fillosaurus wrote on 2024-04-11, 20:42:

Wrangled with some ISA soundcards today. To be more specific, Yamaha SW20-PC, Reveal SoundFX SC400 Pro 16, Samsung Audio Magic X and an OEM HP Aztech 2320.
And somehow I revived 2 graphic cards, a no name S3 Trio 3D/2X AGP and a Matrox G200 PCI.

Nice, what did you do to get the video cards working again? I've had non-working cards work from just cleaning the PCI contacts or re-seating the BIOS.

I've been doing the same thing today on my slot 1 test bench setup - testing out the Yamaha DB50XG that I got inside of a computer I bought recently. I thought it had some fault because the main chip gets relatively hot going up to 60c?

It sounded very wrong playing passport.mid and canyon.mid but sounded fine in games. It seems to just be some mapping issue with those midi files since XG midis from FF7 play fine and games sound excellent with it.
Just to be safe, the Yamaha chip by the ROMs now has a flat ceramic 40mm heatsink on it since I've got a bunch of them for free and they do work.
*sigh* the card it fits best is the CT2950 soundblaster 16 but I heard the hanging note bug within just a few minutes of gameplay - time to modify the soundblaster's DSP with one from the Soundblaster DSP thread.

Then thought I'd finally test out my Tualatin mod on my MS6905 slotket - last time I used this slotket I thought I had completely broken it since neither tualatin or P3 worked so I read up and re-did it. Sadly I did that ~4 months ago and have no idea which mods I did. I plugged it into the board to test it and it works! 1.2GHz Tualatin celeron working on my Aopen AX6BC board.

Oh it was toasty though for some reason after about 2 minutes of operation and it's got a junk bin HSF on it right now just for basic testing.
Why is it so toasty??? The CPU voltage on the slotket is set to auto, that should mean it will go at 1.5v like the CPU says right???

Uh... checked with multimeter and it's running at 2.08 volts...
This 1.2GHz Celeron may not work for very long now - the slotket has manual voltage settings and it's now running at about 1.53 volts being set for 1.5v operation. Oops!

luckybob wrote on 2024-04-11, 21:48:

And this is where I keep assorted lengths of wire... https://i.imgur.com/RaKE8rjm.jpg […]
Show full quote

And this is where I keep assorted lengths of wire...

So, that's what it would have been like if I had collected all those assorted lengths of wire.

A man can dream though.

A man can dream.

Ozzuneoj wrote on 2024-04-12, 02:01:

So, that's what it would have been like if I had collected all those assorted lengths of wire. […]
Show full quote

luckybob wrote on 2024-04-11, 21:48:

And this is where I keep assorted lengths of wire... https://i.imgur.com/RaKE8rjm.jpg […]
Show full quote

And this is where I keep assorted lengths of wire...

So, that's what it would have been like if I had collected all those assorted lengths of wire.

A man can dream though.

A man can dream.

Try the 9700 on some different motherboards, the 9700 has hardware level incompatibility with some chipsets. I know this because I bought one new and it didnt work. ATI actually told me the 9800 was a revision that should work with my chipset and I should wait for that. I did and it did. So test it on some different boards. I wish I could remember what I was running at the time but I can't sorry.

Thermalwrong wrote on 2024-04-12, 00:52:

Nice, what did you do to get the video cards working again? I've had non-working cards work from just cleaning the PCI contacts […]
Show full quote

fillosaurus wrote on 2024-04-11, 20:42:

Wrangled with some ISA soundcards today. To be more specific, Yamaha SW20-PC, Reveal SoundFX SC400 Pro 16, Samsung Audio Magic X and an OEM HP Aztech 2320.
And somehow I revived 2 graphic cards, a no name S3 Trio 3D/2X AGP and a Matrox G200 PCI.

Nice, what did you do to get the video cards working again? I've had non-working cards work from just cleaning the PCI contacts or re-seating the BIOS.

I've been doing the same thing today on my slot 1 test bench setup - testing out the Yamaha DB50XG that I got inside of a computer I bought recently. I thought it had some fault because the main chip gets relatively hot going up to 60c?
1712874066305_100.JPG
It sounded very wrong playing passport.mid and canyon.mid but sounded fine in games. It seems to just be some mapping issue with those midi files since XG midis from FF7 play fine and games sound excellent with it.
Just to be safe, the Yamaha chip by the ROMs now has a flat ceramic 40mm heatsink on it since I've got a bunch of them for free and they do work.
*sigh* the card it fits best is the CT2950 soundblaster 16 but I heard the hanging note bug within just a few minutes of gameplay - time to modify the soundblaster's DSP with one from the Soundblaster DSP thread.

Then thought I'd finally test out my Tualatin mod on my MS6905 slotket - last time I used this slotket I thought I had completely broken it since neither tualatin or P3 worked so I read up and re-did it. Sadly I did that ~4 months ago and have no idea which mods I did. I plugged it into the board to test it and it works! 1.2GHz Tualatin celeron working on my Aopen AX6BC board.
IMG_3155 (Custom).JPG
Oh it was toasty though for some reason after about 2 minutes of operation and it's got a junk bin HSF on it right now just for basic testing.
Why is it so toasty??? The CPU voltage on the slotket is set to auto, that should mean it will go at 1.5v like the CPU says right???
IMG_3157 (Custom).JPG
Uh... checked with multimeter and it's running at 2.08 volts...
This 1.2GHz Celeron may not work for very long now - the slotket has manual voltage settings and it's now running at about 1.53 volts being set for 1.5v operation. Oops!

Very cool that the slotket adapter works! And glad you found and fixed the voltage before it was too late too 😁

What graphics card are you running alongside it?

Today I installed copper heatsinks (9mm x 9mm) on the VRM mosfets on this P2B-B 440bx motherboard to help keep them cool. One of them gets quite hot at 2.0V, too hot to touch, so hope these will help. A case fan will blow over them.

I'm slowly overclocking the FSB, starting at 100MHz and am at 120MHz now. Trying to reach 133MHz while taking some CPU-Z and 3DMark99 measurements along the way.

Uhg... digging through a parts box that I got last year and it was all in really rough shape so I put it on the back burner.

Lots of dead late-AGP cards.

One card that seems to at least power on is a Sapphire X1650XT AGP. Problem is, I can't find and stinking drivers for it. I have four different versions (9.12, 10.2, 14.4 and one other version) and none of them have support for the X series??? They have support for everything all the way back to the Radeon 9000, up to the HD4000 and HD5000 series, but no X800, X1600 etc.... what the heck is wrong with ATI drivers? Why are two generations of cards randomly missing from all of these driver packages? Why did they make this so hard? I could, presumably, drop in a Radeon 9000 Pro, 9700 Pro, HD2600XT or an HD5670 and they'd all work. Even using the manual install method to load an .inf, there is just nothing listed for the entire X series in any of these packages. 😠

How did this user use all of these versions for these cards?

I just got the 10.2 package from AMD's website and it does list X1650 Series in the XP inf, CX_95952. It's the same driver I used on my Sapphire X1650 Pro until its death alongside a HIS X1950 Pro, both AGP.

Thermalwrong wrote on 2024-04-12, 00:52:

I've been doing the same thing today on my slot 1 test bench setup - testing out the Yamaha DB50XG that I got inside of a computer I bought recently. I thought it had some fault because the main chip gets relatively hot going up to 60c?
1712874066305_100.JPG

The main sound chip does get quite toasty on DB50/60XG etc. and it is quite normal. I have heatsinked it on my devices that have it. At least it doesn't get as hot as early EMU8000 chips on AWE32 cards, those get even hotter, enough to even make PCB change color over time D:

PcBytes wrote on 2024-04-12, 21:53:

I just got the 10.2 package from AMD's website and it does list X1650 Series in the XP inf, CX_95952. It's the same driver I used on my Sapphire X1650 Pro until its death alongside a HIS X1950 Pro, both AGP.

Thank you. When I Googled ati 10.2 xp I got this Guru3D link:
https://www.guru3d.com/download/ati-cat ... -(32-bit)/

I have gotten drivers from there for years and I didn't see any reason to believe it was modified, however, the downloaded file is only 46MB compared to 77MB for the one straight from AMD's site. Checking the .inf I see that this particular package is missing everything prior to the HD2000 series. The file on guru3D was uploaded 14 years ago. 😕

*sigh*

Was someone really going around and modifying drivers to remove things? Why are the X series also missing from all of the other packages I have which should support them? 😕

I will try installing the 10.2 set from AMD.

Is it possible that AMD actually modified the "Legacy" packages to include as many cards as possible, where as the old downloads from 10-15 years ago lacked certain devices originally without modifying the inf? So confusing...

EDIT: Yep, 10.2 "Legacy" direct from AMD works fine. I still don't understand why the X series was removed from all of the other packages I have though. I'm checking all the .inf files manually, and they either only go back to the HD2000 series or they only cover the 9000 series. This must be why Windows was saying I only had drivers for those, with the X series missing entirely. This is what I get for going to threads about how to find the "best" ATI drivers, ending up with some "best package" from a third party site, and it turns out it only supports like a third of the cards the driver package is supposed to support.

EDIT2: Woohoo, the card just blue screened during 3Dmark 2001SE. Awesome! 🙄
I left this thing running a game for 3 hours by accident the other day with an FX 5950 Ultra in it and it never crashed. As soon as I try to test ATI cards it's crash crash crash. I will try using Driver Cleaner and starting with just the working 10.2 driver. If that doesn't work, I'm moving on. 🤷

Yeah, if your test machine is the NF7, you might wanna change it. I found out after being close to going insane that nForce 2 and certain Radeons (I think 9600, X1300/1550/1600/1650) won't play nice unless you use a certain Catalyst version. IIRC 9550 and 9600s will only work with 6.2. No idea which works with X series unfortunately 🙁

PcBytes wrote on 2024-04-13, 00:03:

Yeah, if your test machine is the NF7, you might wanna change it. I found out after being close to going insane that nForce 2 and certain Radeons (I think 9600, X1300/1550/1600/1650) won't play nice unless you use a certain Catalyst version. IIRC 9550 and 9600s will only work with 6.2. No idea which works with X series unfortunately 🙁

Yeah, it is the NF7-S.

Do you know if Nforce 3 Ultra chipsets (S939 with dual core) have any issues like this?

Also, its strange because I used a 9600 Pro in this board as my main PC back in the day and it was rock solid.

Cosmic wrote on 2024-04-12, 04:51:

Thermalwrong wrote on 2024-04-12, 00:52:

Nice, what did you do to get the video cards working again? I've had non-working cards work from just cleaning the PCI contacts […]
Show full quote

fillosaurus wrote on 2024-04-11, 20:42:

Wrangled with some ISA soundcards today. To be more specific, Yamaha SW20-PC, Reveal SoundFX SC400 Pro 16, Samsung Audio Magic X and an OEM HP Aztech 2320.
And somehow I revived 2 graphic cards, a no name S3 Trio 3D/2X AGP and a Matrox G200 PCI.

Nice, what did you do to get the video cards working again? I've had non-working cards work from just cleaning the PCI contacts or re-seating the BIOS.

I've been doing the same thing today on my slot 1 test bench setup - testing out the Yamaha DB50XG that I got inside of a computer I bought recently. I thought it had some fault because the main chip gets relatively hot going up to 60c?
1712874066305_100.JPG
It sounded very wrong playing passport.mid and canyon.mid but sounded fine in games. It seems to just be some mapping issue with those midi files since XG midis from FF7 play fine and games sound excellent with it.
Just to be safe, the Yamaha chip by the ROMs now has a flat ceramic 40mm heatsink on it since I've got a bunch of them for free and they do work.
*sigh* the card it fits best is the CT2950 soundblaster 16 but I heard the hanging note bug within just a few minutes of gameplay - time to modify the soundblaster's DSP with one from the Soundblaster DSP thread.

Then thought I'd finally test out my Tualatin mod on my MS6905 slotket - last time I used this slotket I thought I had completely broken it since neither tualatin or P3 worked so I read up and re-did it. Sadly I did that ~4 months ago and have no idea which mods I did. I plugged it into the board to test it and it works! 1.2GHz Tualatin celeron working on my Aopen AX6BC board.
IMG_3155 (Custom).JPG
Oh it was toasty though for some reason after about 2 minutes of operation and it's got a junk bin HSF on it right now just for basic testing.
Why is it so toasty??? The CPU voltage on the slotket is set to auto, that should mean it will go at 1.5v like the CPU says right???
IMG_3157 (Custom).JPG
Uh... checked with multimeter and it's running at 2.08 volts...
This 1.2GHz Celeron may not work for very long now - the slotket has manual voltage settings and it's now running at about 1.53 volts being set for 1.5v operation. Oops!

Very cool that the slotket adapter works! And glad you found and fixed the voltage before it was too late too 😁

What graphics card are you running alongside it?

It's a powerful Geforce FX5200 😁 From this post but since then it got some 'ramsinks' that were sitting around in my parts bin: Re: What retro activity did you get up to today?
Since this one has the best overclocking RAM out of all the FX5200's I've got, it gets to look cool.
This is just a thrown together test computer since this Aopen AX6BC the most accessible slot1 board with full CPU voltage support that I consider kind-of expendable - it couldn't run games with the Tualatin CPU and would just restart, it would also only identify the CPU as a P-III MMX with some NaN MHz value.

So I thought I'd update the bios using my TL866 programmer. It was all going great until I put the EEPROM in backwards. The POST analyser card now only shows "--" with it. The rom still reads okay on the programmer and I was very worried I'd killed the board? since the ROM connects directly to the southbridge. My one spare 2mbit DIP EEPROM still runs the computer but it's unusable since it's got an AMI bootblock that can't be reprogrammed.
Gonna have to wait until some replacement EEPROMs come in until I can test this computer setup out more - such a shame, I was gonna test out the 1GHz 100MHz bus coppermine CPU on it after this.

Tiido wrote on 2024-04-12, 21:56:

Thermalwrong wrote on 2024-04-12, 00:52:

I've been doing the same thing today on my slot 1 test bench setup - testing out the Yamaha DB50XG that I got inside of a computer I bought recently. I thought it had some fault because the main chip gets relatively hot going up to 60c?
1712874066305_100.JPG

The main sound chip does get quite toasty on DB50/60XG etc. and it is quite normal. I have heatsinked it on my devices that have it. At least it doesn't get as hot as early EMU8000 chips on AWE32 cards, those get even hotter, enough to even make PCB change color over time D:

Thanks for confirming, that makes sense since it'd be an effects processor?
I don't think I've ever seen a cooked EMU8000 chip although I've only got later SB32 and AWE64 cards, perhaps those are less toasty.

Thermalwrong wrote on 2024-04-13, 02:23:

Thanks for confirming, that makes sense since it'd be an effects processor?
I don't think I've ever seen a cooked EMU8000 chip although I've only got later SB32 and AWE64 cards, perhaps those are less toasty.

Not just effects but all of the sound generation. The other big chip is just the CPU issuing commands to this one according to MIDI input.

There are 3 types of bare EMU8000, the earliest one with Creative logo and only CT1971 (runs hot) but no "EMU-8000", later one with "EMU-8000" + "IC405A" written on it which runs hot and latest one with "EMU-8000" + "IC405B" written on that runs cool. Later ones inside a big AWE chipsets also run cool.

There is a short thread about this here (with pictures):

Overheating EMU8K on AWE32 CT2760

Thermalwrong wrote on 2024-04-13, 02:23:

So I thought I'd update the bios using my TL866 programmer. It was all going great until I put the EEPROM in backwards. The POST analyser card now only shows "--" with it. The rom still reads okay on the programmer and I was very worried I'd killed the board? since the ROM connects directly to the southbridge. My one spare 2mbit DIP EEPROM still runs the computer but it's unusable since it's got an AMI bootblock that can't be reprogrammed.
Gonna have to wait until some replacement EEPROMs come in until I can test this computer setup out more - such a shame, I was gonna test out the 1GHz 100MHz bus coppermine CPU on it after this.

Oh wow that's crazy! So it sits at "--" with the original BIOS chip installed but it still sort of boots with the other 2mbit EEPROM? I guess that's a good sign that it's not totally dead. I just wonder how that works if the original chip still reads correctly in your programmer.

Hope you have success with a new EEPROM when it arrives!

Ozzuneoj wrote on 2024-04-13, 00:12:

Yeah, it is the NF7-S. […]
Show full quote

PcBytes wrote on 2024-04-13, 00:03:

Yeah, if your test machine is the NF7, you might wanna change it. I found out after being close to going insane that nForce 2 and certain Radeons (I think 9600, X1300/1550/1600/1650) won't play nice unless you use a certain Catalyst version. IIRC 9550 and 9600s will only work with 6.2. No idea which works with X series unfortunately 🙁

Yeah, it is the NF7-S.

Do you know if Nforce 3 Ultra chipsets (S939 with dual core) have any issues like this?

Also, its strange because I used a 9600 Pro in this board as my main PC back in the day and it was rock solid.

IIRC NF3 should be issue free. Not necessarily 939 but I know my K8N ran a 9550 with no crashes whatsoever.

As for a platform that you could run the X1650XT crash-free - 865PE or 865G. Mine ran on a Gigabyte 8IG1000 Pro2 w/ 2.8GHz Prescott HT (MAXIMUM CRISPY) through its life and the last few months it was alive, on an ABIT IS7-E v1.2.

The nF2 crash issue isn't only on the NF7 btw - I had managed to replicate it across 5 different boards - ASUS A7N8X, MSI K7N2 Delta-ILSR, Epox 8RDA6 Plus, Gigabyte 7N400-L, Soltek SL-75FRN2-RL "Golden Flame". All would crash with a 9550/9600, and maintain stability with a 9700 Pro. For the 9550/9600, Catalyst 6.2 cured the BSOD issues. Again, no idea what Catalyst version would prevent X1650XT from crashing, unfortunately.

PcBytes wrote on 2024-04-13, 07:41:

IIRC NF3 should be issue free. Not necessarily 939 but I know my K8N ran a 9550 with no crashes whatsoever. […]
Show full quote

Ozzuneoj wrote on 2024-04-13, 00:12:

Yeah, it is the NF7-S. […]
Show full quote

PcBytes wrote on 2024-04-13, 00:03:

Yeah, if your test machine is the NF7, you might wanna change it. I found out after being close to going insane that nForce 2 and certain Radeons (I think 9600, X1300/1550/1600/1650) won't play nice unless you use a certain Catalyst version. IIRC 9550 and 9600s will only work with 6.2. No idea which works with X series unfortunately 🙁

Yeah, it is the NF7-S.

Do you know if Nforce 3 Ultra chipsets (S939 with dual core) have any issues like this?

Also, its strange because I used a 9600 Pro in this board as my main PC back in the day and it was rock solid.

IIRC NF3 should be issue free. Not necessarily 939 but I know my K8N ran a 9550 with no crashes whatsoever.

As for a platform that you could run the X1650XT crash-free - 865PE or 865G. Mine ran on a Gigabyte 8IG1000 Pro2 w/ 2.8GHz Prescott HT (MAXIMUM CRISPY) through its life and the last few months it was alive, on an ABIT IS7-E v1.2.

The nF2 crash issue isn't only on the NF7 btw - I had managed to replicate it across 5 different boards - ASUS A7N8X, MSI K7N2 Delta-ILSR, Epox 8RDA6 Plus, Gigabyte 7N400-L, Soltek SL-75FRN2-RL "Golden Flame". All would crash with a 9550/9600, and maintain stability with a 9700 Pro. For the 9550/9600, Catalyst 6.2 cured the BSOD issues. Again, no idea what Catalyst version would prevent X1650XT from crashing, unfortunately.

Modded omega driver versions have been crash free from my experience. Catalyst version XP 32bit 6.2, 6.5 and win98 rad_w9x_omega_2637. Used boards Abit NF7 with mod bios and Chaintech Zenith.

I had crashes with both Omega and standard on Radeon 9550/9600 with anything past 6.2. Stock BIOS for all boards listed.

i'm dusting the old 940 opteron board, trying to max the ram and install the os to feel the power of all that ram stick 😁.
i got the feeling that my board all having some quirk somewhere.

the H8DCi will run the memtest just fine but will absolutely won't boot windows and boot linux sometime in a while or kernel oops when it like to oops.
the DK8N will recognized the ps/2 keyboard but after bios post, the keyboard will froze, at least usb keyboard works without having to change the bios setting. also usb controller is not working under linux for this one while the usb port of the H8DCi work fine under linux.

also it seems like my 750w psu (antec eag pro 750w) is having problem, after running DK8N with that PSU, the board will absolutely won't post (no post code at all and post card display four dash) and that gave me a fright,
changing the PSU to the other one in the picture revive the board so at least no permanent damage had been done yet.

PcBytes wrote on 2024-04-13, 07:41:

IIRC NF3 should be issue free. Not necessarily 939 but I know my K8N ran a 9550 with no crashes whatsoever. […]
Show full quote

Ozzuneoj wrote on 2024-04-13, 00:12:

Yeah, it is the NF7-S. […]
Show full quote

PcBytes wrote on 2024-04-13, 00:03:

Yeah, if your test machine is the NF7, you might wanna change it. I found out after being close to going insane that nForce 2 and certain Radeons (I think 9600, X1300/1550/1600/1650) won't play nice unless you use a certain Catalyst version. IIRC 9550 and 9600s will only work with 6.2. No idea which works with X series unfortunately 🙁

Yeah, it is the NF7-S.

Do you know if Nforce 3 Ultra chipsets (S939 with dual core) have any issues like this?

Also, its strange because I used a 9600 Pro in this board as my main PC back in the day and it was rock solid.

IIRC NF3 should be issue free. Not necessarily 939 but I know my K8N ran a 9550 with no crashes whatsoever.

As for a platform that you could run the X1650XT crash-free - 865PE or 865G. Mine ran on a Gigabyte 8IG1000 Pro2 w/ 2.8GHz Prescott HT (MAXIMUM CRISPY) through its life and the last few months it was alive, on an ABIT IS7-E v1.2.

The nF2 crash issue isn't only on the NF7 btw - I had managed to replicate it across 5 different boards - ASUS A7N8X, MSI K7N2 Delta-ILSR, Epox 8RDA6 Plus, Gigabyte 7N400-L, Soltek SL-75FRN2-RL "Golden Flame". All would crash with a 9550/9600, and maintain stability with a 9700 Pro. For the 9550/9600, Catalyst 6.2 cured the BSOD issues. Again, no idea what Catalyst version would prevent X1650XT from crashing, unfortunately.

Thanks, this is great info and should help quite a bit with testing.

I wasn't ready to accept defeat and decided to do my best to rule out overheating, so I actually disassembled the Sapphire X1650 Pro (sorry, thought it was an XT, it is a Pro), applied new thermal paste, cleaned and oiled the fan... and it still crashes at the same points in benchmarks. I forgot to mention that it also stutters heavily in some benchmarks where it shouldn't. Like 3Dmark2k1SE, Car Chase Low Detail. When the jet starts shooting at the trucks it "hitches" with each shot and the frame rate goes from ~200fps to almost nothing, then returns to normal. Also, Car Chase High is not as fast as it seems like it should be. Definitely something going on there.

Also, while messing with the card there were a couple odd things I ran into (unrelated, but kind of funny):

1. What is with the weird thermal paste they used on these? Not sure if mine had been repasted at some point in the past, but the stuff that was on it was actually still spreadable but had a distinct metal-flake look to it. Like a grease+metal combination that had probably changed and separated over the years. I wasn't too pleased to have to clean all the possibly conductive flakes from the GPU (which has exposed SMD caps all around the die). I have never seen paste like this in my life, but this is the first X series I have dismantled.

2. ATi drivers are kind of... funny. After getting them installed properly, I went to the SMARTGART settings in CCC and figured maybe changing something here could help if it was caused by some AGP weirdness (I had already tried with FastWrites on, and currently had them off in the BIOS). I had never seen some of these particular settings since I only ran an ATi card in my main PC for about one year (bought a 6800GT on release day). I switched off "AGP Write" and "AGP Read", rebooted and was presented with a 16-color desktop, as if I'd gone back to Windows 9x without video drivers. The kicker was that CCC no longer recognized the video card at all and provided no way to undo this setting, so I had to remove\uninstall the broken video devices through device manager. They were some super generic name I'd never seen before (not even a Microsoft Display Adapter)... something like "Video Adapter". After rebooting, everything was back to normal... but, hilariously, the SMARTGART page was now entirely absent from CCC. It's like they said "okay, yeah, that was on us... why would we give users settings that can do nothing but break the drivers? We'll just pretend we never saw that..." 🤣

EDIT: There's definitely something funky going on with ATI cards\drivers and this system. I just threw in the HD3450 AGP I was having issues with the other day and I realize now it has exactly the same problems. Stuttering in very specific places and blue screening suddenly. Too bad... I really really wanted to have 3dmark test results for as many cards as possible on the same PC. At this point I have probably 20+ results saved that are all kind of minor\obscure variants of cards (64bit vs 128bit cards for example). I can probably just use Catalyst 6.2 drivers and at least test the 9000 series on this one. I will try the X and HD series AGP cards on my Nforce 3 Ultra system.