Probably related to the bridge chip. There was a year of Catalyst drivers that didn't work right with those cards. I ran into that with an X800XL. Maybe someone here can give driver ideas. I would search to see what drivers people say they used with the X1950 AGP.

Could it be a flaky motherboard throwing things off/conflicting with the bridge chip?? This is an OEM (eMachines) board after all. The only thing that's gotten anything other than a sudden hang was disabling agp spread spectrum in the bios. I managed to get a flickering screen then a black screen and a hang when I had Catalyst analyze the card for optimal oc settings. Before then all I got was just a hard freeze when I did that or started a bench mark. Benchmarks still give a sudden freeze while running, no flickering screen. Seriously considering going with a FireGL X3 and just setting this one to the side.

It certainly could be. I'm not very familiar with the quirks of the AGP X1950 though. I haven't used a bridged ATI card in ages.

I think ideally you would use Catalyst 7.11 or older. Catalyst 7.12 has a major OpenGL overhaul that breaks some older games like KOTOR. They dropped support for a bunch of old OpenGL extensions.

However I think it was that year that had the slew of drivers with hosed Direct3D support on the AGP bridge cards. I found a forum post that said 7.6 works ok.

X1950 is pretty tame when it comes to AGP compatibility or so I found. You can check my K8 system’s configuration, driver version and 3DMark01 run results here: My 3DMark01 Mega Thread

When someone says K7 and lockup my instinctual suspicion is a weak 5V rail PSU..

Also the nForce2 while being much faster than most VIA offerings is shitty in terms of compatibility with RAM and pcş/agp bus devices. Try changing nForce2 drivers or using a KT400A/600/880 motherboard.

swaaye wrote:

It certainly could be. I'm not very familiar with the quirks of the AGP X1950 though. I haven't used a bridged ATI card in ages.

I think ideally you would use Catalyst 7.11 or older. Catalyst 7.12 has a major OpenGL overhaul that breaks some older games like KOTOR. They dropped support for a bunch of old OpenGL extensions.

However I think it was that year that had the slew of drivers with hosed Direct3D support on the AGP bridge cards. I found a forum post that said 7.6 works ok.

Ok, so I'll try those 2 revisions. I think the version I have installed is 7.0. something then try a driver from 05 and 06 and see if I have any luck

appiah4 wrote:

X1950 is pretty tame when it comes to AGP compatibility or so I found. You can check my K8 system’s configuration, driver version and 3DMark01 run results here: My 3DMark01 Mega Thread

When someone says K7 and lockup my instinctual suspicion is a weak 5V rail PSU..

Also the nForce2 while being much faster than most VIA offerings is shitty in terms of compatibility with RAM and pcş/agp bus devices. Try changing nForce2 drivers or using a KT400A/600/880 motherboard.

Yeah, power is a possible issue, but I thought the nf2 boards got most of their juice from the 12v rail like intel boards? Unfortunately, the only psus I have handy are older OEM units that haven't been run for close to a decade, and these aren't good seasonic or enermax units. Ones a Bestec the other is branded Newton I think. And yeah, I've heard the nf2 can be picky. The Usb 2.0 hasn't work properly on it either since installing the drivers (and yes, I did install DX9 beforehad). Combine that with it being a OEM mobo from a company that I don't think is in business anymore and yeeeah... If driver changes don't help I'll probably just set it aside and go with a Firegl x3 or a X800 XT. Judging by Phil's video on the X3 it seems to be a good match to a fast K7. Hopefully I'll have better luck with it . This is the second card that hasn't turned out right on this build. As far as the X1950 is concerned, I have a Intel D845PT desktop board from my old Gateway (the K7 rig is actually in its case right now) that needs recapping, once its done I'll try the x1950 again. If the same issues are present I'll assume the card is effectively shot.

appiah4 wrote:

X1950 is pretty tame when it comes to AGP compatibility or so I found. You can check my K8 system’s configuration, driver version and 3DMark01 run results here: My 3DMark01 Mega Thread

When someone says K7 and lockup my instinctual suspicion is a weak 5V rail PSU..

Also the nForce2 while being much faster than most VIA offerings is shitty in terms of compatibility with RAM and pcş/agp bus devices. Try changing nForce2 drivers or using a KT400A/600/880 motherboard.

Oh, and FWIW, 10.2 (the most recent XP driver I think) was the first one I tried.

Yeah, power is a possible issue, but I thought the nf2 boards got most of their juice from the 12v rail like intel boards

Depends on the motherboard, not the chipset. If the board used the 4-pin 12v connector it should be using 12v for the CPU. Seems like your FIC does.

The X1950 Pro also takes most of its power from the 12v rail.

Core temp remains flat at around 44 degrees

I assume that is the GPU? Sounds weird that it doesn't go up when you start a 3D app. What are you using for measurements?

lost77 wrote:

Depends on the motherboard, not the chipset. If the board used the 4-pin 12v connector it should be using 12v for the CPU. Seems […]
Show full quote

Yeah, power is a possible issue, but I thought the nf2 boards got most of their juice from the 12v rail like intel boards

Depends on the motherboard, not the chipset. If the board used the 4-pin 12v connector it should be using 12v for the CPU. Seems like your FIC does.

The X1950 Pro also takes most of its power from the 12v rail.

Core temp remains flat at around 44 degrees

I assume that is the GPU? Sounds weird that it doesn't go up when you start a 3D app. What are you using for measurements?

I'm using GPU-Z for temp monitoring. It does go up a little bit I think, but it maxes out at only a few degrees higher. The highest reading I've noted was maybe 51-52 degrees C.

Sounds about right for a X1950 Pro.

Try setting FSB to 133Mhz in bios. If the computer takes longer or completely stops locking up at least you will have ruled out AGP incompatibilities.

Hmm, interesting, I'll try that and report back in a bit.

lost77 wrote:

Sounds about right for a X1950 Pro.

Try setting FSB to 133Mhz in bios. If the computer takes longer or completely stops locking up at least you will have ruled out AGP incompatibilities.

Ok, just tried it. It still locks up, but it acts differently. While it was loading test 3 in 3dmark 01SE I got a black screen for about 10+15 seconds , the fan on the card ramped for a second before dropping down to normal and it appeared to recover for about 10 seconds before locking up again as normal (frosen image). Normally it just freezes up during test 3.

Update. I ordered a FireGL x3. If it doesn't work either, I'm going to be strongly suspicious that this is actually a PSU issue. I have a hard time believing I could wind up with 3 bricked cards in a row! Also, I just finished recapping my i845 board, so i'll soon have another board to compare with this one as to AGP compatability.

Jackal1983 wrote:

Update. I ordered a FireGL x3. If it doesn't work either, I'm going to be strongly suspicious that this is actually a PSU issue. I have a hard time believing I could wind up with 3 bricked cards in a row! Also, I just finished recapping my i845 board, so i'll soon have another board to compare with this one as to AGP compatability.

Told you so.. 😀

Honestly, I took a look in some older threads and saw that there were some Socket A boards that had a EATX 12 volt line that actually didn't take as much juice from that port as as a Intel board would. It seemed like it was Via based boards that were under that suspicion primarily, but there's so little data on my FIC nF2 board that I suspect it's possible FIC cut corners and recycled an old board design and just slapped an EATX 12 volt input on it. My main evidence for this is HWINFO consistently reporting lower vcore than is specced for my CPU: 1.54-1.55 vs. 1.60. I suppose the vrms could just be trash, too. I also have an Athlon 3200+ on the way which will also be informative I suspect. If I wind up with about 1.60 vcore vs. the 1.65 specced, then its probably inefficiency, if it's more like 1.55 volts, then that'll be another mark in favor of a board not compliant to specs as far as power supplies go.

You probably shouldn't believe what the onboard measurements say though. It's unlikely to be accurate. Maybe there is some value in watching what happens to the voltage when the CPU goes from idle to load, but I'm not sure about that either.

swaaye wrote:

You probably shouldn't believe what the onboard measurements say though. It's unlikely to be accurate. Maybe there is some value in watching what happens to the voltage when the CPU goes from idle to load, but I'm not sure about that either.

Yeah, but it tracks pretty closely to what is reported in the bios. Edit: Anyhow, I'll basically be looking for a consistent error, not an absolutely accurate reading. If it's still .05 low when I install the 3200+, I'll assume the CPU is getting enough juice, however if it's still around 1.55 (ish) volts, then that suggests to me that the VRM is either shit and can't handle the 3200+ or the board doesn't pull as much juice from the +12v EATX plug as, say, the high quality ABIT nForce boards and its a PSU issue at the core. Oh, and my FireGL showed up and it looks to be actually shot: Graphical anomalies after driver install and lockups in 3d loads (3dmark outputs graphical garbage then locks up and restarts).

Well, an update: the seller of the FireGl got around to testing it and apparently its functional (It'would have been nice if he had tested it before hand it would have saved both of us alot of trouble). This suggests I have a broken/dying Mobo, pending testing the Firegl X1 I have on order in it. The seller actually tested it before selling it so I won't be returning it. I tried testing the X1950 on the intel system I just built but I'm getting issues with Direct3d on the system as well as problems with windows update (issues I didn't have with my AMD rig), so I might try reinstslling XP on it. A question: Should I install direct3d first then drivers or the reverse? Anyone have a source for .NET framework 2.0, Catalyst gets cranky without it and I can't find the 32 bit XP version on the MS website.

Try installing driver from device manager (manually choosing .inf file).

agent_x007 wrote:

Try installing driver from device manager (manually choosing .inf file).

The driver itself SEEMS to install fine, but I get an error when I try to run 3dmark saying Direct3d isn't installed. Dxdiag says everything is fine,though, so I thought there might be something going on with the install sequence: i.e., I installed D3D 1st when I should have done the driver first, or I tried to stack too many things on a single boot.