stupid question?
- Thread starter nicnac1
- Start date
Ok cool, so reboot into BIOS and manually dial-in 1.35v for MCH (aka Memory Controller Hub, also know as Northbridge , vNB etc) save & exit and enter BIOS again to check Health Page and see if the MCH voltage is raised as you expected, if so then work your way up from 333MHz-FSB straight to 340MHz-FSB and then just keep raising the FSB 5MHz at a time trying to get to 355Mhz (or beyond). If that doesn't work boost the vNB up to 1.4v and try again, you can go up to 1.55vNB* for this testing process, keep an eye on the NB temps which should be ok but may rocket under load!
If the 355MHz attempt isn't happening with memory at [1:1] sync then change it again to [3:2] and try to work your way up from 333MHz to 355MHz remembering to start from about 1.35v on the Northbridge, see if this gives us a bit more ground!
Your board has offficial support for 333MHz-FSB (1333MHz System Bus) so anything above that is not guaranteed, having said that when you started this thread you were already running 368MHz-FSB (1472MHz System Bus) so that kinda points out there is some headroom, maybe 355MHz is a bad number and we may need to try a bit higher or lower than this.
If you select higher than a 355MHz FSB at the moment then don't forget to keep the memory under DDR2-1066 speeds as otherwise you will be overclocking it which hasnt been tested yet!
* Work your way up the vNB (1.35v, 1.40v etc), don' just whack it straight up to max!
If the 355MHz attempt isn't happening with memory at [1:1] sync then change it again to [3:2] and try to work your way up from 333MHz to 355MHz remembering to start from about 1.35v on the Northbridge, see if this gives us a bit more ground!
Your board has offficial support for 333MHz-FSB (1333MHz System Bus) so anything above that is not guaranteed, having said that when you started this thread you were already running 368MHz-FSB (1472MHz System Bus) so that kinda points out there is some headroom, maybe 355MHz is a bad number and we may need to try a bit higher or lower than this.
If you select higher than a 355MHz FSB at the moment then don't forget to keep the memory under DDR2-1066 speeds as otherwise you will be overclocking it which hasnt been tested yet!
* Work your way up the vNB (1.35v, 1.40v etc), don' just whack it straight up to max!
Last edited:
Ok cool, so reboot into BIOS and manually dial-in 1.35v for MCH (aka Memory Controller Hub, also know as Northbridge , vNB etc) save & exit and enter BIOS again to check Health Page and see if the MCH voltage is raised as you expected, if so then work your way up from 333MHz-FSB straight to 340MHz-FSB and then just keep raising the FSB 5MHz at a time trying to get to 355Mhz (or beyond). If that doesn't work boost the vNB up to 1.4v and try again, you can go up to 1.55vNB* for this testing process, keep an eye on the NB temps which should be ok but may rocket under load!
If the 355MHz attempt isn't happening with memory at [1:1] sync then change it again to [3:2] and try to work your way up from 333MHz to 355MHz remembering to start from about 1.35v on the Northbridge, see if this gives us a bit more ground!
Your board has offficial support for 333MHz-FSB (1333MHz System Bus) so anything above that is not guaranteed, having said that when you started this thread you were already running 368MHz-FSB (1472MHz System Bus) so that kinda points out there is some headroom, maybe 355MHz is a bad number and we may need to try a bit higher or lower than this.
If you select higher than a 355MHz FSB at the moment then don't forget to keep the memory under DDR2-1066 speeds as otherwise you will be overclocking it which hasnt been tested yet!
* Work your way up the vNB (1.35v, 1.40v etc), don' just whack it straight up to max!
Morning Big Wayne,
After a few hours testing last night i've not got much good news to report.
I tried all the methods above but it always bluescreened during prime95.
Vcore is still only 1.3 though, should that be upped? Also ddr2 voltage was 1.8
upped the MCH to 1.44 but only got as far as 345MHz FSB without it being unstable...
how come i was getting it stable at 368MHz before? what are we doing differently?
is it worth getting it back to the stable 3.31Ghz OC and working up from there? I'll leave it to your expertise though
No worries!
from the sounds of it I am thinking the problems we are encountering are chipset/northbridge related. I don't think its the processor and I don't think its the memory although there is an outside chance it could be a combo between the memory and the chipset.
If the chip was undervolted I would expect the computer to just reboot without a blue screen and if it was just the memory you may have troubles booting windows or just have randon things crashing on your desktop.
If you want to eliminate the CPU and Mem then you can boost the vCore so that it reads between 1.4v and 1.5v in CPU-z/uGuru-BIOS and the same for Memory. It may be even though we have the same memory yours needs more voltage than mine so for testing purposes your good to boost that up to around 2.1v taking care the sticks don't get too hot (heat can cause memory errors also).
I'm running a microATX board here based on the G31/Bearlake chipset which is the same generation as your P35 board and I have the exact same issue as your having when trying to run above 333MHz-FSB. I guess what you need to do it experiment with the Memory ratios also, we tried [3:2] and [1:1] but the board isn't liking it when used with a high FSB?
Previously when you started this thread you were running a [5:4] memory ratio which allowed you to boost your FSB to 368MHz so maybe we need to try that out again. If you set your board back to 333MHz-FSB the [5:4] memory ratio will show up as DDR2-835. In theory that would let you raise the FSB all the way to 426MHz (1705MHz System Bus) without your ram running beyond its rated spec.
The [5:4] memory ration is designed to run on a base 266MHz-FSB (1066Mhz System Bus) with DDR2-666 ram and is not really ideal for what were trying to do but it does seem to be working the best atm?
Your board is setting itself up based on your 266MHz-FSB processor and is using the 266-Strap which has associated tight northbridge timings. On newer boards its possible to manually over-ride this and *force* the chipset to run a higher and more relaxed northbridge strap but it seems we dont have that luxury. On some boards just manually setting the Memory ratio also forces the northbridge strap to change.
I reckon the problems we are having is either a voltage issue which we can try to work through or most likely a strap issue. If we popped out your CPU and replaced it with a native 333MHz-FSB (1333MHz System Bus) chip the board would default straight away to the 333-Strap and you may find certain problems we are having now would just vanish. I'm not suggesting you actually change chip but just used that as an example. Regardless the 266-Strap shouldn't be a showstopper as I've used that on newer P45 chipset boards all the way up to 500MHz with the right voltages.
So at this point I suggest loading CMOS defaults and starting again. With everything on [Auto] I would first change the FSB from 266MHz and go straight to 333MHz, save & exit, back into BIOS set vNB to 1.4v and the vCore to 1.4/1.5v (these are for testing purposes and can be reduced later). So you will be at 3.0GHz CPU (9x333), boosted vNB and vCore and that just leaves the Memory ratio which I suggest you experiment with, try all of them to see which allows you to gain the highest FSB (obviously not the ones that boost the sticks beyond rated DDR2-1066 spec). Also throw a lowered CPU multi into the mix starting with [x9] then [x8] and [x7] etc . . . . . with these manual voltages, different memory ratios and different CPU multis we may make some progress!
Get some scrap paper and make some notes of what you are doing, be patient, methodical as possible and observant and if you find your getting worn down by all this testing just reset your system back to stock with the memory running at full DDR2-1066 speeds and relax for a bit do some gaming etc.
I am hoping we can make a few breakthroughs today!
from the sounds of it I am thinking the problems we are encountering are chipset/northbridge related. I don't think its the processor and I don't think its the memory although there is an outside chance it could be a combo between the memory and the chipset.
If the chip was undervolted I would expect the computer to just reboot without a blue screen and if it was just the memory you may have troubles booting windows or just have randon things crashing on your desktop.
If you want to eliminate the CPU and Mem then you can boost the vCore so that it reads between 1.4v and 1.5v in CPU-z/uGuru-BIOS and the same for Memory. It may be even though we have the same memory yours needs more voltage than mine so for testing purposes your good to boost that up to around 2.1v taking care the sticks don't get too hot (heat can cause memory errors also).
I'm running a microATX board here based on the G31/Bearlake chipset which is the same generation as your P35 board and I have the exact same issue as your having when trying to run above 333MHz-FSB. I guess what you need to do it experiment with the Memory ratios also, we tried [3:2] and [1:1] but the board isn't liking it when used with a high FSB?
Previously when you started this thread you were running a [5:4] memory ratio which allowed you to boost your FSB to 368MHz so maybe we need to try that out again. If you set your board back to 333MHz-FSB the [5:4] memory ratio will show up as DDR2-835. In theory that would let you raise the FSB all the way to 426MHz (1705MHz System Bus) without your ram running beyond its rated spec.
The [5:4] memory ration is designed to run on a base 266MHz-FSB (1066Mhz System Bus) with DDR2-666 ram and is not really ideal for what were trying to do but it does seem to be working the best atm?
Your board is setting itself up based on your 266MHz-FSB processor and is using the 266-Strap which has associated tight northbridge timings. On newer boards its possible to manually over-ride this and *force* the chipset to run a higher and more relaxed northbridge strap but it seems we dont have that luxury. On some boards just manually setting the Memory ratio also forces the northbridge strap to change.
I reckon the problems we are having is either a voltage issue which we can try to work through or most likely a strap issue. If we popped out your CPU and replaced it with a native 333MHz-FSB (1333MHz System Bus) chip the board would default straight away to the 333-Strap and you may find certain problems we are having now would just vanish. I'm not suggesting you actually change chip but just used that as an example. Regardless the 266-Strap shouldn't be a showstopper as I've used that on newer P45 chipset boards all the way up to 500MHz with the right voltages.
So at this point I suggest loading CMOS defaults and starting again. With everything on [Auto] I would first change the FSB from 266MHz and go straight to 333MHz, save & exit, back into BIOS set vNB to 1.4v and the vCore to 1.4/1.5v (these are for testing purposes and can be reduced later). So you will be at 3.0GHz CPU (9x333), boosted vNB and vCore and that just leaves the Memory ratio which I suggest you experiment with, try all of them to see which allows you to gain the highest FSB (obviously not the ones that boost the sticks beyond rated DDR2-1066 spec). Also throw a lowered CPU multi into the mix starting with [x9] then [x8] and [x7] etc . . . . . with these manual voltages, different memory ratios and different CPU multis we may make some progress!
Get some scrap paper and make some notes of what you are doing, be patient, methodical as possible and observant and if you find your getting worn down by all this testing just reset your system back to stock with the memory running at full DDR2-1066 speeds and relax for a bit do some gaming etc.
I am hoping we can make a few breakthroughs today!
Last edited:
glad to hear what we're encountering is normal...
i really like the way you give background info before we try something (telling me why we're trying something) It's really helping me learn. do you do this for a living?
I'll get on to this tonight when i get back from work.
what sort of result am i looking for in this step? the highest FSB and fastest DDR2 speed combined? do i do that then start lowering the multiplier?
Just out of interest, are the blue screens and crashes damaging at all to the system in the long term?
i really like the way you give background info before we try something (telling me why we're trying something) It's really helping me learn. do you do this for a living?
I'll get on to this tonight when i get back from work.
what sort of result am i looking for in this step? the highest FSB and fastest DDR2 speed combined? do i do that then start lowering the multiplier?
Just out of interest, are the blue screens and crashes damaging at all to the system in the long term?
The Blue Screens won't damage your hardware but as mentioned before its the O/S that is fragile. If we were boosting loads of voltages and burning the hell out of everything then after a few weeks and months of running at full load you risk degrading your hardware . . .so in short we are good for the moment!
The more I think about it the more I reckon we are having Northbridge issues, this becomes especially obvious when you do that Cold Boot test as described earlier. I am also thinking that your board may actually be changing the FSB-Strap when we change the memory ratios which if thats the case we need to concentrate on the native 333MHz-FSB memory ratios in the hope that we fool the board into setting the 333-Strap which should allow us to to get some FSB action happening.
The [1:1] ratio is designed to run an FSB of 333MHz (13333MHz System Bus) with 333MHz Ram (DDR2-666). The [6:5] ratio is designed to run an FSB of 333MHz (13333MHz System Bus) with 400MHz Ram (DDR2-800). In theory the latter [6:5] memory ratio should let you run an FSB up to 444MHz-FSB (1776MHz System Bus) while keeping your Kingston HyperX in spec.
So basecamp is now 333MHz, at that speed in BIOS select the memory ratio that puts the Rams at 400MHz (DDR2-800) and try again. Give the CPU/Northbridge/RAMs extra voltage and see what gives. Also try your original settings that you started the thread with and try to get higher with the FSB, drop down the CPU multi to see if that helps but keep the NBCC maths in mind as when you mix a high FSB with a low CPU multi that could open another can of worms which we don't wanna deal with atm!
I'll be suprised if by hook or by crook we can't get your chipset running at 400MHz-FSB, lots of things to try so just need you home and fed and cracking on with the testing!
The more I think about it the more I reckon we are having Northbridge issues, this becomes especially obvious when you do that Cold Boot test as described earlier. I am also thinking that your board may actually be changing the FSB-Strap when we change the memory ratios which if thats the case we need to concentrate on the native 333MHz-FSB memory ratios in the hope that we fool the board into setting the 333-Strap which should allow us to to get some FSB action happening.
The [1:1] ratio is designed to run an FSB of 333MHz (13333MHz System Bus) with 333MHz Ram (DDR2-666). The [6:5] ratio is designed to run an FSB of 333MHz (13333MHz System Bus) with 400MHz Ram (DDR2-800). In theory the latter [6:5] memory ratio should let you run an FSB up to 444MHz-FSB (1776MHz System Bus) while keeping your Kingston HyperX in spec.
So basecamp is now 333MHz, at that speed in BIOS select the memory ratio that puts the Rams at 400MHz (DDR2-800) and try again. Give the CPU/Northbridge/RAMs extra voltage and see what gives. Also try your original settings that you started the thread with and try to get higher with the FSB, drop down the CPU multi to see if that helps but keep the NBCC maths in mind as when you mix a high FSB with a low CPU multi that could open another can of worms which we don't wanna deal with atm!
I'll be suprised if by hook or by crook we can't get your chipset running at 400MHz-FSB, lots of things to try so just need you home and fed and cracking on with the testing!
sadly i've not had a lot of time to do much testing tonight (only just got in from work)!!
although... out of curiosity i set back to my original settings at 3.31MHz, booted, fine, went back into the bios and upped the FSB by 10. Booted, prime95 all ok except core 2 which had an error (sum was 5.394866373950674e+016 , expected 5.391359305433142e+016 if that means anything to you)...
am now running at 3.4GHz DDR2-925 and it seems to be working ok apart from prime95. still no blue screen though.
shall i go back to 333MHz and work up from there as suggested or use this as a base and play around with these settings?
although... out of curiosity i set back to my original settings at 3.31MHz, booted, fine, went back into the bios and upped the FSB by 10. Booted, prime95 all ok except core 2 which had an error (sum was 5.394866373950674e+016 , expected 5.391359305433142e+016 if that means anything to you)...
am now running at 3.4GHz DDR2-925 and it seems to be working ok apart from prime95. still no blue screen though.
shall i go back to 333MHz and work up from there as suggested or use this as a base and play around with these settings?
Hey no worries!
Ok so you said you had a prime error? which Prime test was this? (Small. Large, Blend?). An error is an error btw, if everything is sweet Prime will run forever but if there an error it will stop on one or all cores etc.
So your roughly at 3.4GHz (378x9) CPU, 378MHz-FSB/1511MHz System Bus with the ram running 472MHz [4:5] DDR2-945
Drop the CPU multiplier down from [x9] to [x8] and try again, if it was a pure CPU error from it running an 3.4GHz then now it should pass ok at 3.0GHz(378x8). To mainly test the CPU select the Small In-Place FFTs test from Prime 95, if you wanna test CPU/Chipset/Memory then select Blend
The other option if it is a CPU error is to boost the vCore value a few notches which you may or may not prefere to try first.
If neither dropping the CPU multi or boosting vCore helps Prime run Stable then we must look towards boosting the Nortbridge voltage (vNB).
See how you get on with that. I am most curious to know how many of the Memory ratios you can get working, this [4.5] one we are using seems a bit weird so let me know of any others that work ok and let you boot/reboot without issue!
400MHz-FSB is starting to seem real close now, keep checking your vCore value in CPU (idle/load) and also keep checking your temps (idle/load) and let me know!
Ok so you said you had a prime error? which Prime test was this? (Small. Large, Blend?). An error is an error btw, if everything is sweet Prime will run forever but if there an error it will stop on one or all cores etc.
So your roughly at 3.4GHz (378x9) CPU, 378MHz-FSB/1511MHz System Bus with the ram running 472MHz [4:5] DDR2-945
Drop the CPU multiplier down from [x9] to [x8] and try again, if it was a pure CPU error from it running an 3.4GHz then now it should pass ok at 3.0GHz(378x8). To mainly test the CPU select the Small In-Place FFTs test from Prime 95, if you wanna test CPU/Chipset/Memory then select Blend
The other option if it is a CPU error is to boost the vCore value a few notches which you may or may not prefere to try first.
If neither dropping the CPU multi or boosting vCore helps Prime run Stable then we must look towards boosting the Nortbridge voltage (vNB).
See how you get on with that. I am most curious to know how many of the Memory ratios you can get working, this [4.5] one we are using seems a bit weird so let me know of any others that work ok and let you boot/reboot without issue!
400MHz-FSB is starting to seem real close now, keep checking your vCore value in CPU (idle/load) and also keep checking your temps (idle/load) and let me know!
dropped the muitiplier to x8 and it's now stressing fine @378MHz
Vcore idle - 1.45
Vcore load - 1.39
Temp Idle - 43C
Temp load - 66C
been using the blend test each time
does this confirm it was a CPU error? is it fixable?
should i try going up or down with the memory ratios?
Vcore idle - 1.45
Vcore load - 1.39
Temp Idle - 43C
Temp load - 66C
been using the blend test each time
does this confirm it was a CPU error? is it fixable?
should i try going up or down with the memory ratios?
3.4GHz (378x9) CPU, 378MHz-FSB/1511MHz System Bus with the ram running 472MHz [4:5] DDR2-945 (Failed Prime 95 Blend)
3.0GHz(378x8), 378MHz-FSB/1511MHz System Bus with the ram running 472MHz [4:5] DDR2-945 (Currently running Prime Blend)
If reducing the CPU multi from [x9] to [x8] stopped Prime Blend from finding an error then it's 98% probability it was the CPU struggling to cope. You would need to let Prime run for a few hours really to be sure!
After you are sure its stable at your current settings then lets go back to running the CPU multi at [x9], to stop the error this time bump up the processor voltage [vCore] one or two notches. Keep a beady eye on your temps as 65°C in Prime Blend is getting high. I think you should be ok and finger crossed you can get Prime running stable at 3.4GHz (9x378).
We can do some memory tests once we worked out how to get the processor stable!
3.0GHz(378x8), 378MHz-FSB/1511MHz System Bus with the ram running 472MHz [4:5] DDR2-945 (Currently running Prime Blend)
If reducing the CPU multi from [x9] to [x8] stopped Prime Blend from finding an error then it's 98% probability it was the CPU struggling to cope. You would need to let Prime run for a few hours really to be sure!
After you are sure its stable at your current settings then lets go back to running the CPU multi at [x9], to stop the error this time bump up the processor voltage [vCore] one or two notches. Keep a beady eye on your temps as 65°C in Prime Blend is getting high. I think you should be ok and finger crossed you can get Prime running stable at 3.4GHz (9x378).
We can do some memory tests once we worked out how to get the processor stable!
i've put the multi back to 9 and upped the Vcore a couple of notches to 1.4575 however CPU-z is reading 1.440 at idle and 1.390 at load
Prime blend displays same error for core 2 after 30 seconds (it's always core 2)
i don't understand why upping the vcore in the bios is not reflected in cpu-z
Prime blend displays same error for core 2 after 30 seconds (it's always core 2)
i don't understand why upping the vcore in the bios is not reflected in cpu-z
That vCore seems less than when you had the multi at [x8] and you have raised it a few more notches? What does the reading say in uGuru BIOS page?
If you notice we are using two sets of figures for vCore, the one you set in BIOS and the one you actually get, it's the latter we are interested in although its good for reference to also mention what you dialled in the BIOS, the figure you see in CPU-z should correspond with uGuru BIOS and we refere to this as vCore-Actual as oppsed to vCore-BIOSset.
I think that chip may be Thirsty so your gonna need about 1.45vCore-actual for testing purposes maybe more! Nows probably a good time to ask . . .what is your CPU cooling? it could start to get hot in here!
If you have a picture of the inside of your case that may be helpful, I wanna have a better mental image of what kit we are dealing with and how its set up!
If after boosting vCore your Prime Blend load temps are hitting 70°C we need to improve your processor cooling, for sure if we try for 3.6GHz
Anyway try 3.4GHz again with more vCore, don't worry as your CHIP is robust and can take this testing, just make sure the temps don't start running away!
Have you any notes from your original overclock, I wanna know what vCore you were running!
If you notice we are using two sets of figures for vCore, the one you set in BIOS and the one you actually get, it's the latter we are interested in although its good for reference to also mention what you dialled in the BIOS, the figure you see in CPU-z should correspond with uGuru BIOS and we refere to this as vCore-Actual as oppsed to vCore-BIOSset.
I think that chip may be Thirsty so your gonna need about 1.45vCore-actual for testing purposes maybe more! Nows probably a good time to ask . . .what is your CPU cooling? it could start to get hot in here!
If you have a picture of the inside of your case that may be helpful, I wanna have a better mental image of what kit we are dealing with and how its set up!
If after boosting vCore your Prime Blend load temps are hitting 70°C we need to improve your processor cooling, for sure if we try for 3.6GHz
Anyway try 3.4GHz again with more vCore, don't worry as your CHIP is robust and can take this testing, just make sure the temps don't start running away!
Have you any notes from your original overclock, I wanna know what vCore you were running!
for my original OC the vCore was running at 1.4250... with the NB at 1.37
i upped the core a few notches and it made no difference to the uguru utility... the voltage stayed around 1.41... didn't increase even after saving, booting and checking again. Is it being bottlenecked somewhere? how far can i push it?
At work again so unable to take a picture atm but i can give you a run down of my spec
Icute jet turbine case
Abit IP35 pro
Q6600
Tuniq tower (w/ arctic silver thermal paste
4GB Kingston Hyper-X 8500 Ram
OCZ StealthXtreme Silent PSU 600W
Hitachi 320GB HD
Zotac 260-216 Maxcore
Asus Xonar Essence STX
Philips DVD-RW drive
Samsung 22" Monitor
Vista 32bit
Will get a picture this evening
i upped the core a few notches and it made no difference to the uguru utility... the voltage stayed around 1.41... didn't increase even after saving, booting and checking again. Is it being bottlenecked somewhere? how far can i push it?
At work again so unable to take a picture atm but i can give you a run down of my spec
Icute jet turbine case
Abit IP35 pro
Q6600
Tuniq tower (w/ arctic silver thermal paste
4GB Kingston Hyper-X 8500 Ram
OCZ StealthXtreme Silent PSU 600W
Hitachi 320GB HD
Zotac 260-216 Maxcore
Asus Xonar Essence STX
Philips DVD-RW drive
Samsung 22" Monitor
Vista 32bit
Will get a picture this evening
Hi nicnac1,
when you say your original overclock vCore was that 1.4250 vCore-actual or 1.4250 vCore Bios-set? I think we may need to get your chip approx 1.5vCore-actual to really get the most speed from it, don't worry as long as we can keep the chip cool enough it will be fine, the Tuniq Tower you have is a premium aircooler and should be fine with keeping a 65nm overclocked Quad-Core running safely!
Can you tell me if you are using a 4-pin +12v ATX cable or 8-Pin +12v ATX cable connected to your board? (the cable nearest the CPU socket). I ask because it seems you are getting hit with some vDroop which can be helped a little if you have the 8-Pin +12v ATX cable connected. Other than that it's either your PSU or motherboards fault (or combination of both?). The vCore running smoothly and as expected is really important when trying to keep a big overclock stable, if the vCore drops below a certain amount when the chip is blazing away your gonna get a fail or crash, better too much than too little!
When you get a minute have a play around with the different Memory Multipliers to see which work and which don't not forgetting to reboot a few extra times and doing the Cold-Boot thing!
I think you are doing well, there is a lot of information to deal with here so I'm trying to drop things in on a need-to-know basis otherwise you will meltdown heh!
There are two things going on that maybe are hindering us, the first is the memory multipliers are not operating as I would expect them too, the second is we haven't managed to take full control over your vCore and thats important if were gonna get that chip to 3.6GHz. To some extent your gonna need to use the force luke and not pay too much attention to your vCore Bios-set value and instead use the vCore-actual (uGuru) value.
Waiting for your feedback and looking forward to having a crack at 400MHz-FSB (1600MHz System Bus).
when you say your original overclock vCore was that 1.4250 vCore-actual or 1.4250 vCore Bios-set? I think we may need to get your chip approx 1.5vCore-actual to really get the most speed from it, don't worry as long as we can keep the chip cool enough it will be fine, the Tuniq Tower you have is a premium aircooler and should be fine with keeping a 65nm overclocked Quad-Core running safely!
Can you tell me if you are using a 4-pin +12v ATX cable or 8-Pin +12v ATX cable connected to your board? (the cable nearest the CPU socket). I ask because it seems you are getting hit with some vDroop which can be helped a little if you have the 8-Pin +12v ATX cable connected. Other than that it's either your PSU or motherboards fault (or combination of both?). The vCore running smoothly and as expected is really important when trying to keep a big overclock stable, if the vCore drops below a certain amount when the chip is blazing away your gonna get a fail or crash, better too much than too little!
When you get a minute have a play around with the different Memory Multipliers to see which work and which don't not forgetting to reboot a few extra times and doing the Cold-Boot thing!
I think you are doing well, there is a lot of information to deal with here so I'm trying to drop things in on a need-to-know basis otherwise you will meltdown heh!
There are two things going on that maybe are hindering us, the first is the memory multipliers are not operating as I would expect them too, the second is we haven't managed to take full control over your vCore and thats important if were gonna get that chip to 3.6GHz. To some extent your gonna need to use the force luke and not pay too much attention to your vCore Bios-set value and instead use the vCore-actual (uGuru) value.
Waiting for your feedback and looking forward to having a crack at 400MHz-FSB (1600MHz System Bus).
Evening Big.Wayne
Using the 8-pin connector
I've been playing with the vcore voltages and i've realise i need to really step up the indicated voltage for it to impact on the actual voltage. i.e. I've set the voltage to 1.5350 which is displayed as 1.490 in the uguru utility and 1.510 in CPU-z.
With that i've managed to get it stable at 380fsb x9 1:1.25 (DDR2-950) although load temps are hitting 72C and vcore drops to 1.460V. I'm considering a case upgrade soon cos this one was fairly cheap (i was a student at the time). Will that improve temps?
Haven't played with the memory dividers yet
going to leave prime95 to run for another hour or so to make sure its fairly stable.
after that, should i push on with the FSB?
what's the very maximum voltage/temperature i can work to?
Using the 8-pin connector
I've been playing with the vcore voltages and i've realise i need to really step up the indicated voltage for it to impact on the actual voltage. i.e. I've set the voltage to 1.5350 which is displayed as 1.490 in the uguru utility and 1.510 in CPU-z.
With that i've managed to get it stable at 380fsb x9 1:1.25 (DDR2-950) although load temps are hitting 72C and vcore drops to 1.460V. I'm considering a case upgrade soon cos this one was fairly cheap (i was a student at the time
). Will that improve temps?Haven't played with the memory dividers yet
going to leave prime95 to run for another hour or so to make sure its fairly stable.
after that, should i push on with the FSB?
what's the very maximum voltage/temperature i can work to?
Hey there,
that's some serious vCore fluctuation you got going on there!
vCore
It's the lowest voltage we are interested in, and 1.460v is a decent amount to help you get some more MHz from your CPU, I know people that have pushed 1.5v-1.55v vCore-Actual through their Quads but they had good cooling.
For the moment I suggest we set a max limit of 75°C full load temps, if we breach that then your gonna need to improve your cooling. A good thermally advantaged chassis really helps out especially when your using an Air-Cooled Heatsink. If your on a budget then the Antec 300 is a great case for the money!
Here is a picture of one of my machines built into the Antec 300 case, enough space even for a meaty graphics card and with two 120mm fans blowing in and a 140mm fan in the roof sucking warm air out your onto a winner!
So yeah a good case will improve your temps a lot and if your gonna upgrade anything then thats an obvious area and it will last you for years!
So back to the overclocking! . . . If your happy that 3420GHz (380fsb x9) 1:1.25 (DDR2-950) is fairly stable then we got a few things to try. I keep banging on about these memory dividers but I guess that can wait for the moment.
I think maybe we should have a stab at 400MHz-FSB (1600MHz System Bus) but with the processor running at [x8] instead of [x9]. Pop into BIOS and drop the multi (save & exit) then back into BIOS and turn up the FSB (save & exit) and keep your fingers crossed!
3.2GHz (8x400) - 500MHz Memory (DDR2-1000)
that's some serious vCore fluctuation you got going on there!
vCore
- Bios-set= 1.5350v
- uGuru BIOS= 1.490v
- CPU-z (idle)= 1.510v
- CPU-z (load)= 1.460v
It's the lowest voltage we are interested in, and 1.460v is a decent amount to help you get some more MHz from your CPU, I know people that have pushed 1.5v-1.55v vCore-Actual through their Quads but they had good cooling.
For the moment I suggest we set a max limit of 75°C full load temps, if we breach that then your gonna need to improve your cooling. A good thermally advantaged chassis really helps out especially when your using an Air-Cooled Heatsink. If your on a budget then the Antec 300 is a great case for the money!
Here is a picture of one of my machines built into the Antec 300 case, enough space even for a meaty graphics card and with two 120mm fans blowing in and a 140mm fan in the roof sucking warm air out your onto a winner!
So yeah a good case will improve your temps a lot and if your gonna upgrade anything then thats an obvious area and it will last you for years!
So back to the overclocking! . . . If your happy that 3420GHz (380fsb x9) 1:1.25 (DDR2-950) is fairly stable then we got a few things to try. I keep banging on about these memory dividers but I guess that can wait for the moment.
I think maybe we should have a stab at 400MHz-FSB (1600MHz System Bus) but with the processor running at [x8] instead of [x9]. Pop into BIOS and drop the multi (save & exit) then back into BIOS and turn up the FSB (save & exit) and keep your fingers crossed!
3.2GHz (8x400) - 500MHz Memory (DDR2-1000)
Last edited:
okey dokes.... got it on 400MHz-FSB x8 with DDR2-1000
cpu currently at 74C and the Vcore under load is 1.470 (i upped it one more notch)
like the look of that antec i suppose i need a clear path for the air to to go through the intake, across the tuniq and straight out the exhaust... at the moment (because of the position of the fixed case fans) i don't have that. (would show you a picture if i could find my damn digital camera lead!!!)
so, been priming for 30 mins and so far so good
Edit: Just started getting warning sounds from the MB... CPU got up to 75C and PSU was at 80C! have stopped prime95 cos i'm guessing alarms aren't good (unless switched off)
cpu currently at 74C and the Vcore under load is 1.470 (i upped it one more notch
)like the look of that antec
i suppose i need a clear path for the air to to go through the intake, across the tuniq and straight out the exhaust... at the moment (because of the position of the fixed case fans) i don't have that. (would show you a picture if i could find my damn digital camera lead!!!)so, been priming for 30 mins and so far so good
Edit: Just started getting warning sounds from the MB... CPU got up to 75C and PSU was at 80C! have stopped prime95 cos i'm guessing alarms aren't good (unless switched off
)okey dokes.... got it on 400MHz-FSB x8 with DDR2-1000
cpu currently at 74C and the Vcore under load is 1.470 (i upped it one more notch
/interrupts mumbling the equation, "compile, plus edit, multiplied by amend equals sticky(!)"
Encouraging inputs/results - good luck guys.
