undervolting pinmod for old P4
Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee
undervolting pinmod for old P4
Hey i recieved a computer from a friend and it has an old P4 2.0 A in it with a default vid of 1.5v... im wondering how i would go about to undervolt it because the mobo has 0 options for vcore and fsb adjustments...
thnx
thnx
Locate the datasheet on the Intel website, and locate the pinout. Find pins VID0 to VID3(or 4, can't remember), and the chart listing possible voltages. Find the pin settings for current voltage, and see what lower voltages are available by disconnecting the pins which are currently high.
Be careful though. Pin removal is a one-way street, and if you go too low, the processor won't have enough jungle-juice to power up...
From 1.5v, if VID4 was disconnected, then default voltage would drop to 1.1v. Probably too low, but should give you an idea how to read the datasheet. Assuming I'm reading it right...
--Rich
Be careful though. Pin removal is a one-way street, and if you go too low, the processor won't have enough jungle-juice to power up...
From 1.5v, if VID4 was disconnected, then default voltage would drop to 1.1v. Probably too low, but should give you an idea how to read the datasheet. Assuming I'm reading it right...
--Rich
ill give it a go... if i can get the thing to run at like 1.3 v or lower, i could probably have a fully passive system easy, as i have an older passive gforce 4 mx 420 layin round.[F]bernZ wrote:You can probably just put some scotch tape around the pin itself and then use a little at the bottom of the pin.
nice more ideas
ok acording to the spec sheet my current vid pins (0-4 respectivly) are 0 1 1 1 0 (1.5v default vid)
Now im shooting for 1.3v as my 2.8c will run at stock clocks at that voltage, then i will go from there. the vid pinouts for 1.3v would be 0 1 1 0 1
so vid3 would need to be made to a 0 somehow, and then i would need to isolate vid4 to make the mobo think its a 1... now how would i go about making vid3 a 0 from a 1?
could i possibly bend pin 4 over to touch pin 3?
ok acording to the spec sheet my current vid pins (0-4 respectivly) are 0 1 1 1 0 (1.5v default vid)
Now im shooting for 1.3v as my 2.8c will run at stock clocks at that voltage, then i will go from there. the vid pinouts for 1.3v would be 0 1 1 0 1
so vid3 would need to be made to a 0 somehow, and then i would need to isolate vid4 to make the mobo think its a 1... now how would i go about making vid3 a 0 from a 1?
could i possibly bend pin 4 over to touch pin 3?
well i just decided to go for it... ya kno what the heck its old hardware, and if i fried the stuff i would have an excuse to throw it out. pluss i literally had 2 of the same computer (2 older 478mobos, and 2 p4 2.0a's)
What needed to happen was the mobo needed to see Vid pin4 as a 1, and vid pin 3 as a 0 (exact opposite of what is stock vid4 is 0, and vid3 is 1 stock)
SO what i ded was bend vid pin4 over to touch vid3, this killed 2 birds with one stone, it isolated pin4 making the mobo think it was a 1, and made pin 3 a 0.... and it actually booted right up at 1.3v... YES I ROCK!
now to see how much lower it will go with some easier non destructive pin mods.
What needed to happen was the mobo needed to see Vid pin4 as a 1, and vid pin 3 as a 0 (exact opposite of what is stock vid4 is 0, and vid3 is 1 stock)
SO what i ded was bend vid pin4 over to touch vid3, this killed 2 birds with one stone, it isolated pin4 making the mobo think it was a 1, and made pin 3 a 0.... and it actually booted right up at 1.3v... YES I ROCK!
now to see how much lower it will go with some easier non destructive pin mods.
Well done .
I have a mobile Celeron 1.8Ghz Northwood-256 here that'll run at 2.4Ghz with 1.225v. Getting a decent motherboard today, and already have a gig of PC3200 sitting here, so we'll see what she's really made of. Wonder if I'll hit 3.2Ghz with 1.1v...
One piece of advice - CPU pins are NOT made for bending. It's okay if the CPU is disposable, but they will snap with repeated bending and straightening.
Thinking about it, maybe the easiest non-destructive method for isolating pins would be to poke the pin into the end of a tube of superglue - it'll pick off the pin quite easily, should the 'tweak' need to be reversed...
--Rich
I have a mobile Celeron 1.8Ghz Northwood-256 here that'll run at 2.4Ghz with 1.225v. Getting a decent motherboard today, and already have a gig of PC3200 sitting here, so we'll see what she's really made of. Wonder if I'll hit 3.2Ghz with 1.1v...
One piece of advice - CPU pins are NOT made for bending. It's okay if the CPU is disposable, but they will snap with repeated bending and straightening.
Thinking about it, maybe the easiest non-destructive method for isolating pins would be to poke the pin into the end of a tube of superglue - it'll pick off the pin quite easily, should the 'tweak' need to be reversed...
--Rich
ya its disposable...
but using super glue to isolate the pin wont work.... the stuff is conductive, even when dry, at least thats what i have found out.
if i snap the pin off its right on the edge, i could always solder it back on... i am actually tempted to snap it off... and see if the thing will run at 1.1v
lol and good luck with 3.2ghz and 1.1v!
but using super glue to isolate the pin wont work.... the stuff is conductive, even when dry, at least thats what i have found out.
if i snap the pin off its right on the edge, i could always solder it back on... i am actually tempted to snap it off... and see if the thing will run at 1.1v
lol and good luck with 3.2ghz and 1.1v!
Well, I've been out of this game for years now. But I've had good success in the past:-nd4spdbh wrote: lol and good luck with 3.2ghz and 1.1v!
Celeron 300a stable at 600mhz(but with stupid voltage - happy at 560mhz(?) with something reasonable...
Celeron 566 stable at 1133mhz, almost stable at 1200mhz, and would benchmark at 1275mhz
Athlon XP 1700+(1.46Ghz) stable at 2400+(2Ghz), benchmarks at 2600+
But in recent years, my need for CPU horsepower has reduced. I spent several years with distributed.net doing RC5 - but abandoned it back in 2003 or 2004. All I aim for now is the most power I can get whilst being as silent as I can - and the Northwood-256 will be used in my media centre, so the same logic will apply.
Ideally, would like to go fanless. All fans in my desktop are running at 5v, including the PSU. And I run with the case off, so nothing gets too hot. But the media centre will be cased, so I can see a pair of 120mm fans being used in a custom MDF case, running at a low voltage. I have some LM317 adjustable voltage regulators left over from a PSU I was making, so I can add a potentiometer and get the voltage down so it's as near to silent as possible.
Anyway, I digress...
--Rich
well after prelim testing of the voltage i found that 1.3 was about as low that i could go without errors during stability testing.
So i poped off the stock intel hs and put on the better Zalman CNPS6500 flower hs
I had the case laying on its side so optimal thermal convection was no where near reached, but fully loaded with no fans the cpu got to just under 60*c which is just fine... i have no doubt if i were to set the case on its feet thus putting the fins vertical, that temps would drop...
I just need to figure where i can put the PSU, cus right now ima cut the bottom out of the case and raise it up about an inch and a half then cut the top so cool air can come in the bottom, and rise out the top... with minimal restriction, as i have a low profile passive geforce 4 mx 420.
So i poped off the stock intel hs and put on the better Zalman CNPS6500 flower hs
I had the case laying on its side so optimal thermal convection was no where near reached, but fully loaded with no fans the cpu got to just under 60*c which is just fine... i have no doubt if i were to set the case on its feet thus putting the fins vertical, that temps would drop...
I just need to figure where i can put the PSU, cus right now ima cut the bottom out of the case and raise it up about an inch and a half then cut the top so cool air can come in the bottom, and rise out the top... with minimal restriction, as i have a low profile passive geforce 4 mx 420.
My little Northwood-256 is happily running at 2.56Ghz and 1.18v. I thought this motherboard would let me mess around with voltages, but there are no options at all .
Quite surprised that I'm seeing 1.18v - with vid4 bent, it should be 1.225v. Hey-ho... if I can find a way to get 1.3v or 1.4v, then I'll have a proper stab at overclocking her, as I have PC3200 RAM fitted but am trundling along with an FSB of 570mhz or so.
Board is good for up to 248mhz, which makes for just short of 1ghz when quad-pumped.
I have a couple of spare banks of DIP switches - may take a look at hardwiring the board to let me play with voltages, if I can't find a BIOS flash with more options.
--Rich
Quite surprised that I'm seeing 1.18v - with vid4 bent, it should be 1.225v. Hey-ho... if I can find a way to get 1.3v or 1.4v, then I'll have a proper stab at overclocking her, as I have PC3200 RAM fitted but am trundling along with an FSB of 570mhz or so.
Board is good for up to 248mhz, which makes for just short of 1ghz when quad-pumped.
I have a couple of spare banks of DIP switches - may take a look at hardwiring the board to let me play with voltages, if I can't find a BIOS flash with more options.
--Rich