Cheap Processor Upgrade? - Page 8

could someone provide a picture? i tried taking pictures the last time i did it but they all came out terribly. my canon a75 sucks. or maybe i do at taking pictures


edit: but in words, youre making bridges with the wire and dropping them into the socket on the motherbaord itself.


lga sockets must be a pain in the ass. would have to tie knots id imagine

Yes, take a look at the pin-in and pin-out diagram from the Intel specsheet for the 533FSB Pentium-M.

ftp://download.intel.com/design/mobi...s/30526201.pdf

We already know that bridging the C15 and C16 with a fine copper wire tricks the motherboard into running the processor at 533FSB. This did increase the FSB of my pentium-M 1.7 to 533, but left it unstable at the core speed of 2.26ghz. So, I bridged G3 and H3 as well.

Now, there are reasons why I chose this particular pin. By bridging G3 and H3, you force the Vid3 pin to always report to the motherboard that it (the Vid3 pin) is low (or 0). The Pentium-M works by letting the software (Windows XP, Bios, etc) tell it when to shift states to change it's multiplier. When the pentium-M receives a request to change multipliers, it sends a request to the motherboard to increase (or decrease) the voltage based on some predetermined internal settings. It uses the 5 vid pins (F2, F3, G3, G4, and H4) to communicate to the motherboard what voltage it would like. If you look around in the Intel specsheet for the voltage table (I think its page 16), and study the voltages that the PM 1.7 shifts through (using CPU-Z for example), then you will eventually figure out that you can affect minor changes in voltage requests by changing these pins.

Now, the pin bridging I chose is not the only one that will work. When my PM1.7 is at the 6x multiplier, the voltage is 0.988, so my Vid pins are 101101 (moving from vid5 to vid0). When it is at 17x multiplier, the voltage is 1.324, which is 011000. If I permanently ground vid3, then my vids change to 100101 (which is 1.116 voltage), and 010000 (which is 1.452 voltage, a 0.128 increase).

Now, its possible to actually lock the processor down to only one voltage, but that would make it run hot all the time (like a prescott). That's why I chose vid3. It still varies the voltage when the multiplier changes, so it is still somewhat efficient.

have you tried using rmclock at this point to undervolt the cpu to only what it needs? 0.7v should be enough at 6x133=800. not sure about the high side though. would have to prime to find out

Good idea, Ziddey. Using the voltage chart, I can run the 6x multiplier at .828 (anything lower will just shift 8 spots up because vid3 is permanently grounded). The maximal voltage is more tricky because vid3 is grounded. RMclock doesn't know that vid3 is permanently grounded, so it reports the wrong voltage. However, using the chart, you can figure it out. I found that the exact voltage this CPU needs at 17x is 1.404 (in RMClock it is 1.276). At 1.388, Prime95 finds an error. So I was able to shave .048 volts off of the maximum and .16 volts off the minimum.

Thanks Ziddey!

Hey stomper congrats, now I wish I woulda sprung the extra 50 for the 1.7, it would be awesome if you could post step by step photos on how you did this, I think I have an idea but a guide would rule.

Definately gonna have to get a 1.7 400fsb P-M when I get some dough, so that I can try this. Is it still running fine for ya stomper?

i bridged G3 and H3, and its still failing prime95. Also cpuz isnt showing any change in voltage

What do you think I am doing wrong Stomper?

If buying a new notebook, consider taking a lesser P-M (often saving a few hundred) and then popping in the 2.26 when it comes in June

Ok right that wouldn't be nearly as much fun!!

Toy-yoda, CPU-Z and RMClock report the voltage based on what the CPU says it SHOULD be at. If you ground VID3, the CPU isn't aware of it, but the motherboard is.

After bridging the G3 and H3 pins, you are still getting errors in prime95? Did your temps rise at all? My temps jumped considerably.

If you are sure that everything worked, you could try grounding vid4. That will result in a much higher voltage, but you should be able to use RMClock to bump it down.

Grounding vid4 will result in 001000 which is 1.580 volts at 17x. At that voltage, my CPU reach 70C in about 45 seconds, so I would not recommend keeping it there. However, if you ground Vid4, and use RMClock to set it for 1.212, you will actually be at 1.468 volts. You can then increase the voltage (in RMClock) step by step until Prime95 stabilizes.

Last night, I played WOW for almost 6 hours nonstop. No problems at all. This morning, I ran Prime95 at full tilt for 4 hours without any problems. So far, so good.

Hmm, I already gave up after messing with it for a couple hours.

I tried vid3 and vid4 but each time I got the exact same error in prime95 at the same time. Either I did it wrong the five times I pinmodded it or maybe something else happened.

The memory speed doesnt increase, does it? I also took some pics, I will post them later

Here are the pics, including the error I got

Can you run a superpi benchmark so we can see how much the speed increase is?

Yah, SuperPI 1M in 36 seconds.

Toy-Yoda, sorry that yours is not working. You have the FSB wire in the right spot. The memory is not affected by either pin bridges.

Hey, Toy-Yoda and I both have the XPS2 northbridge heatsink. Which means, no heatsink mods necessary to put the 6800 Ultra in our laptops.

Thanks stomper. I hope with time I may have this solved but I wont be working on it for a few days.

Congrats on your OC though, I posted this on other forums and people are calling insane/stupid for overclocking a laptop

Congrats Stomper! I have a couple questions. Will this mod work on other brand notebooks with sonoma or just the new dells? After you have a successful physical hardware OC is there a way to turn it down to what you regularly need with software? How did your voltage increase effect your batt life? Any heat issues? I think I understand the whole proccess, but you should make a step-by-step tutorial for this mod with pics. Thanks

PM760 Core Voltage - 1.308V

SiSoftware Sandra says my PM760 Core Voltage is 1.308V.

Take note in the processor's name it says;
P3D (Dothan) Pentium M 90nm 900-2.2GHz 0.9-1.4V