• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Confusing 4790k behavior on Gigabyte Z97 UD5H

T

teiresias

Limp Gawd
Joined
Sep 27, 2010
Messages
182
I THINK this is a motherboard thing, so I'm putting it here, but it may be a CPU thing, so I apologize if this is the wrong forum.

In any case I am thoroughly confused by how my 4790k is behaving on this motherboard on two fronts. First, the vcore, and second the clock boost, so I'm wondering if anyone can enlighten me. Let's be clear though, my cpu isn't misbehaving or unstable, I just don't "get" what's going on based on what the actual BIOS settings are telling me.

First, the Vcore, which I'm readily willing to admit I may be looking at wrong. I felt the default Vcore under "auto" settings was too high (sitting at 1.215V in the BIOS), so I first set it manually, but the CPU wouldn't idle at all when running, so then I decided to use voltage offset.

My first time into the BIOS the displayed voltage while in "Auto", which stayed the same when I went to "Normal" to utilize the offset, was 1.215V. I tried an initial -0.035V offset and it was fine. So I went back to the BIOS to increase the negative offset, however, the display Vcore had inexplicably changed to 1.100V, as shown here:

http://i.imgur.com/VEdcF99.png

The only way to get it to set back to 1.215V (besides doing it manually) was to set it to Auto again and reboot the PC. However, the minute I went back to offset adjustment, rebooted and re-entered the BIOS again it had changed back 1.100V again. However, it obviously is not ACTUALLY using a 1.100V core voltage as the pre-offset maximum as show by my HWInfo while running AIDA64 stability test:

http://i.imgur.com/xhX9WKy.jpg

I thought the reported 1.100V is the max vcore available with the offset limiting the max to actually be lower. However, as shown above, my vcore under load is 1.188V so it can't be offsetting from 1.100V. Is this just the BIOS just misreporting its actual setting?

Finally, the frequencies. I thought turbo boost only worked to its advertised frequency on 1 or 2 cores at most. This is what the BIOS ratio settings indicate as well while in Auto:

http://i.imgur.com/sUw2sjf.png

However, when under load, all four of my cores are going to 4.4GHZ (well, 4.389GHZ actually, but who's counting):

http://i.imgur.com/V9X4Rko.jpg

I would have thought that a 4-core boost would only get a 4.2GHz boost. In fact, I've only ever seen a 4-core boost or nothing, even if I start up only two threads in Prime95 all four cores boost to 4.4GHz.

Is this unusual behavior or am I just misinformed?

Anyway, just hoped someone could enlighten me on these things here. Coming from Lynnfield maybe there's some Haswell voodoo I simply am missing.
 
Haswell is complicated.

Your proc is running turbo multiplier on all 4 cores because you have the Enhanced Turbo on or whatever, it is a BIOS setting that allows the turbo to be applied to all cores.

I can't really explain your board's BIOS behavior. It does sound odd. But what I can tell you is that if you are doing things like prime95 for testing, the general first rule is to not use AUTO or offset voltage modes. z87/97 boards alike will overvolt the crap out of a 4770/4790k in stress tests using these voltage modes. If you want to tell your BIOS what the capped maximum voltage will be, use manual or the equivalent.

On my ASUS boards, which should be similar, setting c-states to enabled instead of just auto for some reason lets the c-states function (voltage reduction @ idle) independently of Turbo/Speedstep state.

I've gone off topic but to address your confused BIOS voltage setting I would attempt a couple of things: first off I would completely reset the BIOS using optimized defaults. Secondly I tend to use XMP, as it seems to be involved with the processor clock as well, and not just the RAM frequency. RAM settings under XMP mode can still be altered manually in ROG BIOS. I believe the ASrock BIOS should be very similar.
 
Aw, completely missed enhanced turbo. Thanks for that.

I've reloaded optimized defaults a few times on the BIOS so that doesn't really help, it still behaves the same with the voltage. However, I haven't experienced runaway vcore while running Prime95 using offset voltage, it seems very well behaved to me and stays within the boundaries I'd expect (when I choose to not believe the 1.100V base as reported by the BIOS anyway and just go by what HWInfo or CPU-Z is telling me and temps are well behaved).

I will try the XMP thing though.
 
Hmm I was thinking as rock. Gigabyte seems to have a much better handle on auto voltage than the other vendors.
 
If you want default intel behavior, you need to DISABLE enhanced turbo.
 
Is there a problem with using enhanced turbo so long as temperatures are ok?

Actually, I can't find any reference to Enhanced Turbo in the manual (not at home to pick around in the BIOS). Maybe it's the "K OC" setting? But some googling tells me that just is useful for non-k processors. Perhaps I just need to manually set the boost multipliers in the BIOS instead and having Auto on all of them results in the "enhanced" behavior.
 
Last edited:
nothing wrong with running enhanced at all, but it locks all the cores to the same turbo, so you dont get that extra multiplyier when only one or two cores are loaded.
 
Well, in my case it's locking all four cores to the marketed 4.4GHz boost, so I'm not really losing anything in that regard.
 
which is exactly what is supposed to do.

But if its unlocked, and you only have 2 threads loaded, youd get 45 or 46.

Thats the difference.

I get 48 if only one core is loaded, and 46 if 2.
 
The Mac said:
which is exactly what is supposed to do.

But if its unlocked, and you only have 2 threads loaded, youd get 45 or 46.

Thats the difference.

I get 48 if only one core is loaded, and 46 if 2.
Click to expand...

Hmm, actually that's not what happens with mine. If I start only two threads in Prime95 all the cores still go to 4.4GHz, it doesn't seem to selectively boost cores at all.
 
becaue its locked (enhanced) it has to be unlocked (stock intel) for selective boost.
 
Its the "all cores" option, turn it off.

On asus boards, you can disable XMP, then renable it to get it to reset, not sure how it works on Giga.

If you are going to OC however, its better to leave them locked.
 
I still haven't found a global "all cores" option in the BIOS (K-OC is not it, or at least it didn't cause any change in my rig's behavior). I guess the only way to do it is to set all the Turbo Boost ratios manually and try from there.
 
nothing with "enhanced" in it?

did you try disabling xmp then reenabling it?
 
Nope, nothing with "enhanced" in it, and XMP settings aren't even enabled in the BIOS since my RAM doesn't have XMP (it's older G.Skill 1333MHz RAM from like four years ago).
 
teiresias said:
Nope, nothing with "enhanced" in it, and XMP settings aren't even enabled in the BIOS since my RAM doesn't have XMP (it's older G.Skill 1333MHz RAM from like four years ago).
Click to expand...

Yes, Enhances turbo core is equal to K OC in Gigabyte.
 
youll need to grab a per core clock monitor and see what its doing with assorted threads loaded.

if they are all boosting simultaneously even though only one or two cores are loaded, then they are still locked.
 
I've done that, that's why I'm saying K OC isn't the same as "enhanced boost", because that setting does not change that behavior with less than fully loaded cores and boost behavior.
 
Update the BIOS; the early batches of GB boards have the F2 (or F3) initial release BIOS that has issues with the VID and it goes way overboard. First ever boot, all stock, mine was 1.39 Vcc which is ridiculous. Thought I had a horrendous CPU, did some research, found others complaining of the same with the fix being to update the BIOS. Did so, rebooted and it set it to 1.12.

K OC is for non-K CPU's according to Sin's Z97X OC guide.
K OC(1): Unlocks multipliers on some multiplier locked CPUs
Click to expand...

IME yes, you set the multi in 'Frequency', it runs at that at all times (no matter how many cores you're using). If you want to run x multi 1-2 cores, y multi on 3 cores and z multi on 4 cores, you'll have to set that under the 'Advanced CPU Core Settings'.
 
Yeah, I'm on the latest BIOS (as of two weeks ago anyway, haven't checked lately), and I had to manually adjust the vcore even on that (it wasn't awful on the latest BIOS, but I was able to get a good deal of temp headroom under load by adjusting the offset).
 
You must log in or register to reply here.
Back
Top