New Skylake 6700k & MSi M9 ACK Build Will NOT Overclock at all. Major issues.

[Mandrake4565]

So I loaded optimized defaults for the 2nd time. Then set multiplier to 45 to get 4.5ghz just left vcore @ auto. Turned xmp on (gskill 3200, 16gb ddr4). I saw after running aida for 25 minutes no crash cpu vcore showed 1.352. So I went into the bios and changed the vcore from auto to 1.375. Ran aida again. No crashes.

So I upped the multiplier to 47. 4.7ghz. cpu vcore running in aida shows 1.360 vcore. Stable. Running for 15 minutes now. Not sure what is going on. If it runs stable for 2 hours. Then I am going to change the base clock to 109, and multiplier to 44 to get the memory up to 3488 mhz like I did on the xpower. If it crashes I assume the ram is bad?

GenieGOR slow down, you're only going to drive yourself crazy if you keep changing multiple parameters. Start with what ED suggested, reset to default and then change only the Cpu V to manual at 1.40, change nothing else and see if it will boot.

 

[GenieGOR]

Why would you assume the ram is bad? What ram do you have? Can you create a signazure with your hardware specs please...

Perhaps post some cpuz shots of memory and SPD tabs...

Why are you touching bclk? You should slow down a bit and focus on one thing before bouncing around... Scary how scattered this testing seems (I need some coffee too).

The bclk I typically have at 109. Same with the other build. Just jumped ahead of myself. So far everything right now is good. I had it stable 4.7 ghz @ 1.375 vcore. I jumped it to 4.8thz and it failed right away and watchdog timeout. So I set vcore to 1.390. This time the test ran for a few minutes and then watchdog. Set vcore to 1.42. Running stable now @ 4.8ghz still running the test.

I noticed the max vcore peaked around 1.408 that's why my 1.390 failed as it needed a bit more. Pretty consistent around 1.392 tho. Temps showing 67-68 range.

It looks to me that I have missed something but this chip seems to be clocking cooler with less voltage now then my other computer I built with the X power last month. I had a few peaks go to 1.418 but strong and stable @ 4.8ghz.

Vcore set to 1.43. Temps are showing peak sometimes around 69-71 and vcore sometimes gets to 1.42, most of this test it's been pretty steady at 1.400.

Also I am reading the Cpu temps from the digital readout on the board.

 

[EarthDog]

What is it with people taking screencaps (as opposed to screenshots like prt scn or alt+prtscn) circa 2000?

Not sure why you randomly set BCLK @ 109...no harm in it of course.

This is a new chip right? That would explain the voltage and temperature differences. Even if it isn't a new chip, you have a different mount and TIM application which can make a difference as well.
EDIT: new board only... same thing really. board does make a difference as to what voltage you need to set it at... same thing on the mount as well.

We need the SPD tab please.

 

[Mandrake4565]

GenieGOR, typical and overclocking doesn't work hand in hand. I have rarely seen a setup that will Oc the same even if they are exactly alike. In the post above are you on Auto voltage still or manual?

 

[GenieGOR]

What is it with people taking screencaps (as opposed to screenshots like prt scn or alt+prtscn) circa 2000?

Not sure why you randomly set BCLK @ 109...no harm in it of course.

This is a new chip right? That would explain the voltage and temperature differences. Even if it isn't a new chip, you have a different mount and TIM application which can make a difference as well.
EDIT: new board only... same thing really. board does make a difference as to what voltage you need to set it at... same thing on the mount as well.

We need the SPD tab please.

Here are better screenshots. I thought the same thing. New chip, takes time to massage it and heat it up? My other rig was similar I left it at 4.4ghz for the first 3-4 days. anyway here are the screenshots.

 

[BonaOver]

Wait, did you sell your 6700k and got another 6700k? Why?

I'm sorry but this just looks like silicon lottery to me, you shouldn't have changed the chip.

 

[EarthDog]

He said he builds systems or something...

That said, genie, link the exact ram you purchased. The SPD tab show its 2133 mhz, and you have it running at 3200 mhz...

 

[BonaOver]

He said he builds systems or something...

That said, genie, link the exact ram you purchased. The SPD tab show its 2133 mhz, and you have it running at 3200 mhz...

But really it is kinda obvious the "same" chip can't achieve the same OC, one chip was just better than the other

Also, this mobo is pretty cable of a very good OC, this is just a "bad" chip, compared to the other one.

 

[Mandrake4565]

That said, genie, link the exact ram you purchased. The SPD tab show its 2133 mhz, and you have it running at 3200 mhz...

+1

 

[GenieGOR]

Here is the ram I have. Xmp enabled 3200mhz.

http://www.newegg.com/Product/Product.aspx?Item=20-231-914

Also under AIDA64 I get an average of 1.41 vcore through the whole test. The only time it changes is the first five seconds when I start the test and when I stop the test the vcore sometimes can spike to 1.43 or 1.44.


What confuses me about this is the fact that this chip was giving me trouble but it turns out this chip may actually be better, because my old chip clocked at 4.8 GHz around 1.440 however this chip is doing the same thing at 1.416.

I'm not sure what happened but it seems things are going much better now than they were and I really truly haven't changed anything I'm doing it's like to chip just stabilized and that's very weird Lol

Also it seems this chip might still have some more headroom left and it may be 4.9 GHz. My temperatures under AIDA64 show all cores to be between 69° and 77°. The package Temp on the Corsair link show is 75°, and the digital readout on the board hardly ever goes above 68°

Edit: bios I have set to 1.440 and the Cpu idles at 1.440 and full 4.8ghz. When I stress it goes down to 1.416. I don't know why it won't drop down at idle like it should. Normally it will drop to 800 MHz and much lower vcore.

 

[Mandrake4565]

You either have the power saving features disabled in the bios or have windows on high performance.

 

[GenieGOR]

You either have the power saving features disabled in the bios or have windows on high performance.

OK I see it. Yesterday it wasn't doing this but now it's stuck at full throttle all the time and under windows power management that shows minimum processor at 100% not sure why windows did this all on its own. What's a good recommended setting for windows? I've never had to mess with the setting before usually I just leave it on balanced in the process or slows down when not using it.

Edit: I think I just found out what may be causing my problem. With the windows power performance settings. If I leave the minimum processor at 100% everything is stable, as soon as I put it under power saver or allow the processor to slow down while idle I cannot boot into windows unless the vcore is set to auto.

With windows set to balanced power and processor to 100% all the time no energy savings I get a pretty damn stable 1.42 @ 4.8 GHz.

I guess this is something going on with windows power management and when it brings the processor out of the whole state it adds extra voltage to speed it up or something I'm not sure but every time I take power saving out of the equation and pack the processor to Maxpeed all the time I have zero issues. What do you guys think is this a processor issue or a Windows install issue?

 

[Mandrake4565]

Sounds like at the lowest P-State the processor isn't getting enough voltage. Though I'm unfamiliar how the power management works with windows as far as start up goes. You could try changing the minimum processor state to something like 10-15%, I believe 5% is where it is when on balanced and see if it helps. The other thing you can try is using offset voltage or adaptive. I've never used adaptive so ED may better be able to help you with it. With offset you'll need to figure out the stock VID of your chip then subtract it from 1.42 and that is what you'll set the offset to. So for example if your stoce Vid is 1.2 the off set would be +.22.

 

[GenieGOR]

Sounds like at the lowest P-State the processor isn't getting enough voltage. Though I'm unfamiliar how the power management works with windows as far as start up goes. You could try changing the minimum processor state to something like 10-15%, I believe 5% is where it is when on balanced and see if it helps. The other thing you can try is using offset voltage or adaptive. I've never used adaptive so ED may better be able to help you with it. With offset you'll need to figure out the stock VID of your chip then subtract it from 1.42 and that is what you'll set the offset to. So for example if your stoce Vid is 1.2 the off set would be +.22.

It's weird how the other Cpu and X power board didn't need any of this. How do I set voltage for low state of power?

Also I only see the following on my board:
1.cpu vcore
2.cpu sa voltage
3.cpu io voltage
4.cpu pll of voltage
5.cpu pll sfr voltage
6.dram voltage
7.pch voltage

 

[Mandrake4565]

You really shouldn't need to change any of those for ambient overclocking. It's only when one is really pushing that you need to start pushing other voltages. You did say the BIOS was up to date correct? (too lazy to go back and look) Try changing the minimum power state in windows from 5% to 10 or 15 % and see if it help. I have no idea if it will, but worth a shot.

 

[GenieGOR]

You really shouldn't need to change any of those for ambient overclocking. It's only when one is really pushing that you need to start pushing other voltages. You did say the BIOS was up to date correct? (too lazy to go back and look) Try changing the minimum power state in windows from 5% to 10 or 15 % and see if it help. I have no idea if it will, but worth a shot.

I do have the newest bios. I changed the minimum power and windows to 15% and still the same issue.

If I have the windows power limit set to 5% or even 15% and I have the vcore set to auto at full stress test it sits at 1.51 vcore at idle it's 1.16 @ 800mhz.

If I have windows set to 100% all the time under processor management, I can set the vcore to 1.42 and at full load under stress test it reads 1.416 vcore @ 4.8ghz.

What's strange is when I'm not stressing the CPO with windows at 100% that vcore sits at 1.432.

If I change windows power management at all while the computer is running and the vcore is set to anything other than auto or above 1.52 windows will crash.

This is extremely Bizzarre behavior that my other system did not encounter it's quite obvious that this processor is capable of staying below 1.42 at 4.8 GHz. It's just all this weird behavior.

With everything Stabile 100% in windows, If I jump to 4.9ghz it wants to jump from 1.416 to 1.475 vcore.

I think I'm going to RMA this processor and board and may go back to the X gaming titanium. The only problem I have is I need to find a really reliable Wi-Fi/Bluetooth card that I can put in and I don't know of any.

Or I might give the MSi M9 one more chance, I just want solid over walking ability and I'm not sure if it's my board, or the processor

 

[OC-NightHawk]

Isn't it an abuse of the RMA system to return a processor and board that operate at specification? Just because these two items lost the lottery doesn't mean you should RMA them. Sell them and try again if it pleases you but don't return a perfectly fine processor as defective.

 

[EarthDog]

Isn't it an abuse of the RMA system to return a processor and board that operate at specification? Just because these two items lost the lottery doesn't mean you should RMA them. Sell them and try again if it pleases you but don't return a perfectly fine processor as defective.

this....

 

[GenieGOR]

Turns out the board had issues. I had it looked at by the techs at my suppliers warehouse. Don't know but something was wrong with the board. Its not in good standing on newegg reviews either. Anyhow,

I got a Asus ROG Maximus Extreme. I am able to run 4.7/4.8 ghz at a really good vcore it seems. SIDE NOTE: I was even able to bump the DDR4 up to 3600 MHz from 3200 (something the Msi xpower couldn't do). I had to change the timings 16-18-18-38 to get it stable. However, surprisingly the ram that I tested was the 16-16-16-36 tridentZ and its faster in Aida64 memory test than the 14-14-14-34 tridentZ when setup to the exact same 16-18-18-38 timings @ 3600mhz. Cheaper and faster pretty cool (I tested many times)


Here are some screenshots. First picture shows vcore on the cpuz(oem) and VID on cpuz(ROG Asus) @ idle. vcore is still at 1.408 but VID is very low as the processor is idle. The next 3 are of stress tests. What do you think? Anything I should tweak? The Vcore is holding steading but the VID(request voltage) is in a good range or so it seems to me.

Here is another one. Just the CPU being tested. The VID on cpuz shows less than 1.375 all the time. The vcore seems to stay @ 1.440 even though I have it at 1.42 in the bios.

 

[EarthDog]

Ok... VID = stock voltage. Anything outside of that value is vCore or Core voltage. But VID is not any value outside of stock voltage. Your VID is (for example) 1.15V. If you are using 1.35V that is vCore. VID = stock.