Guys, I'm in need of your wisdom.. 2600k and Z77 OCF board overclocking...

[phill]

@revin - Thanks for the post and mention, I'll see if I can have a look at that tomorrow.. Too tired for it tonight lol

I am pretty sure that IBT would require a bit more volts than say something like wPrime or Cinebench R11.5 or R15 to run but I'm not so overly bothered about the volts and speeds this CPU gives me. It does everything I ask of it at the speed I use it at and with the crunching as well, I'm sure it's a lot more efficient than my 920 D0's would be... lol They are truly monster power CPUs lol

 

[revin]

That screenie is of XTU, Intel Tuning Utility, I Don't do Intel Burn Test, or them other crazy loading test for the just because. If your system will play and do everything you ask even if it's crunching ect, then good. I Don't believe that running Prime, ect for hours justifies a "stable" OC.
Just wanted to give you an example of some changes and how sometimes just a basic change can be all that is needed. I had spent hours bumping System Agent, PCH and on and on. Then after I seen a post that Dave had about doing something different that caused a stir, I thought wait, start over and go smaller.
So in the end the best result was to allow headroom for the Turbo Power limits, and Core Current, then was able to use a small 40 mv voltage bump. All that raising this and that was for naught.
But it did pass the XTU[IBT] Stress testing for 15 minutes every time I would run it. TBH about 5-6 min was all that was needed to test a OC.
Obviously as I find first hand again all systems are different but hopefully it will give another look from a different angle so to speak that less can be more

 

[phill]

I remember back when benching that I'd never use XTU on ANY CPU that you would want to keep as even running it stock, proved to kill CPU's.. It was a massive problem! 'The good olde days' of the Z170 platform and the i3 6320 for it's dual core goodness, god everyone went nuts over that! I know my mate who borrowed my first Z170 OCF before it died under LN2, the CPU would be only running stock but after a few runs (ok, maybe a bit more than a few) of XTU it just killed the CPU.. It would barely run stock speeds with voltage bumps by the end of it.. I think it was generally knackered!!

I've used various programs over the years for testing, some Prime, some Cancer Crunching, even wPrime 1024m at one point and then Cinebench R15, a little tweak in voltage and it was golden! I also found the Asus program that tests multi core and such, that's a good tester for stability Got some good scores on that with my 5960X and 7970... But I digress...

This 2600k CPU is more than capable for what I need right now, until I get the 5960X rig I have here up and running, this will do till then I have a good 4770k (5Ghz @ 1.2vcore I believe or so) but that needs some better thermal paste as it's de-lidded and the temps are massively out of whack.. Difference of like 20C under another H100i I have.. but I've given up with that one for the moment.. Not a spare penny to my name at the moment with my new job, so until I get a pay rise from it, I won't be spending even on thermal paste!!

 

[revin]

I remember back when benching that I'd never use XTU on ANY CPU that you would want to keep as even running it stock, proved to kill CPU's.. It was a massive problem!

First Ever heard of such a thing, @cadaveca You have any knowledge of this ?
Dave has used it vastly more than I have, and I been using XTU since I got my 2600K, at least in January 2012 , and that is how I was able to get the 5.2Ghz OC on mine.

Intel XTU is more for system monitoring and adjusting than for testing.
There is a CPU test and a Ram test included, but it's not like OCCT or other crazy suites.
Albeit they also did release a Benchmark for it to use @Hbot scoring.
You have XTU confused with something else, there is 10's of Thousands of score's with XTU benchmark.
But I think you Still missed the point of the post's I made to give you information relevant to your 2600K @ 5Ghz, anyway good luck with your Sandy !

 

[phill]

There's a test that you can run within XTU that does kill the CPU's eventually as I've had the experience of it! Under the benchmarking tab.. If I remember rightly with the i3 6320's, everyone was chasing the 740 numbers... Then 741 when B die Sammy ram came out and then 742 and 743 where hit but just after countless times trying...

It was back when Intel had locked down the i3's from overclocking as the motherboards bios's were released that could have overclocked them.. Here's the link to the XTU page on Hwbot -

http://hwbot.org/submission/3057206_phill_2_xtu_core_i3_6320_739_marks
http://hwbot.org/benchmark/xtu/rankings?hardwareTypeId=processor_4256&cores=2#start=0#interval=20

XTU had a bad name for a while, not sure if that's changed much as I've been out the overclocking game for a while.. One of the top benchers refuses to run it because it's stupidly easy to gain points.. There was a load of press about it across the HWBot forum if I recall..

But yes, when I've finished backing up my home server and the misses isn't here, I'll be taking a look at seeing if I can improve on the 2600k clock

 

[Regeneration]

IBT (Linpack) is one of the best stress testing software to determine core stability. Just make sure to set threads manually and no more then 9216MB for RAM. 8 hours of IBT is like 24 hours of Prime95.

 

[cadaveca]

There's a test that you can run within XTU that does kill the CPU's eventually as I've had the experience of it! Under the benchmarking tab.. If I remember rightly with the i3 6320's, everyone was chasing the 740 numbers... Then 741 when B die Sammy ram came out and then 742 and 743 where hit but just after countless times trying...

It was back when Intel had locked down the i3's from overclocking as the motherboards bios's were released that could have overclocked them.. Here's the link to the XTU page on Hwbot -

http://hwbot.org/submission/3057206_phill_2_xtu_core_i3_6320_739_marks
http://hwbot.org/benchmark/xtu/rankings?hardwareTypeId=processor_4256&cores=2#start=0#interval=20

XTU had a bad name for a while, not sure if that's changed much as I've been out the overclocking game for a while.. One of the top benchers refuses to run it because it's stupidly easy to gain points.. There was a load of press about it across the HWBot forum if I recall..

But yes, when I've finished backing up my home server and the misses isn't here, I'll be taking a look at seeing if I can improve on the 2600k clock

No offense dude, but while it's great you have this info, didn't you start this thread with a problem that was fixed by something as basic as resetting the BIOS? Like it's great you pay attention to HWBot, but maybe spending some time covering the basics before the advanced stuff will pay off.


That was the point I had tried to make back then that Revin caught on to... and is now trying to make himself since he saw how easy it could really be...

First Ever heard of such a thing, @cadaveca You have any knowledge of this ?
Dave has used it vastly more than I have, and I been using XTU since I got my 2600K, at least in January 2012 , and that is how I was able to get the 5.2Ghz OC on mine.

IBT was an internal tool @ Intel that they used to test the limits of these chips. It'd kill chips. XTU uses the same linpack binary that IBT does, depending on which version you have, so both are capable of killing a chip if you have the right version, or wrong version, depending on how you look at it.

 

[phill]

No offense dude, but while it's great you have this info, didn't you start this thread with a problem that was fixed by something as basic as resetting the BIOS? Like it's great you pay attention to HWBot, but maybe spending some time covering the basics before the advanced stuff will pay off.

That was the point I had tried to make back then that Revin caught on to... and is now trying to make himself since he saw how easy it could really be...

IBT was an internal tool @ Intel that they used to test the limits of these chips. It'd kill chips. XTU uses the same linpack binary that IBT does, depending on which version you have, so both are capable of killing a chip if you have the right version, or wrong version, depending on how you look at it.

Well no offense taken but some of the most basic things we can over look and whilst having knowledge about a lot of things, sometimes you forget to try the simplest of things... Hopefully even for laughs, sometime later on, this thread might be of some use to someone who might be having the same problem

 

[Regeneration]

IBT was an internal tool @ Intel that they used to test the limits of these chips. It'd kill chips. XTU uses the same linpack binary that IBT does, depending on which version you have, so both are capable of killing a chip if you have the right version, or wrong version, depending on how you look at it.

Linpack is a linear equation benchmark tool. I've been using it for years, for long periods (8-12 hours) and never killed a single chip with it. High temperatures and voltages are more dangerous. Excessive stress for long periods can cause degradation for every component. I like my overclocked systems 1000% stable, I normally run Linpack for 8-12 hours, followed by 2 hour test of light load, and finally, Prime95 v26.6 blend with 80% of RAM with a GPU stress test running in the background. A bit overkill, but my systems never BSOD or crash after overclocking.

 

[EarthDog]

 

[cadaveca]

Linpack is a linear equation benchmark tool. I've been using it for years, for long periods (8-12 hours) and never killed a single chip with it. High temperatures and voltages are more dangerous. Excessive stress for long periods can cause degradation for every component. I like my overclocked systems 1000% stable, I normally run Linpack for 8-12 hours, followed by 2 hour test of light load, and finally, Prime95 v26.6 blend with 80% of RAM with a GPU stress test running in the background. A bit overkill, but my systems never BSOD or crash after overclocking.

The problem with it possibly killing chips is platform-dependent these days. Some CPUs increase voltage with AVX loading, and Linpack can cause voltage spikes just due to how it is loading the CPU, and when you add the two, it can overvolt a CPU and kill it in an instant.

Well no offense taken but some of the most basic things we can over look and whilst having knowledge about a lot of things, sometimes you forget to try the simplest of things... Hopefully even for laughs, sometime later on, this thread might be of some use to someone who might be having the same problem

Much respect. That is exactly what I meant. Just, you know, keep in mind the perspective given by HWBot users when you use such information. Using bits for benchmarking is quite different than for 24/7. As @Regeneration posted above, stress for long periods can do damage... that's a fact; and benchmarking at high clocks definitely counts as high stress loading.

 

[Regeneration]

Voltage spikes can be caused by poor implementation of LLC, best thing to do is to test it and aim for 1:1 voltage between load and idle.

AVX stress test generates a lot of heat, but maximum heat isn't maximum stress (AIDA64 for example).

You simply avoid stress tests that abuse AVX much and stick to old versions: Prime95 v26.6, Linpack 2012 (IBT).

 

[phill]

Oh I think I need to change one of the words I put... Instead of Some of the most basic, it should be most!! Us tweakers always seem to go to most complicated problem before looking at the most basic.. Almost reminds me of network cables and printers at work!!

I think with the AVX instructions and the repeated use of that benchmarking button in XTU is exactly why the CPU degraded and basically couldn't be used. Something in the test literally kills the CPU, so when you see the top guy benching it with LN2 and such, they have no worries as they aren't paying for the CPU.. I think it's kinda like a Intel give away that if a test or two manages to give them a few peaked interests, that might turn into sales and so then worth a couple dead CPUs.. Sadly even with all the overclocking I have done, I can't willingly kill my hardware or push beyond too far of the safety limit...
That said, that's probably why I have a few systems here.. Keep benching to one thing, 24/7 to another I'm still trying to work out why tho I have 14 or so rigs at home... It's good to have a spare tho right??

But hey, I digress and we can move on

All I do know is that my 2600k @ 1.2vcore has been crunching cancer for a day and a half at least now and no restarts or issues.. One of the core temps are miles above the rest but meh, it's fine

I might try and push for a little higher but thing with this cancer crunching for me is that it's all about efficiencies not (pardon the term) balls out power

 

[jaggerwild]

I had the 2500K miss that unit!
http://hwbot.org/submission/2234142_philly_cheese_steak_cpu_frequency_core_i5_2500k_5544_mhz/

 

[phill]

That was a damn good CPU!!

 

[jaggerwild]

That was a damn good CPU!!

I know, i had way to many volts into it(1.6XX ) . But had a custom loop and A/C unit running haaa!

 

[phill]

Ah it's fine for short term