stupid question?
- Thread starter nicnac1
- Start date
Hi Mate,
so you are you know at 3.4GHz (8x425) DDR2-1066 stable?
If so you got two things to try, the first is seeing if you can reduce the vCore a bit more at 3.4GHz? would be nice to know if you can get the chips voltage down?
The second is obviously to go for 3.6GHz (9x400) but your chip is maybe not gonna like without some huge vCore increase! . . . at least 1.55vCore Actual/Load if not more . . . . even with the improved cooling I think load temps will rocket!
I'm off out now for some drinks so I'll leave you to play, personally I would be trying to reduce the vCore at 3.4GHz first but I sense that idea bores you and you would prefere to gun for 3.6GHz!
so you are you know at 3.4GHz (8x425) DDR2-1066 stable?
If so you got two things to try, the first is seeing if you can reduce the vCore a bit more at 3.4GHz? would be nice to know if you can get the chips voltage down?
The second is obviously to go for 3.6GHz (9x400) but your chip is maybe not gonna like without some huge vCore increase! . . . at least 1.55vCore Actual/Load if not more . . . . even with the improved cooling I think load temps will rocket!
I'm off out now for some drinks so I'll leave you to play, personally I would be trying to reduce the vCore at 3.4GHz first but I sense that idea bores you and you would prefere to gun for 3.6GHz!
I wouldn't mind seeing an 8 hour prime shot or similar using lowered volts, if you have an accurate vCore figure for 3.4GHz it helps us work out how the chip is scaling and therefore may help working out if 3.6GHz is a possible.
If you wanna cut to the chase then just crank it up to 3.6GHz with all the case fans on full etc and see if you can get at least a 1 hour run under Prime95 Small FFts!
If you wanna cut to the chase then just crank it up to 3.6GHz with all the case fans on full etc and see if you can get at least a 1 hour run under Prime95 Small FFts!
i would post a screeny for you but i'm away again for this weekend. 1.43v under load seemed to cope fine with small FFTs and blend so i imagine we could even go a little lower.
i'm tempted to go for 3.6 when i get back but i'm not sure if the temps are going to go out of control. unless there's something else limiting it and causing it to fail. vNB is 1.47 btw.
with 3.4ghz running at 1.43v under load @ 63C, does 3.6GHz look possible??
Either way, so far we've eeked another 0.1GHz out of the processor and 200MHz out of the RAM. Even that has increased the average FPS in FSX by 5! (which makes a huge difference in FSX)!!!
i'm tempted to go for 3.6 when i get back but i'm not sure if the temps are going to go out of control. unless there's something else limiting it and causing it to fail. vNB is 1.47 btw.
with 3.4ghz running at 1.43v under load @ 63C, does 3.6GHz look possible??
Either way, so far we've eeked another 0.1GHz out of the processor and 200MHz out of the RAM. Even that has increased the average FPS in FSX by 5! (which makes a huge difference in FSX)!!!
Or you could add another string to your bow and lap the HS and CPU? (it really isn't difficult - just very disconcerting, at first
.)It will take a couple of hours, for the 2, and you'll wonder what the hell came over you, when you're about half way through the CPU, but it could make all the difference in reaching 3.6GHz.
Obviously, test if it will do 3.6 in its present unlapped state first. But if it fails (and continues to do so) - i, personally, think that lapping is worth considering as this has been an epic effort on yours and Big.Wayne's part and *perhaps* warrants the slight risk. (Your call of course - and no one would blame you if you decided against it as it voids any warranty and runs the slight risk of killing your CPU.)
EDIT: Having typed all that - 425 x 8 (3.4GHz) with memory @ 1065 is a very respectable conclusion considering how this thread started. You could content yourself with this, if 3.6 doesn't work out, as Big.Wayne achieved your primary goal, which was getting the most out of your memory - he's done that with gold plated knobs on.
Last edited:
Yeah I'm glad we made some progress! 
It does seem likey it's the chip hold us back, it seems to be needing a lot of voltage to unlock the extra MHz which is par for the course unless one is lucky and gets a chip that flies! . . . I'm pretty sure if we pumped enough voltage through the Chip it would reach 3.6GHz although the problem is keeping it cool!
I'm happy to push on for a bit because I think nicnac1 will enjoy finally hitting 3.6GHz although I don't think this speed will be for day to day running but will will nice for a few benches etc . . . . after which the clock can be reduced to something which requires less juice!
Looking at where the system is now the main factor that I see in reaching 3.6GHz is vCore, dropping the multi when changing from 8x425 to 9x400 is actually taking some strain off the northbridge
8x425
9/8= 1.125
1.125 x 425= 478 NBCC MHz
9x400
9/9= 1
1 x 400= 400 NBCC MHz
In the first example (8x425) the NBCC is running nearly 20% faster than the second example (9x400) so if anything you will find that the northbridge voltage (vNB) could actually be reduced when trying for 3.6GHz, heh that's counter-intuitive but that's overclocking for ya!
I think what we need is a freezing cold night, get nicnac1 into his Thermal-gear, get the fans on full speed, get the windows open and get some freezing temps thrown into the mix so that the vCore can be set mega high for a nights testing!
[edit]
I'm also looking at the BIOS and there is still a few more voltages (CPU vTT/GTLREF etc) that may be worth tweaking if the vCore path leads to a dead end!
It does seem likey it's the chip hold us back, it seems to be needing a lot of voltage to unlock the extra MHz which is par for the course unless one is lucky and gets a chip that flies! . . . I'm pretty sure if we pumped enough voltage through the Chip it would reach 3.6GHz although the problem is keeping it cool!
I'm happy to push on for a bit because I think nicnac1 will enjoy finally hitting 3.6GHz although I don't think this speed will be for day to day running but will will nice for a few benches etc . . . . after which the clock can be reduced to something which requires less juice!
Looking at where the system is now the main factor that I see in reaching 3.6GHz is vCore, dropping the multi when changing from 8x425 to 9x400 is actually taking some strain off the northbridge
8x425
9/8= 1.125
1.125 x 425= 478 NBCC MHz
9x400
9/9= 1
1 x 400= 400 NBCC MHz
In the first example (8x425) the NBCC is running nearly 20% faster than the second example (9x400) so if anything you will find that the northbridge voltage (vNB) could actually be reduced when trying for 3.6GHz, heh that's counter-intuitive but that's overclocking for ya!
I think what we need is a freezing cold night, get nicnac1 into his Thermal-gear, get the fans on full speed, get the windows open and get some freezing temps thrown into the mix so that the vCore can be set mega high for a nights testing!
[edit]
I'm also looking at the BIOS and there is still a few more voltages (CPU vTT/GTLREF etc) that may be worth tweaking if the vCore path leads to a dead end!
so is it worth playing with all of the voltages above? see if we get any joy?
I'm back home tonight so i'm going to try the push to 3.6, (windows open and ski-jacket on)! Looking forward to seeing what we can get out of it
I think i'll leave the VNB as it is for the time being and try reducing it if we get anywhere near stable.
Like you say, it would be a shame to not reach our target even if it cannot be left like that for long. I'd certainly be very happy with that achievement and I've learnt so much already!
To pick up on Plec's point, I've been looking around online at lapping and it doesn't look too hard. What are the risks? and will it make much of a difference to the temps?
I'm back home tonight so i'm going to try the push to 3.6, (windows open and ski-jacket on)! Looking forward to seeing what we can get out of it
I think i'll leave the VNB as it is for the time being and try reducing it if we get anywhere near stable.
Like you say, it would be a shame to not reach our target even if it cannot be left like that for long. I'd certainly be very happy with that achievement and I've learnt so much already!
To pick up on Plec's point, I've been looking around online at lapping and it doesn't look too hard. What are the risks? and will it make much of a difference to the temps?
If your clock is thermally limited then lapping should certainly be a 'consideration' – as you can reduce temps as much as ~10 degrees when lapping both the CPU and heatsink. Dramatic changes are not guaranteed though – sometimes the difference is only a few degrees and may not prove to be the aspired clock’s salvation. It’s a relative ‘shot in the dark’ but one that will affect your temps beneficially either way – but to what degree is an unknown.
However, if after lapping your temps are dropped significantly you *may* then find yourself voltage limited – but the lapping will, at least, have assisted in more respectable temps with an already very solid clock. But, this is whole different issue and can only be addressed once/if it presents itself.
I don’t know what Big.Wayne’s take on the present situation is but from your last post lapping *may* have a significant impact on your next lot of settings (voltages permitting)…
As I’ve mentioned before, if done correctly and all necessary precautions are taken, there is very little risk involved in lapping – but it doesn’t stop you from having a tense 2 or 3 hours while you're sanding both components.
Last edited:
Hi Guys,
sorry for my absence but I've been running around with work etc
I'm not 100% sure what the next move is but if the problem isn't heat or vCore related we will need to look at the other voltage settings.
I'm not sure why the vNB is currently 1.44v? is that hand set or is that BIOS [Auto] . . . it shouldn't need that much, tempted to stick that on [Auto] if it isn't already.
I think the next thing to try adjusting CPU vTT, if thats on [Auto] I would like to know what voltage it is running at, that may need manually boosting from 1.2v up to 1.3v or even 1.4v. I've only had to use CPU vTT a few times and that was mainly when trying for a 500MHz-FSB so I'm not sure it will help us here? but its certainly worth a try. Adjusting vTT is a brute force approach, we also have the options of adjusting CPU GTLRefs which is a more fine tuned approach but a lot more complicated when your new.
nicnac1, give that CPU vTT a nudge and see if it helps, slowly but surely
Also nicnac1 when you post please include as much settings info as you can, its quite hard work for me to know what settings you have adjusted etc
Feel free to freestyle it a bit, you know roughly what your doing now and you have a few good testing apps there so have a play and see what ground you can break. If 3.6GHz doesn't come then experiment what is the next best setting you can get stable.
Lapping is certainly an option but I'm not sure if it will do the trick, it was also devalue your chip should you choose to sell it heh!
Overclocking can be tough at times and you end up banging your head against a wall, no matter what you do you can't reach the target speed. The annoying thing is that if and when you swap chips out the new chip gets installed and overclocks like a beast with loads less voltage that the previous one and you think "What the heck did I just spend three weeks doing!"
The doubly annoying thing is you never know if the chip you have is good or bad, you have to assume its good and the overclock settings are wrong but how can one know this? . . . the answer is you can't until you test another chip, chicken and egg heh!
sorry for my absence but I've been running around with work etc
I'm not 100% sure what the next move is but if the problem isn't heat or vCore related we will need to look at the other voltage settings.
I'm not sure why the vNB is currently 1.44v? is that hand set or is that BIOS [Auto] . . . it shouldn't need that much, tempted to stick that on [Auto] if it isn't already.
I think the next thing to try adjusting CPU vTT, if thats on [Auto] I would like to know what voltage it is running at, that may need manually boosting from 1.2v up to 1.3v or even 1.4v. I've only had to use CPU vTT a few times and that was mainly when trying for a 500MHz-FSB so I'm not sure it will help us here? but its certainly worth a try. Adjusting vTT is a brute force approach, we also have the options of adjusting CPU GTLRefs which is a more fine tuned approach but a lot more complicated when your new.
nicnac1, give that CPU vTT a nudge and see if it helps, slowly but surely
You mean 400x9 right? . . . . Get the memory running at [1:1] sync when your trying for 3.6GHz (9x400) so its running at 400MHz (DDR2-800) keeps things simple.right... went for 3.6 (400x8 ddr2-1000) kept failing prime small FFTs so racked the vcore up to 1.54 under load. temps are pushing 85c and it's still failing. don't think i can push the vcore any furthervnb is at 1.44v
Also nicnac1 when you post please include as much settings info as you can, its quite hard work for me to know what settings you have adjusted etc
Feel free to freestyle it a bit, you know roughly what your doing now and you have a few good testing apps there so have a play and see what ground you can break. If 3.6GHz doesn't come then experiment what is the next best setting you can get stable.
Lapping is certainly an option but I'm not sure if it will do the trick, it was also devalue your chip should you choose to sell it heh!
Overclocking can be tough at times and you end up banging your head against a wall, no matter what you do you can't reach the target speed. The annoying thing is that if and when you swap chips out the new chip gets installed and overclocks like a beast with loads less voltage that the previous one and you think "What the heck did I just spend three weeks doing!"
The doubly annoying thing is you never know if the chip you have is good or bad, you have to assume its good and the overclock settings are wrong but how can one know this? . . . the answer is you can't until you test another chip, chicken and egg heh!
Like your chicken-egg analogy. Shows why forums like this exist, this certainly isn’t easy!
Unfortunately it’s not possible to have 1 particular voltage on auto. If you want to manually change a voltage then the rest have to be on manual as well. Bit of a pain
I did what you said and increased the the VTT from 1.30v to 1.35v (highest it will go with VNB at 1.37v)
Prime blend ran for a whole 25 minutes before bluescreening although temps only reached 65C (not as bad as i had expected)
Small FFTs ran for 30seconds before an error occurred (also the temps rocketed!)
Best result so far lol!
Also ran FSX for an hour or so and it seemed to work fine...
Here are the details: (please let me know if you’d like me to include any more in future)
FSB: 400
Multi: 9
Mem: 1:1 (DDR2-800)
Vcore (bios): 1.6250v
Vcore (load-actual): 1.540v-1.550v
VTT: 1.35v
VNB: 1.37v
Temps Idle: 40, 38, 37, 37
Temps load (PrimeBlend): 65, 65, 64, 64
Do you reckon pump some more Vcore?
Unfortunately it’s not possible to have 1 particular voltage on auto. If you want to manually change a voltage then the rest have to be on manual as well. Bit of a pain
I did what you said and increased the the VTT from 1.30v to 1.35v (highest it will go with VNB at 1.37v)
Prime blend ran for a whole 25 minutes before bluescreening although temps only reached 65C (not as bad as i had expected)
Small FFTs ran for 30seconds before an error occurred (also the temps rocketed!)
Best result so far lol!
Also ran FSX for an hour or so and it seemed to work fine...
Here are the details: (please let me know if you’d like me to include any more in future)
FSB: 400
Multi: 9
Mem: 1:1 (DDR2-800)
Vcore (bios): 1.6250v
Vcore (load-actual): 1.540v-1.550v
VTT: 1.35v
VNB: 1.37v
Temps Idle: 40, 38, 37, 37
Temps load (PrimeBlend): 65, 65, 64, 64
Do you reckon pump some more Vcore?
Oh that's strange? . . . I thought all your volts apart from vCore had been on [Auto]
I haven't used your motherboard before but on all the ASUS motherboards you can mix n match your voltage settings, some on [Auto] and the others on manually selected ones, hmmm . . .
That's interesting, I've normally been able to selected any vTT regardless of what the others voltages are?
Are you able to increase vTT more than 1.35v if you first pump up the vNB?
So do you think that increasing the CPU vTT helped stability? . . . If in doubt reset CPU vTT to default volts and test again to see if it errors just a quick, if it does error fast with reduced vTT but definitely lasts longer with increased CPU vTT we may be onto something!
Are you able to increase vTT more than 1.35v if you first pump up the vNB?
QUOTE]
exactly, if I increase the vNB more vTT options become available.
I'll try resetting the vTT tonight and see if we fail quicker.
It looks like we're tantalisingly close!!!!!