i7 860 Over-clocking

IceBear said:

Intel actually say you shouldn't run your chip at over 72DegC...

Click to expand...

Thought that was the max air temp inside the case @95W, rather than chip temp sensor reading, or did I misunderstand it?

EDIT:
Temp of case (CPU housing) max 72, not the air!
Max air temp for stock intel cooler 40C!!

humph said:

IceBear said:

Intel actually say you shouldn't run your chip at over 72DegC...

Click to expand...

Thought that was the max air temp inside the case @965W, rather than chip temp sensor reading, or did I misunderstand it?

Click to expand...

I could be wrong, but Intel specs are here: http://processorfinder.intel.com/details.aspx?sSpec=SLBJJ

There is also good discussion here: http://www.tomshardware.com/forum/267871-28-load

Lots of people running them hotter, but in theory shouldn't go above 72.7DegC; or a core temp of 78DegC.

Main point as I see it is that VCore should be max of 1.4V... This is lower than some guides on the web so best to be careful. AND how long is your chip going to run at that temp/load during normal usage? Mine ticks along nice and cold 99.95 of the time.

Just when I'm running mprime does it actually get very hot. Even processing lots of photos through lightroom, doing a batch import/export it doesn't get that hot.

what I don't understand is why on the mac side, performance drop 50% IF I run memory at 1600 and without nullCPU kext in /Extra/extension. Like others, I thought nullCPU is not need.

madguy said:

what I don't understand is why on the mac side, performance drop 50% IF I run memory at 1600 and without nullCPU kext in /Extra/extension. Like others, I thought nullCPU is not need.

Click to expand...

I'll make an attempt but I'm sure somebody out there will be better able to confirm/surmise than I.

My way of explaining this is that the OS fundamentally interacts with the components on the board, using information from the BIOS. Some things are abstracted by standards; SATA etc; others require more direct and specific integration like the CPU and associated instruction sets.

OSX requires us to use an interpretation layer between the OS and the BIOS, the DSDT. We can affect the way the OS sees the machine by changing some things within the DSDT but not everything.

When OSX determines what features are enabled on the CPU it is obviously doing so in a slightly different way to windows. Hence the fact that if you don't hard-code those features to ENABLED they will not be active IF AND ONLY IF you've made changes to the BIOS - LIKE CHANGING YOUR MEMORY SPEED.

It's weird but repeatable.

thnx for the explaination

I posted asking about the imac 11,1 smbios vs. using the DSDT to enable speedstep and then found this thread. Any consensus on differences in behavior:

1. Standard clock, enabling speedstep and turbo

2. OC, behavior with speedstep and turbo

My GB32 scores were around:

8000 with 10.6.2 (stock clocks and 1333mhz)
8800 with 10.6.3 (stock clocks and 1333mhz)
9500 OC'd to 3.3ghz with 1200 mhz memory
10500 OC'd to 3.36 ghz with 1600 mhz memory

Windows 7 seems to be 1000 pts lower (7800 stock) in 32 bit mode than 10.6.3 in 32 bit mode. Not sure how it compares when overclocked.

Unfortunately, I think I just blew my PS in the middle of Prime95 testing.

Currently running the MacProsSmc and Tony's old DSDT with audio and speedstep edits for the UD2.

Tim

cakemonster said:

I posted asking about the imac 11,1 smbios vs. using the DSDT to enable speedstep and then found this thread. Any consensus on differences in behavior:

1. Standard clock, enabling speedstep and turbo

2. OC, behavior with speedstep and turbo

Click to expand...

Not sure what you mean. Both scenarios work fine. Where you go wrong is OC without Speedstep in DSDT AND the 11,1 smbios.

timg said:

My GB32 scores were around:

8000 with 10.6.2 (stock clocks and 1333mhz)
8800 with 10.6.3 (stock clocks and 1333mhz)
9500 OC'd to 3.3ghz with 1200 mhz memory
10500 OC'd to 3.36 ghz with 1600 mhz memory

Windows 7 seems to be 1000 pts lower (7800 stock) in 32 bit mode than 10.6.3 in 32 bit mode. Not sure how it compares when overclocked.

Unfortunately, I think I just blew my PS in the middle of Prime95 testing.

Currently running the MacProsSmc and Tony's old DSDT with audio and speedstep edits for the UD2.

Tim

Click to expand...

Bugger about the power supply!

I think 10,000 is a good 32-bit score for the i7-860. Mine has been stable now for several weeks and seems very good for long term use. You can go higher but damn does it get hot quickly when you start upping vCore.

timg said:

My GB32 scores were around:

8000 with 10.6.2 (stock clocks and 1333mhz)
8800 with 10.6.3 (stock clocks and 1333mhz)
9500 OC'd to 3.3ghz with 1200 mhz memory
10500 OC'd to 3.36 ghz with 1600 mhz memory

Tim

Click to expand...

Not sure how much of this is 64 vs 32-bit or other system configuration, but with my system (i7 860 running 10.6.3 (64-bit) with stock clocks and slow memory (1066MHz) I get a GB64 score of just over 10000 (10050) It was just over 9000 on 10.6.2

Curious, are you running 32 bit OS or 64 bit OS and 32bit GB?