Leadtek 6200TD Vmods here VGPU/VDDR

[overcrash86]

all on pics:



good mods

[Cybercat]

Yes!!! Thank You!

[Cybercat]

OK, I've done the VMod, and managed to get 1.7v to the core. Problem is, at particularly high clockspeeds I'm getting black screens. The solution for this on the 6800 series was to solder on some extra capacitors. No one's (that I'm aware of) done this for the 6200/6600. Could I just get like a 3300u capacitor and solder it over another one already on the card without damaging anything?

EDIT

Alright, I took two 3300u capacitors off my fried Abit NF7 motherboard, and put them on my card. I did some research on how cap mods are usually done. The positive end goes on the measure point, the negative end goes to ground. So I found a measure point (one that was on one end of a capacitor already on the card) and put the two caps on there. There might have been some improvement, but not much.

I wonder, would upgrading my power supply help any? I've got a cheap Thermaltake 420W native 20-pin with only 18A on the 12v line, and my card is really using some serious juice. 1.7v VGPU, 3.4v VDDR. I'm wondering if I got a Enermax 485W with two 12v lines, totalling 36A, native 24-pin, with active PFC, if things would improve any. Any advice or opinions that anyone can give will help.

[seaman]

Sorry,what clock you have on the memory if you use 3.4V?
I guess,is Hynix 3.6

[Cybercat]

You guessed right.

[seaman]

You guessed right.

And what is the clock GPU/memory? I can do more than 600/800 with only 1.55/2.72V.I think you are not able to do 900 or 1000 to use 3.4V

[Cybercat]

And what is the clock GPU/memory? I can do more than 600/800 with only 1.55/2.72V.I think you are not able to do 900 or 1000 to use 3.4V

....not that you would know just by looking at my VDDR.

But as a matter of fact I can't do 900 or 1000. You can't really expect much from TSOP memory. I'm doing 580/740. This is game stable, not benching stable, which means no artifacts or instability. At the moment, I've lowered the VGPU to 1.51v because my card (or maybe my power supply) couldn't do 1.7v stably.

[overcrash86]

OK, I've done the VMod, and managed to get 1.7v to the core. Problem is, at particularly high clockspeeds I'm getting black screens. The solution for this on the 6800 series was to solder on some extra capacitors. No one's (that I'm aware of) done this for the 6200/6600. Could I just get like a 3300u capacitor and solder it over another one already on the card without damaging anything?

EDIT

Alright, I took two 3300u capacitors off my fried Abit NF7 motherboard, and put them on my card. I did some research on how cap mods are usually done. The positive end goes on the measure point, the negative end goes to ground. So I found a measure point (one that was on one end of a capacitor already on the card) and put the two caps on there. There might have been some improvement, but not much.

I wonder, would upgrading my power supply help any? I've got a cheap Thermaltake 420W native 20-pin with only 18A on the 12v line, and my card is really using some serious juice. 1.7v VGPU, 3.4v VDDR. I'm wondering if I got a Enermax 485W with two 12v lines, totalling 36A, native 24-pin, with active PFC, if things would improve any. Any advice or opinions that anyone can give will help.

3.4v ,it's too height for the memory module ,i set 2.7v for 700Mhz ...(no change without vmod ,is it useless)

for the vgpu i have same problems ,black screen over 1.52-1.55v :/
i'havent try with en 3300µF capacitor ,i'll search this week to solve this
for the moment benchies at 1.51-1.52v 634Mhz

[seaman]

I have a Sparkle6600 and I use a modded bios.That is why I can go so high with memory.You can find it in the 6600nonGT thread.It helps a lot overclocking the memory,due to its unique sets of timmings.
What I ment to say is that so high a voltage for memory will never help.I can not do better with 2.9V than I do with 2.75V.Limits seems to be somewhere around 615/815 artifact free,but at 600/800 I can play a year with no problems.
I was stuck for a while at 600/700 till I discovered this bios,and voltage didn't help me much.

[Cybercat]

Damn all you guys and your good cores. 634, damn, I can't even do 600.

3.4v does me good actually. I've got some RAMsinks, and anything lower than that artifacts, but anything higher then heat becomes the limitation. I've done quite a bit of testing with this.

Also it seems AGP 6600s OC better. Maybe it's just me, but the dedicated molex connector seems to do them some good.

[seaman]

Your card is PCI-E?
Guys with Leadtek AGP cards can not overclock more than 450-470 using Leadtek bios.But they overclock normal with other bios.
Did you try to use another bios?
Don't you think is dangerous to feed 3.4V on memory even with heatsinks?
This modules are rated 2.5V with a max of 2.75V.I can understand 3V but 3.4V looks way to much for me.
Try to change timmings in your bios with this set that I use:
TIMING0 : 060b1003
TIMING1 : 05010407
TIMING2 : 00240306
Don't know how you gonna do it because NiBiTor 2.1 doesn't seem to change them.Maybe an older version,or maybe edit the bios in hex.
If you succed,you are not forced to use so high a voltage.

[Cybercat]

Yes, PCIe, check sig.

I've only tried one other BIOS, and it messed me up. Performance went down the tubes, so I flashed it back. Haven't bothered with any other since. Call it laziness if you will, I guess.

Strange how Leadtek AGP 6600 BIOS locks overclocking at a certain point. Mine seems to be doing average, so I don't really think it has any OC lock on mine in particular. That doesn't mean a non-standard BIOS from another vendor wouldn't improve things, that's always a possibility. Who knows yet.

I'll let you know if 3.4v is dangerous when my card finally gives out. Until then, it's running quite stable. As I've mentioned before, I've done quite a bit of testing, and for me 3.4v seems to be right at the threshold of heat tolerance while still giving me benefit. It makes me wonder though if using another BIOS would lower my memory's tolerance though, because of some internal tweak or something. Don't know.

As for changing the memory timings to completely laxed values (whether that actually improves performance or not, I haven't seen, seeing as that usually decreases performance in most cases), I haven't messed with it, and probably won't worry about it until I move to another BIOS that already has higher timings set in it. As you've already mentioned, Nibitor doesn't change them correctly (this I have tried), so I won't bother fiddling with it until I get a BIOS that already has them set in. Changing them myself seems to be out of the question.

[seaman]

You can try to use nTimings to change them on the fly,to see if it helps.
If you take a look in the memory data sheet for this Hynix 3.6 modules,you will see that these timings are not so relaxed,as one will think.
And for me,helped rising clock from 700 to 800 and I got an extra 300 points in 3DMark2005.Plus the posibility to play with AA+AF at resolutions higher than 1024X768.
Is up to you to try.

[Cybercat]

nTimings huh? I'll have to give it a try. BTW what's the favorite BIOS that everyone uses? Well, then again for AGP users it may be different....


woa, lol, that's a complicated mess of numbers there. Uh, any advice? Which ones do I change?

[over@locker886]

My memory timings is same.
Samsung K4D261638F-TC36.
But,...

Nibitor read:
04090F0D/05010405/20230204

Rivatuner read:

050b1410/05010406/20230305

so, what is right??

[Cybercat]

I'd pay attention to nTimings. Not because they're necessarily right, but because they're the only useful ones, seeing as with that application you can actually change their values.

[seaman]

nTimings huh? I'll have to give it a try. BTW what's the favorite BIOS that everyone uses? Well, then again for AGP users it may be different....


woa, lol, that's a complicated mess of numbers there. Uh, any advice? Which ones do I change?

Input the values that I posted above

[Brandybuck]

Well , my card sparkle 6200 clocks are 630/700 and no vmod .. memorys are 4ns ...
thanks to guys for this thread , I'll try ..... I think gpu clock to reach 700mhz

[Cybercat]

Input the values that I posted above

those timings are even tighter than what it comes at.

[Cybercat]

Hey Seaman, you were probably right about above 2.75v not making a difference. I just tested at 2.8v and it was around the same. I don't know happened when I tested earlier that made 3.4v look like it made a difference, but oh well.

Er, then again, 2.9v makes a slight difference. That seems to be the sweet spot for my card.

[seaman]

Wisdom of the older

[Cybercat]

Wisdom of the older

uh huh. too bad that doesn't seem to apply to memory timings.

[seaman]

uh huh. too bad that doesn't seem to apply to memory timings.

I don't think I get you

[Cybercat]

I don't think I get you

The timings you gave me, as I said earlier, instead of being losened were even tighter than stock timings. You want them to be higher, not lower. How you were able to increase your clockspeed to 800 (was it?) by lowering your timings, I may never know. Maybe they were extremely low to begin with, or maybe you gave me the wrong numbers? Who knows.

[seaman]

The timings you gave me, as I said earlier, instead of being losened were even tighter than stock timings. You want them to be higher, not lower. How you were able to increase your clockspeed to 800 (was it?) by lowering your timings, I may never know. Maybe they were extremely low to begin with, or maybe you gave me the wrong numbers? Who knows.

Can you tell me what Riva Tuner reports about your timmings with your original bios?