Hi,
maybe the problem is the overheat of bridge chip.... maybe also this small chip is influenced by the mod, mine is really hot.
Gpu is cooled by a liquid system, the bridge chip by a small iron heatsink.
anyone can reply?
Printable View
Hi,
maybe the problem is the overheat of bridge chip.... maybe also this small chip is influenced by the mod, mine is really hot.
Gpu is cooled by a liquid system, the bridge chip by a small iron heatsink.
anyone can reply?
I got only 625 with 0,45ohm......... maybe is the bridge...
Hi. Sory 4 english :rolleyes:
I have Sparkle GF6600GT interfece AGP and i have problem. I must going higher clock on 600MHz on GPU but card sey not to me when i overclocking GPU on 605MHz. Help me to pencil vmod.
Quote:
Originally Posted by DEVILEK
If you need any help let me know.
I used a 5K rheostat connected here. The measured resistance before the mod was 418 Ohms.
http://www.aocb77.dsl.pipex.com/6600voltmod.jpg
My meter is pretty accurate. Adding just the 5K rheostat let me increase both the GPU and memory speeds for some reason.
These are the measured GPU voltages before mod at the capacitor. I'm not sure why my 2D voltage is so high. Possibly because I have 418 ohms instead of 500, but then the 3D voltage would be higher also but its not.
1.34 volts (2D)
1.45 volts (3D)
My MSI 6600GT AGP could do around 560/1.13 stable before the mod. The optimal test would detect 564/1.03 before the mod.
I lowered the resistance until I got;
1.54 volts (2D)
1.65 volts (3D)
After raising the voltages I tested the optimal overclock settings again and both memory and GPU went up to;
579/1.09
Then I tried to set them manual and I got around 579/1.19
What I noticed is even though the optimal test listed 579 on the GPU as being stable it wasn't stable I got artifacts. That's weird because the drivers list that speed as being OK and it really isn't. The memory seemed to be OK at the faster speed.
What I also noticed is even though I raised the memory from 1.13 to 1.19 with no artifacts. There wasn't really any improvement in benchmarks. So I don't think the GPU is very memory bandwidth limited at the speed I am running it at.
So I disconnected the mod for now and in the future when I get better cooling I might give it another try. My cooling now seems to be pretty good so I'm not sure if a better coolor would make a difference. It idles at around 46C and around 62C load. This is with the driver probe. I have an external probe also that reads 35C idle, 51C load. The external probe is touching the side of the GPU with thermal compound and is accurate to within 1C+-.
Heat is the culprit with the green trapazoidal shapes that appear after increasing the voltage. If I had my card clocked to high when it was stock, these would appear, esp. during the Troll test on 03 and to somewhat lesser degree on Nature. I did the pencil on both sides of the card, Gigabyte 6600gt, and they cleared up. So of course I would increase the speeds and they would reappear. At that point I put a strong 92mmx25 fan infront of the fan on the stock heat sink. It helped alot. Then I duct taped a fan from a TT-Volcano 6 cpu hs to the HS on the card. Anyone who had one of these knows they are abit loud but put out good air and pressure for their size. This coupled with the 92mm in front of that and some AS, it clocks much higher than stock and stays the same temps as far as I can tell. 46 - 47c idle. 600/1200. I moved the outer fan to reposition it during a 3DMark run and in a few seconds my screen looked like a green screen they use for special effects. Quickly replaceing the fan where it was removed the green patches. You have to have more cooling for the heat caused by the increased volts.
Quote:
Originally Posted by coop
So your temps stayed the same in the drivers with better cooling, but you didn't get the green artifacts anymore?
Yes, atm it is not pretty but it works. For some reason it won't go lower than 46-47. I switched hs with a zalman iceberg that was terrible, a forged Tiasol cpu cooler, and the best one was the stock one. That is all I had to work with. I did sand it slightly, put AS on the chip, and increase the tension on the springs slightly. I'm not sure that really did anything, but if you push down on a pot of water on the stove just as it starts to boil you can see the effect. I'm running at speeds that would not have been possible and with the fans can keep the artifacts from appearing. I will say that the room temp makes some difference and if I only use one fan the artifacts are in the first airplane test almost as soon as it starts. Cooling is key. Her is a pic of when I used one fan. Now I have the 60x60x25 duct taped to the front of the hs right over the stock fan. I tried with the stock fan in, out, on off, it works best on.
It looks like someone spilled a box of fans on your computer. :)
I think the idle temps are tough to get lower no matter what unless your using water cooling.
Have you tested the temperature at load? Because maybe with your extra fans your lowering the load temperature which would probably help with the green artifacts.
I know, and I agree about the fans and the lowering of the temps. I'm thinking of getting a water block. The chip seems to respond well to lower temps just from air cool.
Quote:
Originally Posted by coop
I'm gonna have to try changing the fan on my MSI 6600GT. It looks I can take off the cover on the heatsink and screw an 80MM fan to the heatsink.
You can take it apart, the two push pins and the hs itself has screws to hold top and bottom together. The stock hs has been rather well thought out and designed I think it just needs a bigger stronger fan. The spacing of the holes in the card were a problem for me as far as attaching a different hs. They are just not spaced right. I even tried a NB hs from an LP B. It did not work very well at all. Good luck, keep sharp pointy things away and stay grounded. I almost ran a box cutter across mine trying to cut through a wire tie I used for one of the sinks :eek:
Quote:
Originally Posted by coop
I just modded it and it really didn't help anything.
This is what I did I took off the cover plate and removed the fan. That left me an empty area where the fan was. I found a scrap heatsink that I cut down to fit in the empty area. I attached the scrap heatsink to the empty space using Arctic Silver 5 and Arctic Silver thermal epoxy. Then I screwed an 80mm Smart Fan 2 to the heatsinks.
The card runs cooler but I still get the green triangle specks at higher voltages. Unless I run the smart fan 2 at full. I don't want to do that so lowered the voltages back down to default. I think the GPU is just limited.
That was kind of a good idea attaching another hs where the original fan was. You will need to have sufficent air though to make a diff. Did you do anything to increase the voltage to the mem or gpu? Also, I don't know how much the pcie to agp chip will affect what you are doing or wanting to do I should say.
Kind of...lol? It was a good idea. :)Quote:
Originally Posted by coop
The more heatsink mass the better. It filled in the empty space pretty good. I put the Arctic Silver 5 right in the empty space right where the GPU would be, and then used Arctic Silver thermal epoxy on the rest. I have plenty of airflow because I attached an 80mm Smart Fan to the card when I was done.
I increased the voltage for the memory and GPU, and the fan helps but only if the fan is running at full blast or very high. There is no way I could take the fan running like that it makes too much noise.
I think the problem is the heatsink. It has to be either copper or well designed aluminum. Pure copper would perform so much better because it would absorb the heat faster. The MSI6600GT is aluminum or something like that....but its not copper for sure.
check your pm
Hi,
My Sparkle 6600GT was on reclamation and returned with another revision and layout. I`m lama in electrical things( and english too :-) ). Is there anyone, who know where I must apply pencil to increase voltage? I have searched all day long, and I find nothing, that looks like my card...
thanks for any answer
Is it always C37? Or isn't it :confused: . I'm a n00b in voltmods.
The Regulator chips where on the back on the previous GT's grab some shots of the back of the card.
I have found on my MEDION 6610XL (NV43 for PCI-Express.) the C37 too, but on the place, where should be a resistor, there are only two solder points. My PCB-Layout looks very strange, when it compare it with the Standard-PCI-Express PCB-Layout. It is a Special-Version from MSI of the 6600GT card for a big german company like DELL.Quote:
Originally Posted by DesertShooter
Could anybody help me and say defenitly, that C37 is the Point to put a Resistor on? I would start with 1 Kilo-Ohm ore more and slowly lower to the famous 550 Ohm in little Steps with OC- & Temperature-Checks beetween.
My Card goes powered by Infineon GDDR3 (128 MIB) up to 561 MHz GPU & 515 MHz VRAM (DDR1030) - from 400 MHz GPU & 400 MHz VRAM. I hope to hit 666 MHz GPU, the Card have a lot of potential. :D
P.S.: Please help me, all you VOLTMOD-Sepcialists.
Is there nobody online?Quote:
Originally Posted by TtecX
can someone post the pic of the 6600GT agp vdimm mod please?Quote:
Originally Posted by zakelwe
tnx,
No, but I found this: http://sg.vr-zone.com/?i=1912Quote:
Originally Posted by Jort
This applies to most 6600GT AGPs, including my XFX and it looks like the BFG as well. My resistor measured a little differently but I did the pencil trick and here are my results:
ohms on resistor (see the link above):
stock: 538 modded: 497
Stock: 1.458vcore
Modded: 1.566vcore
60ºC max under load, either way
nVidia 91.47
"Detect Optimal Frequencies" 2nd pass
Core: 585 Mem: 1.17Ghz (585mhz) < - stock
Core: 589 Mem: 1.18Ghz (590mhz) < - modded, yawn
GPU (stock volts, Mem@500mhz):
reached 616 before persistent artifacts
ATITool held at 595.90 artifact free (5+ mins)
GPU (1.566v, Mem@500mhz):
reached 633 before persistent artifacts
ATITool held at 617.13 artifact free (5+ mins)
Mem (stock volts, GPU@500mhz):
reached 615 before persistent artifacts
ATITool held at 604.48 artifact free (5+ mins)
I have yet to mod my vdimm but we'll see. I'd like to push the core even further just for kicks, but I might leave it alone to avoid a kick in the pants, instead :D