2500K 12+ hours prime stable @ 4.8, but has problems at boot-up/sleep!

996GT2

Diamond Member
Jun 23, 2005
5,212
0
76
As the title says, my 2500K is 12+ prime95 blend stable @ 4.8 GHz, 1.35v. I've tried with PLL Overvoltage enabled and disabled; this setting does not make a difference in prime95 stability on this particular chip.

481352vprime95.png


However, even though the overclock is seemingly stable, I have issues recovering from sleep and booting up. Sometimes, the system will not recover from sleep. I know Intel has an issue with S3 sleep and PLL overvoltage enabled, but I do not have PLL OV enabled right now. In addition, the system sometimes will freeze while booting Windows and will require a fresh restart.

My specs are in sig. I have the latest BIOS on my Maximus IV GENE-Z.

Any ideas as to why my system is passing 12+ hours of Prime95 blend without any problems, but having boot issues/sleep issues?
 
Last edited:

richierich1212

Platinum Member
Jul 5, 2002
2,741
360
126
I've gotten random BSODs with my 2600K once I occupied all four RAM slots. I had to bump vcore and memory voltage a bit to stabilize. Maybe you might have to increase vcore and memory voltage as well?
 

996GT2

Diamond Member
Jun 23, 2005
5,212
0
76
I've gotten random BSODs with my 2600K once I occupied all four RAM slots. I had to bump vcore and memory voltage a bit to stabilize. Maybe you might have to increase vcore and memory voltage as well?

Perhaps, but it seems strange since it passed almost 14 hours of Prime95 blend when all 4 slots were occupied.

I'm running 1T with all 4 slots occupied, but if it passes multiple cycles of memtest86+ and Prime95 blend, I'd think that was stable...
 
Last edited:

Kenmitch

Diamond Member
Oct 10, 1999
8,505
2,250
136
You could keep playing around with vcore and see if it'll help.

My 2500k would loose the ability to return from sleep with anything over 4.6ghz. My 2700k has no issue returning from sleep at 4.8ghz. I'm thinking it's a luck of the draw thing just like with overclocking ability.

At what clock does yours loose the ability to return from sleep? Did you try lower clocks yet to see? If sleep is important to you then you might just have to run at a lower overclock.

Being prime stable doesn't guarantee your 100% stable. Try Intel Burn Test as it'll hit the chip a lot harder. Just make sure you get the latest version with AVX support.
 

996GT2

Diamond Member
Jun 23, 2005
5,212
0
76
You could keep playing around with vcore and see if it'll help.

My 2500k would loose the ability to return from sleep with anything over 4.6ghz. My 2700k has no issue returning from sleep at 4.8ghz. I'm thinking it's a luck of the draw thing just like with overclocking ability.

At what clock does yours loose the ability to return from sleep? Did you try lower clocks yet to see? If sleep is important to you then you might just have to run at a lower overclock.

Being prime stable doesn't guarantee your 100% stable. Try Intel Burn Test as it'll hit the chip a lot harder. Just make sure you get the latest version with AVX support.

This is a common misconception. LinX/IBT do not stress the CPU more. They make the CPU run hotter, yes, but they are not as comprehensive of a stability test as Prime95 Blend. I can easily pass multiple hours of IBT at even 5 GHz, but to pass 12 hours of Prime95 blend @ 5 GHz requires more VCore.

I'm trying a slight (0.01V) bump in VCore along with a slight bump in RAM voltage. Will report back.
 

Cookie Monster

Diamond Member
May 7, 2005
5,161
32
86
Since its an Asus board, disable PLL overvoltage and C3/C6 states. This could potentially fix BSOD during idle, during light workload, cold boot issues, wakeup issues etc. I think its well documented here.

I think its best to test with IBT 2.53 (one with AVX enabled) first. You can also check the GFlops result to see if your voltage for that overclock is stable (cpu could be "stable" but underperform - also can compare your numbers to others).

Then if it passes 5 tests, do a 20 test run. If that is fine, go ahead with the 12hr prime95 blend. Then test gaming load, browsing/movie etc.

Good luck!
 

996GT2

Diamond Member
Jun 23, 2005
5,212
0
76
Since its an Asus board, disable PLL overvoltage and C3/C6 states. This could potentially fix BSOD during idle, during light workload, cold boot issues, wakeup issues etc. I think its well documented here.

I think its best to test with IBT 2.53 (one with AVX enabled) first. You can also check the GFlops result to see if your voltage for that overclock is stable (cpu could be "stable" but underperform - also can compare your numbers to others).

Then if it passes 5 tests, do a 20 test run. If that is fine, go ahead with the 12hr prime95 blend. Then test gaming load, browsing/movie etc.

Good luck!

Will look into that! So far I've given it a slight boost in VCore (to 1.37), and a slight boost in RAM Voltage (to 1.54V), along with disabling EPU power saving. It didn't have any issues coming out of S3 sleep this morning, but we'll see how it does with more sleep/wake cycles.
 

Rvenger

Elite Member <br> Super Moderator <br> Video Cards
Apr 6, 2004
6,283
5
81
Since its an Asus board, disable PLL overvoltage and C3/C6 states. This could potentially fix BSOD during idle, during light workload, cold boot issues, wakeup issues etc. I think its well documented here.

I think its best to test with IBT 2.53 (one with AVX enabled) first. You can also check the GFlops result to see if your voltage for that overclock is stable (cpu could be "stable" but underperform - also can compare your numbers to others).

Then if it passes 5 tests, do a 20 test run. If that is fine, go ahead with the 12hr prime95 blend. Then test gaming load, browsing/movie etc.

Good luck!

+1 :thumbsup:
 

Kenmitch

Diamond Member
Oct 10, 1999
8,505
2,250
136
Good to see maybe the bump in vcore took care of your issue. Sometimes it does and sometimes it doesn't. :)

I'm not trying to argue with you first off. Everybody has their own technique/style of overclocking. In the past IBT may or may not have been better than prime for stability testing. Things kinda changed a little once the AVX extensions came into play. I know alot of software doesn't take advantage of it yet....But someday you may just install some that does.


Intel Burn Test with AVX support for initial testing....Prime later is the best way to get to stability. Yes it's true sometimes it'll take a tic or two more on the vcore to pass prime in the long run. IBT first speeds things up alot also. Nothing like looking at your rig and seeing prime failed on a core or two after 7hrs and 52min of wasted electricity. Up the vcore a tic or two and try again.

I didn't plan on running my chip at this speed or vcore. I'm just using this for an example. I didn't wanna push anymore vcore than needed to just get it somewhat stable. More for just to see if it would do it more than anything. And to get a screenshot.

With this vcore prime is somewhat stable at this speed. Not sure how long it would have ran if I just let it sit and do it's thing. I just can't do that tho and was posting in this forum, while taking the screenshot and uploading it to photobucket. It lasted a little over 30 minutes before I got a bsod which I expected long before that.

My point is with IBT with AVX support it was hit start = bsod

52ghz.png


If you wanna be 99% sure your overclock is stable then run prime for 12hrs plus and slam IBT on top of it. If they both continue on and you don't get a bsod or crash then it's most likely stable. The other 1% is a windows thing. Some would argue this percentage is greater tho. :)