View Full Version : NF7 v2 chipset Voltage
Okay, what Voltage should I give my chipset with stock cooling? I have Vmodded with a pencil, and am giving 1.85volts right now. I will be lapping as soon as my actic silver ceramic comes in.
thanks
buff
I don't think i would give it any more than that with the piece of crap stock cooler.
IamAnoobieCheez
08-05-2003, 08:15 PM
Originally posted by buff
Okay, what Voltage should I give my chipset with stock cooling? I have Vmodded with a pencil, and am giving 1.85volts right now. I will be lapping as soon as my actic silver ceramic comes in.
thanks
buff
sounds like you're wanting to to overvolt your vdd at 201+ FSB on your rev2.0 board.
nforce2 a1 ultra400s fsb oc well without vdd mod. vdd mod comes handy when you push your fsb to a very high level. Many new overclockers tend to overvolt the vdd when actually the Vcore and cooling is needed on the CPU. vdd will not make things change everything. please consider all these before bumping up too vdd voltages. It is unnecessary.
if you're at around 200~ 210 FSB, lower your vdd to 1.6v.
hedge
08-06-2003, 08:42 AM
what kind of pelt did you use on your nb, MickeyMouse?
I just noticed you FSB is only 201.. i don't think your problem is Vdd then.
With a rev2.0 board 200FSB should be obtainable with 1.6V
The stock cooler isn't that good at all, take it off and see how bad it makes contact actually, it will scare you.
It wouldn't recommand exceeding 1.85V w/ that crappy cooler.
uhm, I haven't updated my sig yet. I am at 217 with 1.7v, 222 with 1.85v. I already did the L12 mod, and I know it worked 'cause it booted up with only "user define" cpu speed. my mem is at 2-2-2-7 @ 2.99v. I tried running single channel, and it made no difference on my max fsb. I'm kinda stummped on how to improve my fsb to 230.
It looks like your chipset wants some more.
If you do the mod with a pot it's more accurate than with a pencil.
If you want to give your chipset more voltage upgrade your NB cooler first.
Oh, have you tried loosen your timings?
IamAnoobieCheez
08-06-2003, 12:35 PM
three of my nf7 rev2.0s do not require more than 1.7v Vdd at 225mhz FSB DC 5-2-2-2.0, with no NB cooling mod, just a standard factory HSF with factory white goop, no lapping, no anything. vdd set to 1.7 only.
long time ago, I thought superior cooling the NB would do the trick in high FSBs but not necessarily. Turned out to be almost useless. Similar experience reported by Oppainter with his high FSB overclocking on the rev2.0s.
Supercooling the NB came handy with the A2 & A3 nforce2 chipset(rev1.0, 1.1, & 1.2) boards. They require a lot of vdd voltage and their chipsets heat up. Supercooling it(Switech MCW5000A + superchilled water on the NB), and no need to raise the vdd voltage.
I'm at 225fsb with the 14 bios. I tried relaxing my timings, nadda
saaya
08-07-2003, 07:54 PM
IamAnoobieCheez where do ya live? mind sending me one of those boards? :D i got a nf7-s 2.0 some days ago that wont do more than 215dc with 1.7vdd :(
i live in germany...
Looks like you got a point there, IamAnoobieCheez.
My rev1.2 board defititly needed a Vdd mod to get 200FSB stable, this current rev2.0 doesn't seem to need it.
The contact between the NB chip and the cooler is horrible, lapping the chip like i did won't hurt though. :cool:
Any idea on hot to get farther on the fsb? My mem is just fine. I am using the 14 bios dual channel, slots 1&3. cpu interface enabled. going to sinlge channel does nothing.
RacerX
08-08-2003, 09:56 AM
Originally posted by buff
Any idea on hot to get farther on the fsb? My mem is just fine. I am using the 14 bios dual channel, slots 1&3. cpu interface enabled. going to sinlge channel does nothing.
More juice! I need 1.930v to run 225 dc stable. That being said some NF7's need more than that. Luckily the Nvidia ultra chipset runs quite cooler than the first version. Anyone running 230+fsb either has one hell of a MB or pushing over 2v thru the chipset. Better cooling is involved as well.
Bios 14 is one of the better OCing bios. Some say disabling APIC when installing OS improves a more stable OC. I have no problems with APIC enabled and cpu interface enabled. Gonna reformat this weekend and try MickeyMouse's theory of better performance with APIC disabled.
IamAnoobieCheez
08-08-2003, 05:21 PM
i've tried apic enabled(before the format + instal OS) before, and got worse result. With the enabled APIC, the perfomance seemed to "gain" at lower FSBs but caused more instabilities at same high FSBs as the one without the APIC enabled, the perfomrance dropped due to it. Interesting finding though. And of course, I'm not saying this will apply to all other people, but i guess you give it a try and see how it goes? :) i'm now back to APIC disabled.
EDIT: fixed the typing error.
What is the benifit of apic enabled?
I ordered a vantec iceberg, how high of voltage should I give my chipset with that? I am also gonna lap the chipset.
Thanks
IamAnoobieCheez
08-09-2003, 07:20 PM
Originally posted by buff
What is the benifit of apic enabled?
I ordered a vantec iceberg, how high of voltage should I give my chipset with that? I am also gonna lap the chipset.
Thanks
i don't know the detail of APIC's functionality. Oppainter had said that it improves performance a bit, but more unstable at higher FSBs. Disabling the APIC would be the solution he's said. I have the similar experience as he had.
yes, 1.7v vdd is good enough. try that first.
I'm at 1.87v right now with stock cooling at 225dc. That wont die quickly, will it?
IamAnoobieCheez
08-09-2003, 07:26 PM
Originally posted by buff
I'm at 1.87v right now with stock cooling at 225dc. That wont die quickly, will it?
no, that voltage isn't bad enough to hurt your chipsets. Not a bad voltage.
st0nedpenguin
08-10-2003, 07:15 AM
Originally posted by IamAnoobieCheez
Many new overclockers tend to overvolt the vdd when actually the Vcore and cooling is needed on the CPU
Is it true that bumping up the CPU voltage a notch or two will usually get a higher stable FSB?
I have heard this a few times but never got round to trying it.
IamAnoobieCheez
08-10-2003, 01:56 PM
Originally posted by st0nedpenguin
Is it true that bumping up the CPU voltage a notch or two will usually get a higher stable FSB?
I have heard this a few times but never got round to trying it.
generally when you up the FSB you get higher clockspeed as we know, which needs that voltage + cooling to ensure stable operation. Abit board(and maybe some other brands) for example, the vcore often undervolts and fluctuates down further when you run at load, and especially at extensive period of time. This is pretty much common and applies to all computer systems, but the abit nf7 series boards tend to put out less voltage than what you command in the BIOS. Many people who are used to with other type of boards such as abit KT266/333 board for example, the vcore output is good, above what you put in. But try it.. it may or may not work for you. I can't say for sure because there are other variables. Also plenty of cooling is a plus. Keep the temperature within range.
st0nedpenguin
08-12-2003, 07:03 AM
Well, before I couldn't get anything above a 200Mhz FSB, no matter what I tried.
And now I have managed to get a 3DMark03 run to pootle along overnight at 210Mhz, after bumping the vCore up a notch and upping the chipset to 1.7v.
I can't guarantee it's stable, I'll know better when I get home, but considering it never managed longer than an hour before, I think there's a chance it's ok.
Now all I need to do is get the L12 mod done, and see how far I can go.
Cheers IamAnoobieCheez!
IamAnoobieCheez
08-12-2003, 10:48 AM
cool, i'm glad that 210mhz worked out for you. :)
st0nedpenguin
08-12-2003, 10:53 AM
Well, 210Mhz is nothing compared to most people on here, but it's as far as I have managed so far, If it's ok when I get home, I'll crank it up and see how far it'll go, I'm planning on doing the L12 mod tonight, so hopefully, the sky's the limit.
felix88
08-12-2003, 11:11 AM
Originally posted by IamAnoobieCheez
the abit nf7 series boards tend to put out less voltage than what you command in the BIOS.
this isn't necessarily true for all NF7-S boards. i know a few people, myself included, that have measured the actual Vcore from the mosfet and it is actually a bit higher than what it is set to. i think Abit just used an inaccurate hardware monitor.
anyway, i agree that you shouldn't need more than 1.7v to get 210-225MHz. it all depends on your board. i've hit 220MHz, but it locks up after a few hours.
I measured all the voltages with my nf7 and every one was higher than selected and shown in bios.
saaya
08-12-2003, 02:07 PM
how much vdd did you meassure when seting it to 1.7v?
Originally posted by saaya
how much vdd did you meassure when seting it to 1.7v?
Setting mine to 1.7 gave me 1.73V measured on the little solder square pad.
felix88
08-12-2003, 02:41 PM
yup, mine overvolts .02-.03v too.
2fast4u
08-12-2003, 09:13 PM
It's all in the cooling! Try putting a pelt on your nb and see what you get. I am running 230@1.93v stable for over 12hrs in Prime with an 80w pelt\waterblock.
IamAnoobieCheez
08-12-2003, 09:38 PM
Originally posted by felix88
this isn't necessarily true for all NF7-S boards. i know a few people, myself included, that have measured the actual Vcore from the mosfet and it is actually a bit higher than what it is set to. i think Abit just used an inaccurate hardware monitor.
anyway, i agree that you shouldn't need more than 1.7v to get 210-225MHz. it all depends on your board. i've hit 220MHz, but it locks up after a few hours.
ok.. i admit, i was a bit off in making the statement there.
but read this part :
Originally posted by IamAnoobieCheez
Abit board(and maybe some other brands) for example, the vcore often undervolts and fluctuates down further when you run at load, and especially at extensive period of time.
I've said, under load, not idle.
Did you measure the voltage at idle?
IamAnoobieCheez
08-12-2003, 09:42 PM
undervolting it due to voltage spikes down to 0.11V or so is a huge variation. Seems to fluctuate more than the other boards i've owned. Do not measure the voltage when you just start up your machine. That is not accurate way to do. Run it in the middle of the day while the system has been running at load. better yet, run prime95 and observe your Vcore voltage line.
felix88
08-12-2003, 09:47 PM
Originally posted by IamAnoobieCheez
Did you measure the voltage at idle?
i measured it when i was running F@H, and the computer had been on for several days. i'm pretty sure that the hardware monitor that Abit uses is just very inaccurate.
IamAnoobieCheez
08-12-2003, 09:51 PM
Originally posted by felix88
i measured it when i was running F@H, and the computer had been on for several days. i'm pretty sure that the hardware monitor that Abit uses is just very inaccurate.
and what was the reading on the Vcore voltage on yours while running F@H for several days? what was the lowest point and highest point you have observed in that rail?
felix88
08-12-2003, 09:56 PM
1.777-1.778(it bounced between the two), i had it set to 1.775. maybe i've just got a fluke board and the rest do undervolt, but my original point is that the hardware monitor isn't to be trusted.
IamAnoobieCheez
08-12-2003, 10:02 PM
Originally posted by felix88
1.777-1.778(it bounced between the two), i had it set to 1.775. maybe i've just got a fluke board and the rest do undervolt, but my original point is that the hardware monitor isn't to be trusted.
that is amazing.
only fluctuates by 0.001v. I find it completely b.s. i'm sorry.. but i don't buy it. maybe you didn't spend enough time to measure the voltage before you noticed the fluctuations at load.
I have never seen any board that fluctutes that small and overvolts while under load. Try Prime95 also..
another thing, you are at low vcore. How about in the range of 2.0~ 2.1V? Try overclocking the system, at higher Vcore, run Prime95 and try measuring the Vcore voltage. Running the system at default(or small overclock) at 1.7v @ F@H puts very small stress on the board.
felix88
08-12-2003, 10:31 PM
well, my Fluke 112 DMM is very accurate(at least compared to the hardware monitor), so it's not a problem with my meter. i tested the voltage at the mosfets 3 or 4 times, watching the display for at least 10 seconds each time.
if i can get the probes behind this 92mm fan i've got on my SLK900 i'll test it at a higher voltage some time. i'll also measure while running P95, although i can run 220FSB Prime stable, but i have to knock it down to 200FSB for F@H.
Originally posted by felix88
my original point is that the hardware monitor isn't to be trusted.
;)
IamAnoobieCheez
08-12-2003, 10:43 PM
Originally posted by felix88
well, my Fluke 112 DMM is very accurate(at least compared to the hardware monitor), so it's not a problem with my meter. i tested the voltage at the mosfets 3 or 4 times, watching the display for at least 10 seconds each time.
if i can get the probes behind this 92mm fan i've got on my SLK900 i'll test it at a higher voltage some time. i'll also measure while running P95, although i can run 220FSB Prime stable, but i have to knock it down to 200FSB for F@H.
;)
Folding and Seti'ing is easy. Try Prime95.. That's the real stress. ;)
I'm not saying your Fluke is inaccurate, but i don't think you're setup is the right way to measure the voltage. I'm not saying your method in terms of where to check the voltage is incorrect, but you're running the system at such low vcore, with low mhz, running F@H is what I find too easy. No matter what kind of power supply you have, your Vcore voltage fluctuation by 0.001v difference is completely obsurd.
Overclock the system to at least 2.4~ 2.5ghz since you're on air. Crank up the voltage to around 2v, and run Prime95 for few hours. Do not check the voltage at idle nor the first part of the prime test.
my original point is that the hardware monitor isn't to be trusted.
This issue becomes moot. Do you want to explain to me why I need at least 0.1v more to run the Abit NF7 series boards stable than my abit/epox KT266/333 & Asus A7N8X boards need with a same cpu and cooling? ;)
felix88
08-13-2003, 11:28 AM
Originally posted by IamAnoobieCheez
Folding and Seti'ing is easy. Try Prime95.. That's the real stress. ;)
i can run Prime95 for days without getting an error, but i have to raise my vcore and lower my FSB to get my system stable enough for F@H. anyway, i trust F@H as a stress test just as much or more than Prime.
i'm running my XP2500 at 2.3GHz(200 x 11.5) using 1.85v. the winbond hardware doctor reports 1.82-1.84v under load(F@H or Prime). CPU-Z reports 1.80-1.84v under load(again, F@H or Prime). and my Fluke DMM reports 1.911-1.912v under F@H, and 1.913-1.916v under Prime. i have also noticed that the hardware monitor reports 11.74v on my 12v rail, but if i measure the 12v rail from the 20 pin connector on the motherboard i actually get a bit over 12v.
Programs are always completely off on readings... i would never even begin to trust one. If you say your nf7 undervolts at load than i think it's your PSU's fault, if your reading it from a program than it's completely invalid.
Originally posted by saaya
how much vdd did you meassure when seting it to 1.7v?
1.77