Does anyone have this motherboard and if so, I have a qx9650 and i need help with bios settings. There are way to many and I dont have a clue as to what some of them mean. I am trying to get a stable o/c, all I can do is 4.3ghz with 1.52v never goes over 65c b/c I just bought a Swiftech H20-220 compact and it seems to be doing the job but I cant get a stable configuration to go past 4.3ghz. I need bios settings and system tweaks. memory is gskill pc2-8000 so going over a 420mhz fsb is killing me. Maybe you guys know something I dont. Thanks for the replys.
Try here:
http://csd.dficlub.org/forum/showthread.php?t=6689
http://csd.dficlub.org/forum/showthread.php?t=6050
And the "Out Of Warranty" section gets rather good:
http://csd.dficlub.org/forum/forumdisplay.php?f=70
Just *how* high is that FSB!?!?
http://valid.x86-secret.com/show_oc.php?id=356943
Regarding your memory: You should be able to get the FSB significantly higher than that on a DFI board. So if your RAM is stopping you, set the memory divider so your RAM is UNDER~clocked. That will give you the headroom you need to raise your buss and work on your CPU. Once you have your CPU stable at it's target, *then* you can go back and tweak the RAM.
The order I would approach it is: (1) FSB (2) CPU Multi, and (3) Memory Work on only one thing at a time, and I think you'll find far fewer headaches.
You are getting 4.3Ghz from a 2.66Ghz processor? Seems like a pretty hefty overclock already. But, your memory is probably the biggest problem.
What Scotteq said.
Read you manual, very closely. All those settings on your DFI board will allow you to do some pretty big stuff, but you have to know what they all do.
If you need explanations try here:
The Definitive BIOS Optimization Guide
http://www.techarp.com/freebog.aspx
My memory is not holding me back. On evongugg, where do i look on that list of things, thanks, I have read the manual but it is chineese to me. LOL
Steeda - You should be able to get to a 500 (2000)Mhz Front Side Buss on that mother board with little trouble. But what happens is when you raise Front Side Buss speeds, it takes the memory along with it. You can do a 30~40% overclock on many CPUs. But memory?? Don't count on it. Not at all. I suggest you revisit your Memory Divider.
I can run it at 1150mhz, so I know its not the memory, can get 9500 mbs on everest across the board.
65C on water is quite high.
And it's a QX9650 so 4.3GHz is already a very good overclock without extreme cooling.
| Andrius wrote : 65C on water is quite high.
|
Agreed. I get a stable 4.1ghz on air with a TRUE on my QX9650 and the temp only goes to 48c loaded up on Prime95. By the way, that Swiftech compact cooler is not designed for heavy duty use. As far as the FSB problem goes, many motherboard exhibit "dead zones" which won't work regardless of what you do. A way around them is to use a lower multiplier and then a higher FSB.
As far as the ram goes, don't know what voltage you have it at, but 2.1v-2.2v seems to be common for higher overclocks. Balancing the ram divider is also a bit of a guessing game, as you bump it around in search of what works. Be aware that any ram clocked faster than DDR2-800 is overclocked ram, so what you're searching for is a sable overclock on the ram and not every set will go to the very high clocks.
| steeda wrote : I can run it at 1150mhz, so I know its not the memory, can get 9500 mbs on everest across the board. |
<shrug> Suit Yourself..
i know, do u think i did something wrong, looked at the cpu and the thermal paste looks good, do u think I tighten it down to tight
True, I here you on that, but getting back to my temps, do u think I did something wrong, b/c I am getting around the same temps with my tuniq tower except they are a little lower. Any advice?
Sorry, Scotteq, didnt mean to be rude.
Ok, i am running prime95 right now on it, 56c @ 1.18v shouldnt my temps be at like 40c with water cooling? Another note, my temps are always at 55c when i startup, sholdnt they be lower?
IMO the kit is hitting it's limits. I've read a few reviews and a Q6600 at 4.0GHz and 1.5V is infact just below 70C with this kit. So I don't think you did something wrong.
http://www.cluboc.net/reviews/supe [...] /page3.asp
i am at 70c at 1.42v in occt after 20minutes at 4.2ghz, this sucks b/c my tuniq tower did the same thing
Andrius, i read that same article, Im thinking the same thing. 10.5x400 is working quite nicely so far and 1.5v in the bios
Oh, Andrius, no reboots after i plugged the fans into the 5v motherboard and took them out of the 12v rail
^That's scary. But hearing some guy had 5V on the 12V rail because of an Antec tri-cool fan I'm not that surprised.
I have an antec tri-cool fan and it has a 12v attachment but i have it to my motherboard as well, have had my computer for 30minutes on everest stress test at 4.3ghz at 70c at 1.46v 1.51v in bios
As long as it's stable, the 70C is a bit high for my taste but I hate to see 50C on my Q6600.
If you need better performance look into the D-Tek Fuzion Waterblock and a Laing DDC pump (expensive DIY but no other way to get better performance).
http://www.bit-tech.net/modding/20 [...] aterblock/
1.5v seems like alot for a 45nm processor. I know it's a quad and it'll need a little more, but it still seems like alot. Maybe you just have a not so great OC'ing CPU.
^It's a QX9650.
What scares me is the Vdrop of 0.05V.
I'm used to such values from cheap boards and cheap PSUs not X48 and PC P&C. But I've never seen a QC at 4.3GHz before so I don't know...
http://valid.x86-secret.com/show_oc.php?id=361583
it cant be that bad
ok i have been running everest stress test for a hour now, i took off the vdroop control and my temps have dramatically increase and my system seems more stable now, it wont pass prime95, the first 2 cores die on me but my over all temp is staying at 61c at 1.46v, the voltage seems to not be bouncing as much and it actually lowers to an exceptable level, I have bench it 4 times now without a reboot as well, its at 4.3ghz, i think that vdroop control is messing with my computer b/c and this is honest, i dropped the vcore voltage 3 notches and noticed the voltage did not drop then i took the vdroop control to disable and the voltage finally dropped.
That looks quite nice. Is the 61C load the maximum?
Vdroop control could be such a usefull feature. Too bad they messed it up.
On a side note : Why are there 4 GPUs listed on that CPU-Z Validation?
Are you doing QuadCrossfire? That 750W psu just looks way more scary right now.
i know i saw that and wondered why it posted that, as well as my core is 900mhz and my memory is 1100ghz, that looks weird too, what is scary about the 750w psu
btw, my temps have dropped down to 40c idle, does this sound more reasonable
^Yes. That's a more like it. But it's still quite high for watercooling.
For scary I figured
2 HD 3870 X2 means ~500W max power draw
and that's a bit scary to have on a 750W and sharing it with a 4.5GHz quadcore the X48 and 2 Raptors.
Yeah, I know, that totally sucks, well I ran everest stress test over night and my computer was still running this morning, at 1.46v 61c at 4.336ghz so I think im going to stay with that, that is a good speed to use for daily use and the fact that it idles at 40-44c and stresses at 61c is an acceptable level to me, I scored around 22750-22900 on 3dmark06 and that vdroop control totally sucks for DFI, I cant believe that when I had it on that the voltage would never drop.
Does CPU VTT voltage play a part in how many volts are going to the cpu. I also noticed when I increased that voltage that it would make the cpu temp on everest go up, do u know what I am talking about everest shows CPU,CPU1,CPU2,CPU3, & CPU4---- when I raise the vvt voltage my CPU temp was 5c higher then my CPU1-4 and when i raised it and lowered my vcore voltage it would register the same voltage as before. The only time I could lower my voltage was when I lowered my vvt and my vcore voltage.
VTT or FSB termination is around 1.2V by default I think (not sure for 45nm chips). It plays it's part with every time the logic bus state changes (every value transfered between cpu and northbridge). Not really sure about the details. High VTT is a chip killer (anandtech lost at least one QX9650 to it).
Since VTT is some sort of reference voltage it's impact is likely much beyond that but I don't really know. You should only change it if you really need to IMO.
Vdroop: That's what Vdroop control is meant to do (prevent it drooping below Vcore - Vdrop). But it should stay a bit under Vcore (as it is intended to work) and protect the chip from voltage spikes.
Im here to tell you that vdroop conrol does not work on my motherboard and from what I am experiencing with DFI's motherboard, i like the fact that that i can take into account vdroop as it is what i am accustom to. Your right on the 9650, i read that article and have set my vvt to 1.10v, north bridge to 1.25v, southbridge to 1.51v and vcore to 1.47v. I am happy with a 4.336ghz o/c. I can use it everyday and not worry about it crashing now. I am just worried why my cpu is so hot even with water cooling. I thought that it would cool it down to at least 30c at idle and you always have that 20-25c rise in core temp when at load. I just have no clue why my cpu is so hot. The kicker of it all is that my Tuniq Tower 120 did the samething this thing is doing. Do you think I could possibly have a bad chip that heats up way to high?
Should I take my q6600 out of the box and hook up the water cooling to it. the only thing is that it would be some work and I have the q6600 already built and put in a case?
I don't think it's a bad chip. In fact it does awesome. Your H20 kit costs about $150 and it's the best kit by many accounts. If you need higher performance you have to do it DIY with probably a higher flow pump and a better waterblock design (d-tek Fuzion). A larger 120.3 radiator would also improve the heat dissipation and increase the turnaround time (thus giving the fluid more time to dissipate heat). The same could be achieved by a large fluidtank like with metal fins for dissipation (like that Zalman all in one thing).
I doubt the Q6600 will do much better (like you read in the reviews). 4GHz appears to be the thermal limit of this kit. You got 330MHz over that by using a 45nm chip. Leave the thermal compound a bit of time and see if it improves (arctic silver takes some 200 hours to set) and some pastes also take a few hours to a few days to cure properly. Zalman's paste is the exception.
Andrius, can you give me a link to the paste, i want to check it out, I did pull the heat sink off and noticed some areas where there wasnt paste so i filled them in, I just wait, anyways that is what the game is all about anyways, right?
Zalman STG
http://www.newegg.com/Product/Prod [...] 6835118010
Arctic Silver Ceramique (no idea if it's good)
http://www.newegg.com/Product/Prod [...] 6835100009
Arctic Silver 5 (I have their thermal glue and it's quite good)
http://www.newegg.com/Product/Prod [...] 6835100007
More items to choose from
http://www.newegg.com/Product/Prod [...] %2f+Grease
There are 928 identified and unidentified users. To see the list of identified users, Click here.
You are about to answer a thread that has been inactive for more than 6 months.
If you still wish to proceed, please ensure that your posting is original and does not duplicate or overlap any prior responses to this thread.

