Easy Overclocking

[Squidge]

By the time you can get your meter out and measure the voltage of the core,you could have already fried the processor. By far the best way is measure both resistors before power is applied, and then calculate the output value using the algorithm I pasted earlier.

Drag, one of the original pentium heatsinks (small flat aluminium, with small fins, and no fan) would probably do nice. Stick it on with some Arctic Silver Thermal Epoxy (about £10 for a few gram tube), or some other kind of thermal adhesive.

Ok, before apply the mod i've looking for the points where the polymeter should read vcore and I think T7 is GND and T8 vcore... at least my flu gives 1.78v in that points. Anyone knows if this is true?

Thanks.

​

 

[DistWave]

That is right, I put two passes of pencil and the reading was the same, 1.78v. Added 2 more but with more pressure and in a second powered, the polymeter read 2.48v

Removed pencil and GP32 is still alive

 

[reaper79]

That is right, I put two passes of pencil and the reading was the same, 1.78v. Added 2 more but with more pressure and in a second powered, the polymeter read 2.48v

Removed pencil and GP32 is still alive 

​

Wow, that just go to show you that this method is WAY to inaccurate. Do it right and get some resistors and solder it or have someone solder it for you. At the very least, attempt to measure the resistance. However, do to the nature of coating a resister with powder, you will probably never get an accurate ohm reading, but at least checking with a multimeter would help.

As far as a heatsink, keep it small enough not to interfere with anything, and make sure you use thermal epoxy or tape... not paste. Anything to increase the surface area, even a flate piece of aluminum. Don't use a penny, the bottom needs to be as flat as possible, Lincoln's head would get in the way.

 

[Titcher]

I tihnk someone should wire a miniature fan in their GP32 for cooling, cut two little holes in to vent air, if the fan was fast enough could probably use the GP32 as a heater then.

 

[frolik]

So the official recommended voltage is between 1.65V and 1.95V, right. It sounds to me like most of you will be far exceeding that without knowing it, which makes me wonder if the CPU's used are more robust than Gamepark give them credit for.

Has anyone actually heard of a GP32 cpu being blown? If so what were the circumstances?

I'm curious about what resistances (and then voltage) people are getting after applying the lead, especially those who have been running for a while. Maybe we could create a test list to compare our findings, for example:

Model: FLU
MaxOC Before: 156Mhz
R48 Resistance: 68k
R52 Resistance: 120k
Voltage: 2.1V
MaxOC After: 192Mhz

and then maybe we will have some hard facts!

Squidge, I've been trying to work out your voltage formula and it produces different results to that of Rob's voltages, can you elaborate on it? Also I'm not sure whether inputvoltage should be 3V or 3.3V....

 

[reaper79]

Has anyone actually heard of a GP32 cpu being blown? If so what were the circumstances?

​

I'd like to know this as well. Especially if it was a result of overclock/volt modding. I was leafing through the white sheets for the processor and saw that the absolute maximum input voltage is 2.7 volts. I'm sure this is under the BEST of circumstances, but I wonder how far these units can be pushed. Especially with proper cooling of some sort. Probably not enough room to attach the proper components... too hard to fit a pump and waterblock in there

The recommended voltage is stated as 1.8 +/- .15, so yes 1.65-1.95. I think that this is the recommended voltage, because it is sufficient for the system to run stable at 133 mhz. The reason for this is that everything but the memory controller is imbedded in the chip, such as USB and LCD, and overclocking the chip overclocks just about everything. Perhaps the external memory controller explains why the overclock isn't dependant on ram, but I doubt it. But anywho, 133 is what they recommend for proper operation, especially of the USB and probably the LCD. Do some ppl have problems with the USB when overclocking??? Could be why.

 

[bast525]

Well I been playing the heck out of my BLU+ since yesterday with this mod... did some more testing and found that it is only stable up to 184mhz in Mame... anything over that and it locks up within seconds. I have played at 180mhz in Mame and DrMD for over 30 min stretches on a single game (no exiting to the menu) with no problems. Again, I would THINK this means the voltage has not been increased so high as to be a danger, but as was pointed out, due to variations in the quality of the processor I guess there's no way to really know.


For everyone trying it and not getting it to work I must stress to you what I said earlier... it is VERY easy to put too little or too much which will result in very similiar symptoms, basically it will seem not to work. So try various thicknesses of the lead over the resistor and if you can't get it to work, then also try putting some on the R48... mine only works if I put some on both R52 and R48, a thin layer on each, though a slightly thicker layer on the R52. Also be SURE to blow hard on the circuit board after every change as loose particles can make a difference.


I've never heard of a CPU being fried in 6 mos of being on this board, and as far as this mod goes it seems that even if you DO put too much lead on there as Trooper did, it is a simple matter of undoing it to get it working. Look I'm not technician, but just thinking about all this logically I think that the risk of frying the processor is not that great, look at how many have done it so far with all manner of different model GP's, and many of us HAVE applied too much lead to the resistor and found that all it does is cause the GP not to work properly... if it didn't fry it at that level it doesn't seem it would fry it at the lower level that actually DOES work. Again I'm no tech but I have a basic understanding of how electricty works and flows and it just doesn't seem likely this will fry anything. Plus, 3v (max voltage from the batteries) isn't DOUBLE 1.8v it's only 60% increase and it's still a pretty low voltage... Reaper says the CPU is rated for a max voltage of 2.7v. At any rate it definately lets you know when it's too much when you're trying this mod...

If anything I guess it IS possible that this may be shortening the life span of the chip by some degree but considering a chip effectively can last forever that could be just a relative thing.

Whatever... I'm keeping the mod and LOVING it and won't undo it until someone (hopefully not me!) fries their GP....

EDIT: anyone tried Quake GP at higher speeds yet? What kind of framerate does it get now?

 

[gaterooze]

Would this mod reduce the life of your gp32_console -- I don't mean a sudden burnout, just through continued use at a higher voltage over time?

 

[Vimacs]

didnt i jsut say that?

@gaterooze yes

 

[fitzsteve]

EDIT: anyone tried Quake GP at higher speeds yet? What kind of framerate does it get now?

​

I have got my GP to 180 and where 166 used to max at 14 fps i get a max of 17 in small spaces/non busy areas. However there isnt much change in busy parts still 7-8 maybe 10 fps

I would expect a better speed at 200-220 but maybe the lack of ram still slows it down, I remember playing quake on my p200 mmx with 16mb of ram, when I upgraded to 32mb the game ran sooooo much better.

Still would be interesting to know what you 220mhz people get fps wise in Quake

Steve.

 

[LOOM]

Sorry 252mhz :rolleyes:

 

[Kylix]

I've been hanging around these boards, checking them regularly, for about 2 years now, and yet I've only made a couple of posts..

As of late, I've pretty much lost interest in my GP32. I'm one of those people who can't play a game without sound, and seeing as my GP32 is really crap at OCing, this isn't too good for me.

Soooo, I'm gonna try this 'mod'. I really don't have much to lose.

I can't remember the last time I actually played an emulator, or anything else, for that matter, on my '32, so it seems that I could try this pencil thing and if it works, then great, It'll breath new life into my GP32. If it doesn't work, I can honestly say I won't be too bothered..

 

[Aaaaaaaa]

not even homebrew? there are some realy great ones!

 

[Kylix]

not even homebrew? there are some realy great ones!

​

Actually, thinking about it, I did play LadyKiller not too long ago...but before that, I have no idea..

Anyway, back on topic...

I managed, after about an hour and a feckload of pencil (completely scribbled all over it ), to get a solid 180 out of it on the clockspeed tester. All seems to be fine, and I'm dead chuffed.

Thanks a lot for this, it's brought my '32 out of retirement

 

[DaveC]

I can tell you haven't played with the GP32 in awhile. That is because there are now ALOT of emus that work full speed WITH sound. Try DrMD for example, full speed with full sound Megadrive emulation at only 144 MHz. Even Lynx is now near full speed w/sound at 156 MHz.

 

[Kylix]

I can tell you haven't played with the GP32 in awhile. That is because there are now ALOT of emus that work full speed WITH sound. Try DrMD for example, full speed with full sound Megadrive emulation at only 144 MHz. Even Lynx is now near full speed w/sound at 156 MHz.

​

Yeah, I have heard about these, but yet I hadn't had the inclination to try them myself.

I think I was just lazy...but this inspired me to get out my GP32, and start using it again.

I'd be grateful to this even if it didn't work, as it rekindled my interest in my GP32..

Or something...I don't know, my mind's all over the place at the minute

 

[schaap2005]

cool, can you let me hear what you find of the emu's you tried now, I don't know if I have to buy an Gp32 or an Gizmondo, so i'd like to hear,

 

[DistWave]

Well, i'm still researching and i've found some things:
- About 20 or 30 blind-attempts with 2v or more and the CPU still lives
- More than 1.96v does not boot my gp32
- Pencil increases conductivity during the first hours applied:
Put some and CPU gets 1.90v.
Two hours after and with no more pencil, CPU gets 1.92v
Four hous after, CPU gets 1.93v

I'll keep informing B)

P.S. I'm using a GP32 FLU with LCD controller broken, so I assume the risk

 

[bast525]

Thanks for the info DistWave.


Well I'm on day three of running this mod, still can get 184mhz but no more, still running fine. Distwave's findings kind of confirm what I said... if you put enough pencil to get a high enough voltage that would be considered dangerous (2v +) the GP will simply not work properly... it definately lets you know that you have put too much pencil.

I'd also like to point out that while regular alkaline AA's are 1.5 volts each, that those of us running rechargeable NiMH batteries have a little less to worry about... they are only rated at 1.2v each (at least, three different brands of NiMH AA's I have around the house are all rated 1.2v), so that means only 2.4v total provided by those batteries... might explain why it seems people using alkalines can OC better than people using NiMH's....

With that voltage fact in mind I'm even MORE convinced that the risk of frying your CPU is very low indeed.....


It's funny how fear of the unknown is going to keep the group of guys owning highly overclockable GP's as still being a pretty exclusive club... just like Robojoe pointed out, how people were afraid to OC at all and look now we ALL try to OC our stock GP's the minute we take it out of the box and not one system fried as of yet.

 

[reaper79]

I'd also like to point out that while regular alkaline AA's are 1.5 volts each, that those of us running rechargeable NiMH batteries have a little less to worry about... they are only rated at 1.2v each (at least, three different brands of NiMH AA's I have around the house are all rated 1.2v), so that means only 2.4v total provided by those batteries... might explain why it seems people using alkalines can OC better than people using NiMH's....

​

This is a misnomer, the power regulator is in charge of keeping the board at a constant 3.3v regardless of which batteries you use (within reason of course, don't wire a car battery to it). The CPU draws power from this regulator, but steps down the voltage to 1.8. The ratio of the two resistors in question is what determines the output of the power regulator that supplies the CPU.

As a side note, since you are effectively running the R52 resistor in parallel with a graphite trace, assuming the graphite trace could approach 0 ohms if thick enough, then the minimum resistance you could possibly achieve is 235k (470k in parallel with 0 ohms). Graphite on R52 will increase voltage... graphite on R48 will reduce voltage.

Cheers