Vcore is set to 1.7v in the BIOS.
The Vcore according to MBM5 ranges from a low of 1.66v to a high of 1.74v. Is this normal or is MBM5 not very accurate. Obviously when the CPU is under full load it tends to stay towards the lower end.
The reading in the BIOS Health Status switches between 1.71v & 1.72v.
I have the extra 4-pin ATX connected to motherboard as well.
AMD XP-M 2500+ @ 2.5GHz (1.7v)
Abit NF7-S v2.0 - ABit D27 Bios
OCZ Powerstream 470W PSU.
MBM5 gets this stuff from the Motherboard.
Mine only fluctuates by 0.02 between min and max, which suggests that maybe your PSU is not giving very stable power.
The NF7-S has never been good at reading voltages (or temp's for that matter), so i wouldnt worry about it too much. It seems to vary more on this board than it does on others, so its "real" varience is probably less than whats being shown.
Originally Posted by Saracen
The sensors in mobos are notoriously innacurate.
Puts my mind at rest.
The Vcore readings never go outside the high & low in the image as well.
Actually I was wrong before - the PSU power rails are showing very steady, as befits an OCZ. Mine is worse there (which is why I'm replacing it).
Still, I'm using the same motherboard and I don't see the same fluctuations in vCore.
That can also depend upon what sampling interval you are using - the shorter the interval the more it seems to fluctuate.
Also the higher the Vcore setting the wider the apparent swings.
That must be true - mine is set at 30 seconds though. And I have it on 24/7, during which time it has long bouts of CPU 100% use, and some periods of CPU next to 0% use.
And with regard to the NF7 showing inaccurate figures, while I'm sure that's true - the relative figures are probably accurate enough.
Yep, thats exactly right.And with regard to the NF7 showing inaccurate figures, while I'm sure that's true - the relative figures are probably accurate enough.
The faster you sample, the more samples you get. You dont miss the fluctuations, if you do more samples.That can also depend upon what sampling interval you are using - the shorter the interval the more it seems to fluctuate.
Also the higher the Vcore setting the wider the apparent swings.
The reson behind the fluctuations is this:
Mosfets. They get HOT. Hotter they are, less effiecent. More energy is wasted. They dont work aswell when there hot.
SocketA mobos are all designed with a vcore to be in between 1.1 and 1.85. AMD set these specs wide, never intending to use 1.85, I dont think.
Board designers wanted boards to be cheap to make, so the part of the mosets that should be heatsinked, is soldered to the mobo. Its just a way to make a cheap, but in-adequate heatsink.
If your going over 1.8, you should add heatsinks to the plastic side of the mosfets. A fan also helps a lot. Fluctuations do go down a LOT after this. There is no way you could run a nf7-s at 2v vcore without heatsinks on mosfets. It would just be instable.
Mine, as an example, when set at 2v, it would dip down to 1.9v, thats 0.1v fluctuation. Its too much!
Last edited by SilentDeath; 19-12-2004 at 12:43 PM.
So... my KT400 Gigabyte board running my CPU at 1.92V on air... without additional MOSFET cooling... is probably a bad idea? Meh, it's been rock solid stable for six months so I think I can deal with it.
Oh, and I like doing this:
MOSFET = Metal Oxide Semiconductor Field Effect Transistor (from memory that one. Kez = teh acronym machine)
PHP Code:$s = new signature();
$s->sarcasm()->intellect()->font('Courier New')->display();
Never used a KT400. I doubt it would have exactly the same design.
Mosfet cooling is often not bothered with - becuase of the cost of mini heatsinks
Anyway, how long does it take to hack up an old heatsink, and glue it on? Its what I did, took about 30mins.
Try drilling 60 5mm holes, 7mm deep into a block of copper with a hand drill
Semiconductors arnt 100% efficent, its more like 80%.
Meaning 20% of the power your cpu is eating, is wasted as heat in the mosfets, just to get it smooth enough. Thats why theres usually 6 of them. Cpus like 3 phase power, but one mosfet for each phase just isnt enough.
Future boards, iirc, might have 4 mosfets+ for each phase (= 12)
Mosfets arnt cheap, so there not going to stick another 3 or 6 on there just for 2% of there potential customers.. - cheaper to have 1% of boards RMA'ed.
Last edited by SilentDeath; 19-12-2004 at 10:18 PM.
There are currently 1 users browsing this thread. (0 members and 1 guests)
Posting Permissions
LinkBack URL
About LinkBacks
Reply With Quote
