Overclock @ 4.2ghz gives Lowest GB score then normal 3.4 ghz

[elracador]

HI,

I've just made an hackintosh for a friend who needs to have Snow Leopard 10.6.8.

DSDT Install with Multibeast 3.10.
Chimera 1.9.2

Here's the specs:

CPU: i7 980 (x6) 3.4 ghz
RAM: 24 Gb at 1600 Mhz
GPU: Sapphire Radeon 6850 1 go
SSD: 120 GB OCZ Agility 3
RAID 0: 8 TB ( 4x 2 Tb)
BACK-UP: 6 TB ( 2x 3TB)
PSU: Cooler Master Pro Gold 1000 Watt

Geekbench scores:

@ 3.4 ghz = 15.400 points
@ 4.2 ghz = 8.200 points ????

I had a similar issue on another hackintosh on Lion 10.7.4 but was solved using the SSDT option on Multibeast for Lion.
Unfortunatly this option is absent on the Multibeast 3.10.1 for Snow Leopard.

How can I fix this to get a more reasonable 18.000 GB score???

thanks in advance

 

[futan]

Your overclock isn't stable. Try higher voltage.

 

[Asus-Freak]

Your overclock isn't stable. Try higher voltage.

If you don't mind I'd love to hear your explanation for this opinion...

 

[futan]

If you don't mind I'd love to hear your explanation for this opinion...

When an overclock isn't quite stable, it will cause errors and have to repeat what it just calculated. Therefore slower performance. If the OP can boot into Windows and run Intel Burn Test or Prime95, I can pretty much guarantee that it will not pass.

It could also be the PSU not being able to provide the extra power needed from the overclock and throttling. However I kind of doubt that because everyone goes overkill with their PSU.

 

[elracador]

Hi,

I'm gettin a stable 19000 points @ 4 Ghz Overclock.

but when I do a "Carbon Copy cloner" of the Mac OSX disc and then
reinstall the DSDT and the drivers I will get the 8000 point again on Geekbench
and the USB dongle Bluetooth doesn't work.

any explanation?

But What do you mean by "try an higher voltage"??

Should I try 4.5 ghz instead of 4.2 or even higher ( like 4.8 ghz).
Isn't it too much for the i7 980??
Isn't it dangerous for the CPU life? ( it will die much sooner doesn't it??

thanks

Thanks.

 

[Ommudra]

"What do you mean by 'try an higher voltage'??"

Voltage is the electric potential difference between two points in a circuit. It's measured typically in Volts (symbol V). It is not measured in Hertz (mHz/gHz). Increasing the CPU voltage increases the amount of current flowing to your CPU from the power supply (also increasing heat generated, and wattage required).

For overclocking an LGA1366 including the Core i7 980, this guide is great. It gives great tips for optimizing voltage, etc., which can be applied to any chip. http://vip.asus.com/forum/view.aspx..._id=1&model=P6T+Deluxe&page=1&SLanguage=en-us

This tweaktown guide shows a 4.2gHz oc can be stabilized for an i7 980 hex at 1.36V.
http://www.tweaktown.com/reviews/3177/intel_core_i7_980x_32nm_lga_1366_six_core_cpu/index4.html

For those Overclocking Ivy Bridge, I recommend the following:
AnandTech's Ian Cutress wrote a detailed, quantitative article on voltage and clock relationships for Ivy Bridge chips, and how the 22nm process and die shrink significantly raises the heat generated by the CPU in comparison to Sandy Bridge, "Undervolting and Overclocking on Ivy Bridge": http://www.anandtech.com/show/5763/undervolting-and-overclocking-on-ivy-bridge

Patrick Schmid and Achim Roos wrote an even more detailed report on Ivy Bridge overclocking "Digging Into Ivy Bridge's Overclocking Issues":
http://www.tomshardware.com/reviews/ivy-bridge-overclocking-core-i7-3770k,3198-4.html

I suggest reading them in full. Based on the Cutress article, you should not need to turn up the voltage from the mother board's stock voltage setting, which is about 1.15V-1.2V if I'm not mistaken. In fact, one should be able to achieve 4.2gHz undervolting your chip to as low as 1.0V. Cutress suggests "1.15 V at 4.6 GHz is a happy medium."

 

[Ommudra]

"Isn't it dangerous for the CPU life? ( it will die much sooner doesn't it??"

As long as the voltage and thermals are within range and not extreme, you shouldn't notice a difference. Think about it - let's say your longevity decreases from 12 years to 8 years. Are you really not going to replace the chip in eight years? And if you did, CPU's depreciate so you'd be looking at what? $10 for that chip in eight years if you wanted the same one. Further, the longevity may be based on either constant usage or more than 12 hours a day. If you're using the CPU for 4 hours a day that would in theory triple longevity anyway. The only practical longevity issue in oc'ing a cpu i've heard of is, well, frying the chip.