the fact that I have 4 DRAM chips could be affecting my OC?

I think its time for me to try another chip. Ordering another 2600K and an ASUS M/B with an H80 cooler while I'm at it. If I do better the second time, I'll sell my current combo as 4.3Ghz "cool and stable"

I have 4 dimms populated on my 2600K rig, 16GB, running DDR3-1866 10-10-10-28-T1 on 1.5V.

Same rig, albeit with an H100 versus your H60, and I'm LinX stable at 4.9GHz, have clocked to 5.2GHz for just playing around with benches (haven't bothered to tweak the OC for LinX stable at those speeds yet, will get to it though).

Maybe you just got an unlucky 2600K?

 

[OVerLoRDI]

I have 4 dimms populated on my 2600K rig, 16GB, running DDR3-1866 10-10-10-28-T1 on 1.5V.

Same rig, albeit with an H100 versus your H60, and I'm LinX stable at 4.9GHz, have clocked to 5.2GHz for just playing around with benches (haven't bothered to tweak the OC for LinX stable at those speeds yet, will get to it though).

Maybe you just got an unlucky 2600K?

What is your 24/7 stable clock and voltage?

 

[Idontcare]

What is your 24/7 stable clock and voltage?

I don't have one yet, still playing around. It's going to be somewhere around 4.8GHz I suspect. Maybe higher.

 

[mrjoltcola]

Maybe you just got an unlucky 2600K?

I think that is exactly it.

Heck, companies are selling 4.6Ghz stable tested kits online using the H60, so I fully expect to get that myself.

When I get the ASUS Z68 and 2nd 2600K I'll try the current 2600K in the ASUS M/B just to see if things differ. May also compare some combinations, like H60 vs. H80.

I've also ordered some thermal compound to alleviate any possibilities with the H60.

Removed 2 of the DRAMs, no difference.

 

[OVerLoRDI]

I don't have one yet, still playing around. It's going to be somewhere around 4.8GHz I suspect. Maybe higher.

Thanks, I know you were working on this and were posting about it somewhere Just ever curious into your temps and voltage, probably looking to push my system higher with the next gen cards this christmas/january