View Full Version : MSI/Creative 5900SE/XT... brick wall @ 450/750...
ragejg
02-09-2004, 07:46 AM
http://www.nvnews.net/reviews/3dblaster_fx_5900/2.jpg
First I figured it was a cooling issue, so I slapped some ramsinks and a better hsf setup on there...
see pic below...
ragejg
02-09-2004, 07:47 AM
looks nice, but... no improvement...
ragejg
02-09-2004, 07:51 AM
Now I'm at teh point where I either:
Look for a better BIOS (might it be TIGHT memory timing issue? This is an engineering sample)
or
Do a vmod...
Now I haven't seen much on SE/XT vmods, and I also haven't heard much on flashing MSI/Creativ BIOS's...
Any words of advice?? I'm seeing others with the 2.8ns ram enjoying much higher frequencies than myself...
btw, another consideration is removal of the heatspreader for some more core mhz, but for right now, I'm more focused on the ram...
Thanks in advance!! :)
ragejg
02-23-2004, 02:56 PM
bump
Bra!nFreeZe^
02-25-2004, 11:57 PM
Ppl obviously dont even bother figuring out how to vmod these nVidia cards ? I only see them working on the Ati cards!
XanderF
03-02-2004, 01:29 PM
vMod on the 5900XT is real easy - I'll post pics tonight of where to get the voltage changes from.
But, don't know where to MEASURE them from. Need help with that.
Bra!nFreeZe^
03-05-2004, 05:30 AM
come on piccies!!!! ;)
Have you tried measuring underneath some of the caps ?
XanderF
03-05-2004, 10:37 AM
http://epsilon.sloweb.net/1nfi9s_5900XT-2-1.jpg
Pic of someone else's card, but it looks like mine.
What I *think* you need to do is the chips identified by the '1' and the '2'.
The pinout for these appears to be here (http://www.intersil.com/data/fn/fn9030.pdf), I think. The chip has a little dot on the top of it to identify pin 1. Looks like you need to connect pin 5 to pin 7.
Now, I did this using a ad-hoc 20k pots. That being, after all, what everyone was using.
Utterly useless here! Did this too late at night to add 2 and 2. Measuring the resistance between those two points showed only .73 kOhm or so. I should have used a 1 kOhm pot to overvolt these chips!
Have you tried measuring underneath some of the caps ?
What, while it's running?!? (Only way to measure voltage I know of!)
HELL NO!! I've lost a card that way! Measured two caps that obviously should not have been connected to each other once, very large spark show and dead card resulted.
Nono, I'll wait until someone with more funds than I has a guide to where to measure the voltage from before I proceed on my volt-mod again.
Bra!nFreeZe^
03-05-2004, 10:52 AM
Some nice info there !! :) have you seen an increase in the overclocking potential ?
XanderF
03-05-2004, 11:08 AM
Not at all, but, as I said, I used the wrong pots.
I had an ad-hoc 20k 'pots'. Basically a 10k pots with a 10k resistor in series. Hence, I could only take it down to 10k resistance, and the default was .73k. Obviously, it didn't get me anything.
In the next couple days, UPS should be arriving with some ramsinks for me, as well as another radiator for my loop (bringing me up to 3x80mm radiators). When that happens, I'll be taking the whole thing apart again and trying it with the 1k pots.
I REALLY wish I had a place to measure the voltage from before then, though!
ALSO! An interesting aside...
It turns out these cards not only have different CLOCKS for 2d and 3d modes, but different VOLTAGE, as well! IOW, don't adjust your vMod until you have the PC running a 3d app in 3d mode. Then, and only then, you'll be adjusting the 3d voltage.
(I guess it's a fixed percentage boost, so you COULD adjust it in 2d mode, just being aware that you are going to get a +.1v or +.2v or so kick when 3d mode comes on.)
XanderF
03-05-2004, 11:12 AM
Hmmm....actually, scratch all that.
I must have the wrong pins identified.
I definately get .73 kOhm or more on the resistence between those two pins.
X-Bit has an article up (http://www.xbitlabs.com/articles/video/display/evga-5950ultra_3.html) that reminds me of something I already was supposed to know. The resistence between the voltage 'reader' leg and ground SHOULD be - stock - 0!
It's by increasing this FROM zero that we overvolt.
Must have the wrong damn chip identified, then.
*sigh* starting over from the drawing board, then
Bra!nFreeZe^
03-05-2004, 11:21 AM
Let me know as soon as you figure something out!! :)