New 4790k and older Gigabyte z97m-d3h... stock voltages are nonsence...

[Kaynar]

Hello all,

I just bought an 4790k to upgrade from a Pentium G3258 using a gigabyte z97m-d3h mobo that i had and Corsair Dominator platium 4x4GB kit at 2133mhz.

Here comes the "problem": After reseting the bios and fixing all the options to standard stuff, the Auto voltages for the CPU are set to 1.24v Vcore and 1.8V Vrin. That looks waaay to high for a stock voltage.

When setting the XMP profile on, the Vcore becomes 1.27v and the DRAM voltage is automatically set to 2.1v (yea lol i dont know how the system even booted like that).

Of course this meant crazy near 100c temps during Prime95 and me panicking. Can someone tell me why Intel or Gigabyte allows such mistakes to happen? Mind that I could never get past 3.8ghz on the Pentium AE i had while everyone seemed to reach 4.5ghz easily so there might be something very wrong with Gigabyte boards.

I had to manually set the clock ratio to x44 and reduce the Vcore = 1.14v, VRIN = 1.7v and the VRAM = 1.6v (seems stable at this voltage, crashes during boot with vcore 1.11v) so now my max temp is 70c after 5 minutes of testing. I can shave another 5c with my Corsair H110 I guess by getting something better than Deepcool Z5 crappy thermal paste but thats all I could find for now.

I would like to know if there is anything else that could have been automatically set wrong in the bios. I mean if there is any other critical voltage for the system's stability and longevity, something that the motherboard could have set excessively high.

Thanks.

 

[tabascosauz]

1.11v is a good place to be at x44. I believe my own VID is 1.24. My temps are perfectly reasonable on a D9L.

Haswell should not be tested with Prime95. Use OCCT instead. Prime95 will burn thru any cooler on Haswell.

Perhaps you should have a shot at BIOS updating. Just maybe. What version of the BIOS do you have?

Your G3258 may have just been the short straw in the silicon lottery. Many G3258 will overclock well, but not every single one.

 

[little cat]

http://www.anandtech.com/show/8227/devils-canyon-review-intel-core-i7-4790k-and-i5-4690k/2

 

[Kaynar]

1.11v is a good place to be at x44. I believe my own VID is 1.24. My temps are perfectly reasonable on a D9L.

Haswell should not be tested with Prime95. Use OCCT instead. Prime95 will burn thru any cooler on Haswell.

Perhaps you should have a shot at BIOS updating. Just maybe. What version of the BIOS do you have?

Well 1.11v didnt make it to the desktop but 1.14v is still okey now. Im just so surprised that the XMP set the VRAM at 2.1v though

Okey I will try OCCT, I tried a few cpu intensive games and it looks fine (temps below 65c always at 1.14v). I've re-sit the heatsink and re-applied thermal paste 3 times to ensure i didnt do something wrong, always get the same temps.

I am using F7 bios, there is only F8 to download which says "improved compatibility with 5th gen intel processors" - 4790k still counts as 4th gen, right?

edit: In OCCT the temps dont go above 65c, and im sure a better TIM would do wonders.

 

[tabascosauz]

Well 1.11v didnt make it to the desktop but 1.14v is still okey now. Im just so surprised that the XMP set the VRAM at 2.1v though

Okey I will try OCCT, I tried a few cpu intensive games and it looks fine (temps below 65c always at 1.14v). I've re-sit the heatsink and re-applied thermal paste 3 times to ensure i didnt do something wrong, always get the same temps.

I am using F7 bios, there is only F8 to download which says "improved compatibility with 5th gen intel processors" - 4790k still counts as 4th gen, right?

edit: In OCCT the temps dont go above 65c, and im sure a better TIM would do wonders.

65 is crazy cool. Anything up to 75 is tolerable for me and not even close to thermally jeopardizing the longevity of the chip.

You never know whether that BIOS might solve your problems. On the other hand (and a HUGE but), every BIOS flash carries a small chance of something going wrong so if you can, avoid flashing BIOS if your issues are small. I'd say you have things under control now.

 

[Kaynar]

@tabascosauz Yea thats why im not updating to F8 without solid reasons. And yes 65c is perfect. Im just so frustrated that the auto voltage values were so high without reason... its the first time this happens since Pentium 4 days, and I've had alot of systems set up for me and for friends. My friend's i5 4690k for example is set to 1.11v on auto with an Asus Ranger mobo and an H110 cooler and never goes above 60c, which is normal for an auto values...

 

[Kursah]

These chips throttle very effectively at their 100C tjunction, and I've yet to see one fail that hit that throttle point. I've handled many of them, seen many in situations where they hit that limit far more than one would like them to. When they throttle at that limit, they lightly throttle back speed, to maintain performance...not cutting down to idle speeds...but rather down -100 to 200MHz at first. That leads me to believe that Intel isn't worried about these chips running hot...and I haven't been after years of seeing many of them run hot and not fail. 65C is nothing in perspective to that.

4790K are 5th Gen Devil's Canyon, which was a little rework of Haswell 4th gen.

65C is a good temp for a 4790K, I wouldn't be worried at all with that temp. Another thing to keep in mind is the smaller die where the wattage heat output is more focused in a smaller area, making harder to effectively cool with standard solutions...but not impossible.

What OCCT test did you do? I recommend Linpack with all logical cores, do NOT do AVX. See what you get after an hour. OCCT is a great stability testing utility, and has been my go-to for years.

 

[BiggieShady]

Well 1.11v didnt make it to the desktop but 1.14v is still okey now.

You should still look at Event Logs for WHEA cpu recovery errors to verify if 1.14v is really enough.

Im just so surprised that the XMP set the VRAM at 2.1v though

XMP Profile is stored in EEPROM on RAM sticks and the BIOS just reads those values ... it's possible that excessive voltage 2.1v setting is written there by error or incorrectly read from there, but manually entering correct values from XMP Profile should work just fine

 

[Kaynar]

XMP Profile is stored in EEPROM on RAM sticks and the BIOS just reads those values ... it's possible that excessive voltage 2.1v setting is written there by error or incorrectly read from there, but manually entering correct values from XMP Profile should work just fine

Yes manually entering everything solved all the issues I initially had...

Ran, OCCT as @Kursah suggested, max temp 63.

Anyway, I guess there is nothing else to discuss about my problem. Thanks all for the feedback

 

[Toothless]

At least you weren't like me and got 1.45v on first boot.

 

[Folterknecht]

You re not the first with problems when it comes to Gigabytes Z97 boards - I also had my share "fun" with them

http://forum.wotlabs.net/index.php?/topic/15590-warning-z97x-ud5h-bk-stay-away-from-it-update/

http://forum.wotlabs.net/index.php?...from-it-update/&do=findComment&comment=425017