Interesting revelation regarding Z97 + i5-4670k

[Faceman]

So, I recently had a Z87-A + i5-4670k.  I was running a 4.5Ghz OC @ 1.2v because well, temps were better than the 4.7 @ 1.275v and I don't really need that power.  Heck, I don't even need 4.5 for gaming.

 

Anyways, my friend recently bought a system and I was able to sell him some of my old components, one of which being my Z87-A, and I upgraded to a Z97-AR.  Everything went smooth, it was easy to upgrade mobos, just uninstall the old drivers, and install the new.  Piece of cake.  The Asus Z97 platform is incredible by the way, especially Fan Xpert III.

 

I decided to re-do my overclock to see if anything would be different, and yes, it is.  I kept the same 45 multiplier, but I was able to drop my voltage down from 1.200(which is already pretty good) all the way down to 1.165v!

 

Anyone able to explain why this happened?  If I had to guess, something about the Z97 being designed for higher TDP processors than the Z87.  I always thought that the quality of the overclock was 99% predicated upon the quality of the chip, but going from Z87 to Z97 has proven otherwise.

[Deletive]

Z97 is a newer chipset with some improvements. 

[TheKDub]

Just a better overclocking board?

[TopplesS]

 I kept the same 45 multiplier, but I was able to drop my voltage down from 1.2(which is already pretty good) all the way down to 1.65v!

 

 

MATHEMATICS!

But yes, the newer board probably has better power delivery to the chip, requiring less to make up for the fluctuations in the previous board. Good work but!

I am also a fan of the new BIOS and such.

[Snickerzz]

What? 1.65>1.2

Am i missing something?

[Faceman]

@Snickerzz

Sorry, sorry, will edit. 1.165

[Snickerzz]

@Snickerzz

Sorry, sorry, will edit. 1.165

ahh much better

[79wjd]

So, I recently had a Z87-A + i5-4670k.  I was running a 4.5Ghz OC @ 1.2v because well, temps were better than the 4.7 @ 1.275v and I don't really need that power.  Heck, I don't even need 4.5 for gaming.

Most people buying K cpus don't need the added power, its all for the E-Peen. 

 

Not that I haven't been tempted to under clock my chip in order to make my computer dead silent... So long as I'm getting 60 fps in all my games, theres no real reason to push the chip harder... 

[VG Docs]

This is irrelevant, but I might as well ask here rather than making a new thread about it - what's the default voltage of the 4670k? Mine's 1.198v but I don't ever remember touching the voltage in my overclock... sounds awfully high to me.

[Faceman]

This is irrelevant, but I might as well ask here rather than making a new thread about it - what's the default voltage of the 4670k? Mine's 1.198v but I don't ever remember touching the voltage in my overclock... sounds awfully high to me.

It can't be that high..

 

Default voltage is 1.0 to 1.1.  1.0 for 3.4 and 1.1 when it boosts to 3.8

[VG Docs]

It can't be that high..

 

Default voltage is 1.0 to 1.1.  1.0 for 3.4 and 1.1 when it boosts to 3.8

Strange... I purposely don't bump up the voltage because I don't need a high OC. That's what it says in CPU-Z though.

Could my motherboard have automatically bumped up the voltage with the OC? Like an adaptive voltage or something?

[Faceman]

Strange... I purposely don't bump up the voltage because I don't need a high OC. That's what it says in CPU-Z though.

Could my motherboard have automatically bumped up the voltage with the OC? Like an adaptive voltage or something?

Yea, but not that high.  For instance, I set my voltage to 1.275 and it would bump up to 1.282 something small like that.  Right now I have it set for 1.165 but it is bumping up to 1.171.

[VG Docs]

Yea, but not that high.  For instance, I set my voltage to 1.275 and it would bump up to 1.282 something small like that.  Right now I have it set for 1.165 but it is bumping up to 1.171.

Okay, well quick update: in the BIOS there was an option to set the vcore voltage mode. At the time it was on "auto", I just now set it to "override" and set it to 1.165v

In CPU-Z it shows 1.163v, so it worked. Let's just hope I didn't lose the silicon lottery on this one... oh please remain stable good 4670k

Also, just to note - it remained at 1.198v before, it never changed at all.

[airdeano]

If I had to guess, something about the Z97 being designed for higher TDP processors than the Z87.  I always thought that the quality of the overclock was 99% predicated upon the quality of the chip, but going from Z87 to Z97 has proven otherwise.

 

i just updated my Z87 Hero UEFI (4th GEN update) and now the operating stock

voltage is 1.008v and not the older UEFI of 1.232v (which is my undervolt 1.005

40 multiplier - 4770k) more TPU/EPU algos at work. 

[SkilledRebuilds]

While I love me some overclocking and in all the time I've been playing with my PC I never really compared lowend stock to highend overclocked.

I was genuinely curious yesterday on differences.

 

[Faceman]

Okay, well quick update: in the BIOS there was an option to set the vcore voltage mode. At the time it was on "auto", I just now set it to "override" and set it to 1.165v

In CPU-Z it shows 1.163v, so it worked. Let's just hope I didn't lose the silicon lottery on this one... oh please remain stable good 4670k

Also, just to note - it remained at 1.198v before, it never changed at all.

Don't set it to override, set it to adaptive 1.165 and keep lowering it until you BSOD, then go back up to the last stable voltage.  Override is for use only when stress testing.

[Faceman]

i just updated my Z87 Hero UEFI (4th GEN update) and now the operating stock

voltage is 1.008v and not the older UEFI of 1.232v (which is my undervolt 1.005

40 multiplier - 4770k) more TPU/EPU algos at work. 

I actually had something similar happen too.  I had my Z87-A on the BIOS from the disc that came with it. At this time, I was doing auto-OC because I was getting amazing results.  4.7Ghz when 1 and 2 cores were in use, and 4.6Ghz when 3 and 4 cores were in use @ 1.275v, that was the auto-OC.  So I updated my BIOS to 1802 I think, the one that makes it compatible with DC chips.  I redid the auto-OC and it set my CPU to 4.7Ghz on all cores.  So the BIOS update must have improved something.  Anyways, I later found out that while the auto OC was good for me to get an idea of what my chip is capable of, it had set my LLC to a maximum 8/8 which is not so good.  I finally decided to do it manually, and I set it to the exact same 4.7 on all cores @ 1.275v except with LLC at 4/8 and everything worked like it should.

 

Still, going from Z87 to Z97 and I was able to get .35v reduction and remain stable is pretty incredible.

[VG Docs]

Don't set it to override, set it to adaptive 1.165 and keep lowering it until you BSOD, then go back up to the last stable voltage.  Override is for use only when stress testing.

Oh, alright. Will do.

 

While I love me some overclocking and in all the time I've been playing with my PC I never really compared lowend stock to highend overclocked.

I was genuinely curious yesterday on differences.

-Snipe-

The only time I really enjoyed overclocking was with my E5300 and a DG35EC, which you couldn't overclock through the BIOS. I like finding ways to get around normal means like that

[SkilledRebuilds]

Oh, alright. Will do.

 

The only time I really enjoyed overclocking was with my E5300 and a DG35EC, which you couldn't overclock through the BIOS. I like finding ways to get around normal means like that

Same here, Q6600 in a DELL Bios.

Had to use the CPU PIN Tape Mod to force 1333Mhz FSB instead of the 1066Mhz, taking the 2.4Ghz up to 2.999Ghz

Beast.

[Faceman]

While I love me some overclocking and in all the time I've been playing with my PC I never really compared lowend stock to highend overclocked.

I was genuinely curious yesterday on differences.

 

While I appreciate the video, and I do think that he is right that overclocking isn't going to give you very tangible results for gaming, the tests he ran are incredibly flawed.  Throughout all of the tests, there was NO ACTION.  He fired his gun twice throughout that entire video and it was only against 1 enemy on the screen in Metro:LL.  I would have liked to see some multiplayer tests run in Crysis3 and BF4.  Again, I don't think OCing will give you very big performance gains, but that video is not a definitive "It doesn't do anything."

[VG Docs]

While I appreciate the video, and I do think that he is right that overclocking isn't going to give you very tangible results for gaming, the tests he ran are incredibly flawed.  Throughout all of the tests, there was NO ACTION.  He fired his gun twice throughout that entire video and it was only against 1 enemy on the screen in Metro:LL.  I would have liked to see some multiplayer tests run in Crysis3 and BF4.  Again, I don't think OCing will give you very big performance gains, but that video is not a definitive "It doesn't do anything."

Yeah, I noticed there weren't really any very CPU intensive moments in his tests. 

I think the 3570k is able to handle those games pretty well still, but he didn't really try to put any real load on it. 

[SkilledRebuilds]

While I appreciate the video, and I do think that he is right that overclocking isn't going to give you very tangible results for gaming, the tests he ran are incredibly flawed.  Throughout all of the tests, there was NO ACTION.  He fired his gun twice throughout that entire video and it was only against 1 enemy on the screen in Metro:LL.  I would have liked to see some multiplayer tests run in Crysis3 and BF4.  Again, I don't think OCing will give you very big performance gains, but that video is not a definitive "It doesn't do anything."

I agree, was a timedemo essentially with zero action.

I'm sure in BF4 64p servers it makes a difference.

[Faceman]

I agree, was a timedemo essentially with zero action.

I'm sure in BF4 64p servers it makes a difference.

Right, and that is what I want to see.  Crysis3 can also be played online.

 

Both BF4 and C3 are incredibly CPU intensive, so until he runs those tests in an online setting it means nothing to me.  Also, he needed to incorporate more action into that video, there was barely any action, and hardly any stress being applied to the CPU, as far as I'm concerned, the guy wasted his time doing an inherently flawed test.  His head is in the right place and I'm sure he means well, but it doesn't prove anything.

[runit3]

I had a roughly similar experience with a system I built for one of my friends. Moved from a Z87-n (gigabyte itx) to a Z97i-plus (Asus itx) and moved up to a 1:1 4.5ghz OC at a slightly lower voltage. The z87 was stuck at 4.5/4.3 with a rather horrendous voltage of 1.27...didn't help that his i7 was a hot one either

 

Not sure if it was moving from Gigabyte to ASUS, or Z87 to Z97, either way can't really complain.

[Faceman]

I had a roughly similar experience with a system I built for one of my friends. Moved from a Z87-n (gigabyte itx) to a Z97i-plus (Asus itx) and moved up to a 1:1 4.5ghz OC at a slightly lower voltage. The z87 was stuck at 4.5/4.3 with a rather horrendous voltage of 1.27...didn't help that his i7 was a hot one either

 

Not sure if it was moving from Gigabyte to ASUS, or Z87 to Z97, either way can't really complain.

That would be interesting to find out if the brand had anything to do with it.  Up until today, I was under the assumption that you are going to get the same OC on any platform and that your result is 99% dependent on the chip you have.  Would be great if someone could do a motherboard roundup overclocking comparison.