Need expert help qith QX6700 and eVGA 680i rev a overclocks please!!!

My story with the EVGA 680i motherboard and a QX6700 is a strange one. Perhaps some people here can give me some insight.

System specs: eVGA 680i rev A2 (Quad core friendly)
Bios P27 (but everything I will write about occurs with P26 and P25 as well)
CPU: QX6700 A stepping
Ram 4G OCZ PC6400 Platinum XTC Rev2
SLi 8800GTX
SIlverstone Olympia 1000W power supply
DTec Fuzion WB, DDC2 with Petras Top, MCR320-QP rad with 3 Yate loon ML
All of the usual speed limiting parameters in the bios are off (speed step etc, GPU ex off also)

I am currently running at 3.5 GHz, FSB 1400 10x multi.
CPU Voltage is set at 1.5V in bios, the bios monitor shows 1.46V, and in windows, speedfan shows 1.46v idle and a drop to 1.42 under load.
Idle temperatures are at 39,39,33,35. Under load with orthos on two cores and prime 95 large fft's on another and prime 95 small ftt's on another temperatures are at 63 maximum over a 14 hour period. The system is very stable.
Ram is running unlinked at 800 mHz 4-4-4-15 2T at 2.1V (stock), but can also run at 850 prime stable.

All voltages are at auto. FSB is 1.4, SPP is at 1.5, MCP is at 1.5, and HT at 1.2.

Ok all indications from this is that I have a fine overclock, with no need to push any part of the system to high voltages.

However, something does not seem right to me.

My FSB is limited to 360 no matter what I do. I can sometimes boot into windows, but then the system will lock up even at a low cpu speed. (I've tried 1600,1601, 1625, 1650, 1700, 1750, 1800...all behave the same) I have increased all of the voltages to high levels (as documented on evga's site) and this does not change at all even with a low multiplier on the cpu. Linked vs unlinked memory makes no difference.

I can adjust the multiplier and FSB(Lower) to get my cpu up to 4 GHz but it is not stable at anything over 3.5GHz. At 3.8 - 3.9 GHz it can boot into windows but will usually lock up and require a shutdown. This is the same at stock voltages or with boosted voltages including the CPU up to 1.6V. Even at 1.6V I cannot get 3.6G stable.

If I try to increase the FSB to get 1600 and using a 9x multiplier, it will usually crash during windows bootup but if I do log in it will freeze. Even with lower multiliers. There seems to be no correlation at all with voltage selections in the bios.

One interesting point, is that the MCP temperature never seems to change in Speedfan no matter what the voltage. But this could just mean it is not actually measring the MCP temperature.

I have exactly the same issue with trying to overclock the ram. It is stable from 800-850MHz 4-4-4-15 2.1V but not at any other speeds such as 900 5-5-5-17 (2.2V) irrespective of the settings of the other voltages. Even with using optimum timings (as determined by the bios).

I am now wondering it there is the possibility that the voltages are actually not changing even though I set them in the bios, and the bios monitor shows that they have changed (although strangely, only the CPU voltage shows any kind of vdroop and the other voltages at all exactly the same as they are set).

It would seem that perhaps the voltage adjustment circuitry is doing nothing. Does this make sense? Is it possible that the CPU voltage can be changed but nothing else?

Or, is this just a characteristic of my particular CPU?

I would appreciate it if the experts here can shead some light on this.
Thanks!

Sorry I can't help but could you do me a favor? Put your ear by the CPU area and tell me if you hear a creeking/buzz sound when you move your mouse or when your computer is under load? I had this issue with my 680i's. :)

Nope, it's silent

well i get pretty much same behaviour also A1 with QX6700. mines an ES, is yours? but i see some change with voltages. MCP temp rises 1-2C with highest voltages on it. also I can improve stability by changing the various voltages though not always by increasing them, sometimes lower is better. but yeah ive hit a wall @ 385MHz, cant get stable any higher.

another strange thing I have 4x1gb sticks of OCZPC8000 XE. thought I would try to find 2 best sticks by testing each stick individually. so I put one in and got it to 1088MHz 4-3-3-x 2T with 2.2V. memtest stabel all day. I then set 1089MHz. no boot, just beeps as with memroy error. so I tried other 3 sticks and exactly the same, they are all stable 1088MHz 4-3-3-x but you cant boot even 1MHz higher. surely this is some sort of board limit and its very weird. didnt try slacker timings yet to see if can boot higher but this behaviour isnt right.

You know, it actually turns out that I see the same, I can get further sometime by lowering the spp voltage, but ultimately it doesn't allow stability above 360 in my case.

maybe you have already reached the max of your chip

This is not much of help but you are not alone having issues... ;)

I have a QX6700 ES. Mine is an B0.

When I run unlinked with RAM@800 I can run 400 and 401MHz FSB. -399 and +402 no go.

Also, I can not use multiplier 13. I do not have cooling for higher multiplier atm so I have not tested any higher.

And at 400MHz FSB I only have one or two cores in windows. Windows device manager shows four cores though.

I am pretty sure EVGA is about to look into if theese problems are related to ES-chips.


EDIT: My CPU can run @+480MHz FSB on the QuadGT with all four cores avaliable in windows.

Quote:

---

Originally Posted by aolforbroadband

Sorry I can't help but could you do me a favor? Put your ear by the CPU area and tell me if you hear a creeking/buzz sound when you move your mouse or when your computer is under load? I had this issue with my 680i's. :)

---

What did u do to fix it?

to give you some more perspective...blossa says he can get his QX6700 over 480 on the QuadGT...my Q6600 can not go over 420 FSB on the QuadGT...i really have to live at 380 FSB so the voltages are not overwhelming to hold an idle condition for most of the time.

i think board and cpu are equal co-conspirators in what you get out of it for max overclock...that is what stopped me from getting the A1 revision....i am no worst off by staying with the QuadGT...it might be worst with the EVGA 680i...it is the luck of the draw!...pure and simple!:)

PS...i have to remind myself that i have two conroe's onboard and therefore, i have twice or maybe 175% the computing power compared to most users who have a conroe.

Quote:

---

Originally Posted by Solarfall

maybe you have already reached the max of your chip

---

doubtfull for me since the same QX6700 does 500MHz on the P5B-D or I wouldnt have suggested its the board. ;)

no there is defo some isuue still with quad core and this baord. it exhibits some strange behaviour and same with ram clocking where there are these walls as well as FSB holes all over the pace.

thing is to OP its worth trying other FSBs. dont assume bacuse you cant run 361,362,365,370MHz etc that it wont be stable with for e.g. 405MHz. I need to test further as im seeing people like posted here getting 401MHz stable but not 402 or 399MHz. ive also heard of someone not stable at 410MHz but is at 420MHz.
and just to add im not knowcking this board or EVGA as I still think its one of the best SLI boards out there. I also think future BIOSes will imporve things. magic can happen with BIOS, DFI 975X proved that for sure.:D

Quote:

---

Originally Posted by t1no

What did u do to fix it?

---

yeah id like to know that too as i get this when its running certain prime calcs.

Well yesterday I played around a little more, and I definetly cannot go above 380 FSB stable. Furthermore I cannot get x10 multiplier for 3.6GHz at 360 (1440) FSB stable either no matter how much voltage I apply to the cpu or spp etc.
However! If I bring the fsb down to 340-350 I can get stability with an 11X multi giving 3.7-3.8Ghz stable! The only thing is that ram i/o as measured with everest goes from about 9400 mb/s to 78 mb/s but my 3dmark06 goes from 17300 to 18450! Also, now I need 1.57 volts instead of 1.5 (in the bios).

For the sake of longevity of my cpu, I am going to stick with the fsb 1400 x10 = 3.5 GHz 1.5V and a slightly lower 3dmark06.
For all practical purposes they are the same.

So many options, so little time