PCI-E Voltage question on 680i

[jolly]

My rig is

evga 680i p28
e6600 1.55v
teamgroup extreem c3 @ 1006 - 5/5/5/18 2.2
galaxy 1000w
alphacool basic dual rad wc loop
2x evga 7950gt ko superclock 512 @ downclocked to 560/700(1400)

heres a screenie



First off I was wondering about what I should set PCIe frequency to in bios, secondly I was wondering why the pcie v is red (im not totally noob, but to this I am, its something ive never looked into)

My cpu does require 1.55 to get it to 3.4 and its pretty stable (well it is stable).

Oddly I was wondering why i keep getting crashes in css after my machine has been on for over 12 hours, I usually turn it off for 5 mins and itll run fine for another 4 or 5 hours. As i noted the gfx are downclocked because clocking them at the factory OC causes lockups (not artifacting!) Its a strange problem and I was wondering if anyone could shed any light /Edit: both cards are recent rmas as the old ones got shorted out

Also think i might have a slightly duff clocking cpu, no amount of voltage will get it to 3.6

 

[raja]

a number of things that could be at play here;

NB core voltage and vdimm needs to be set right and within safe limits, whilst considering overall temps..

Some of the newer stepping Conroe's are speed binned, so top out around 3.5ghz. Nothing to worry about in reality, unless benchmarking to the nth degree. As long as you can stay away from influence beyond reality or sensible thinking that you really have been stuffed by not having an extra couple of hundred mhz, often no real need other than personal satisfaction that one has not been screwed over, or has an equal level of bragging rights to anyone else.

PCIE voltage can help add some stability to the system at times, often an increase of 0.1v is sufficient.

With regards to Orthos, try running the blend test as your settings are now and note any failures, small fft's target specifically one area more than any other, not reflecting how the NB is going to handle full 3d load when every bus is being stressed simultaneously. The blend test is quite a heavy load between CPU/NB /RAM, whilst full 3d stability still requires some additional work on some Conroe boards at various fsb/multiplier combinations.

Anything outside this is usually driver related or patch issues, which should have been disqualified by you already...

Hope this helps a little,

regards
Raja

 

[jolly]

Heh, i know this sounds daft, but I really dont like running the blend test because it makes some component (unidentified, im guessing its my cpu or nb) make a very loud flickery clicking squealing squeaking type noise, which doesnt stop for the duration of testing, so i usually stop the test about 5 mins in

btw i got my conroe back in december

 

[raja]

well thats up to you, but running small fft's in your situation is obviously not giving you the full picture, certainly not close enough for you to nail the issue. The noise is just inductor squeal....


December may have just caught the beginning of the speed binning...

regards
raja

 

[jolly]

its 3 hours stable now, im going to leave it going till this time tomorrow night, that should give me a good indicator