Help me push 2500k past 4.1Ghz

[grandpatzer]

This is what I have achieved so far:

my Mobo has LLC: Disabled, 25%, 50%, 75%, Enabled.
I'm assuming Enabled is 100%(?)

there is vcore in HWINFO and it's simply called Vcore, however this motherboard and 2500k that I today installed seems to always have too low readings
When putting Vcore to AUTO it reads MIN 0.600v, MAX 0.848v and now in IDLE Current: 0.632v

So far I have had BIOS DEFAULT and just increasing multiplier to 42x and IBT (Intel Burn Test) crashes after just 2 runs.

41x is 100% stable at this stressing: 2hour small FFT Prime95 + 20runs IBT 2048mb (High) stable with in BIOS everything at DEFAULT besides the multiplier.

I did try following: LLC:75%, Vcore 1.45v, 44x, Interna PLL overvoltage enabled. Everything else is DEFAULT I think the default PLL is 1.80v.
First time running IBT it passes, 2nd time I try IBT 2048MB it crashes after just 2 runs, warning me that PC not stable.

 

[madness777]

Leaving settings on auto can be very dangerous! Do some ram testing with memtest, see if that's giving you problems...

 

[airdeano]

i dunna know if you have all the right components configured.. like:

set multiplier to 43
go to advanced CPU core features> enable Ratio Change in OS option

C1E-DISABLE
EIST-DISABLE
C3, C6 States-DISABLE
CPU Thermal Monitor-DISABLE

reboot

4.0-4.3GHz 1.300v-1.325v should be the voltage spectrum to run with, verify CPU VCORE.

it should reboot and start windows.. once there now the fun of Vcore(VCC) and LLC, and VTT(VCCIO) voltages..

airdeano

 

[grandpatzer]

i dunna know if you have all the right components configured.. like:

set multiplier to 43
go to advanced CPU core features> enable Ratio Change in OS option

C1E-DISABLE
EIST-DISABLE
C3, C6 States-DISABLE
CPU Thermal Monitor-DISABLE

reboot

4.0-4.3GHz 1.300v-1.325v should be the voltage spectrum to run with, verify CPU VCORE.

it should reboot and start windows.. once there now the fun of Vcore(VCC) and LLC, and VTT(VCCIO) voltages..

airdeano

I have to try another motherboard as the one I have now shows wery low voltage

 

[madness777]

Upgrade the BIOS if you can...

 

[airdeano]

I have to try another motherboard as the one I have now shows wery low voltage

was wondering, the d2h isn't very strong with small phasing power.
how is the stock clock performance?
i got a spare Z68-UD4 to try if you are local..

airdeano

 

[Master&Puppet]

HWinfo64 doesn't give you the vcore figures. That might be causing some confusion:

HWinfo64 gives you VID (Voltage ID) which is what the CPU is requesting. When you get vdroop you don't get what you request, you get vcore instead. VID - Vcore = Vdroop.

It would help if you could take screenshots of your BIOS so that we could see what you have to play with.

M&P

 

[grandpatzer]

Upgrade the BIOS if you can...

I'll have to look into this, I'm thinking either my Mobo or CPU is defective, can't go past 4.1ghz and performance is bad.

was wondering, the d2h isn't very strong with small phasing power.
how is the stock clock performance?
i got a spare Z68-UD4 to try if you are local..

airdeano

I'm sorry I forgot to mention I'm running on a Sapphire Pure P67, it has 8pins for CPU and I think It's alot better then the d2h.

Also I live in Sweden so not quiet local :lol:

HWinfo64 doesn't give you the vcore figures. That might be causing some confusion:

HWinfo64 gives you VID (Voltage ID) which is what the CPU is requesting. When you get vdroop you don't get what you request, you get vcore instead. VID - Vcore = Vdroop.

It would help if you could take screenshots of your BIOS so that we could see what you have to play with.

M&P

Actually there is a sensor in HWinfo it's more down on that picture it's not shown but it should be more down, it shows same vcore value as CPU-Z

 

[Master&Puppet]

Actually there is a sensor in HWinfo it's more down on that picture it's not shown but it should be more down, it shows same vcore value as CPU-Z

You win I missed that somehow.

To be honest it should be relatively simple to overclock beyond 4.1. Your Vcore is very low as you mentioned.

 

[madness777]

Would be fun if you measured the actual core voltage with a DMM on the capacitors on the other side of the motherboard, so you can be completely sure!
You ready for it or no?

 

[grandpatzer]

Upgrade the BIOS if you can...

Ok, guys I finally have some time to spend with my baby 2500k
so far up to 45x but so far only tested 5min of Prime95.

the newest BIOS is from 2011 and I assumed my mobo was shipped with newest, well it was not.
After BIOS update the software in windows7 still does not report correct VCORE but there is HWmonitoring inside the BIOS itself.

Now I don't remember if this vcore monitoring was available always but when I put BIOS Vcore to 1.49v then BIOS HWmonitor reads it as: 1.384v.

I run Prime95 64bit 27.7 Build2 AVX for just 5min for now and it reports succes no errors.

My core temperature is: 70c, 71c, 78c, 73c.
So after 5min Prime95 hottest core is 78c, cooler is a Mugen 2 Rev B with push pull I think the fans are 500 or 800rpm depending on settings.

The Mugen is maybe 1-4c worse compared to the Noctua.

 

[Master&Puppet]

Are you using small ffts for the prime test? Blend takes about 20 mins to get hot. Intel burn test is another one which which will get you quickly to temperature and ballpark stability.

The reason why your vcores are reading differently is because what you set in BIOS is an upper limit, not a target. Read the second article in my sig to find out more

 

[Mrt1989]

Vcore of 1.49 is way too high imo , you should get 4.5 with far less volts.
Mine is @4.4ghz 1.272v under load. 4.5 needed 1.320v.

 

[233Kosta]

Have a look at your LLC settings, there might be a substantial amount of Vdroop present. I'd consult M&P on whether that would be better than just leaving it at 1.49, he's done the study on it.

Just an off topic question, has anyone else noticed a trend with 2500Ks where core#2 is between 5 and 10 *C hotter than the rest?

 

[airdeano]

Vcore of 1.49 is way too high imo , you should get 4.5 with far less volts.
Mine is @4.4ghz 1.272v under load. 4.5 needed 1.320v.

im with you on the VCORE, but the chip he has is used and was run through
the wringer as a folding pc, so it might take a bit more to achive speed than
an unabused CPU.

he's been struggling to get it to be at least stabile, but 5minutes won't tell
much other than it can hold something.

airdeano

 

[233Kosta]

he's been struggling to get it to be at least stabile, but 5minutes won't tell
much other than it can hold something.

airdeano

From my experience, if a chip doesn't fail within the first 30s of Large FFT prime testing, it has a decent chance of remaining stable if the temps are kept in check. That being said, it's possible for the chip to require a nudge or two on the Vcore just to further stabilize it, but a ball park figure is easily achieved in that manner, made easier by the likes of TurboV and equivalent software.

I do however see your point about properly abused CPUs. Maybe the chip has degraded too far to be able to sustain the clock at all :sad: