Is the chipset AGP driver loaded?

Edit: whoops, make that PCIe drivers... I are thinking that extra fiddling is required for PCIe on 2k as it is on 9x.

I don't think it's a PCI driver issue. I think Windows 2000 treats PCIe like an extension of the PCI bus.

My 6800 GS (PCIe Gen 1) functions correctly with the last GeForce driver that will install on Windows 2000.

Windows 2000 is based on the NT kernel. Nothing to do with 9x.

Foster, I would hit up Blackwingcat since he is the one that modified the nvidia drivers to work on windows 2000 for the 980ti but the BWC kernel is required. He should have a greater idea if what is going on. If no luck with vanilla then the bwc kernel is pretty minimal if you just wanted to install that.

For vanilla I just use my quadro fx 1300 which works well.

DosFreak wrote on 2021-11-20, 17:50:

Windows 2000 is based on the NT kernel. Nothing to do with 9x.

I meant the same kind of fiddling, not the exact same fiddling. For example DirectX is required on 2000 and XP as it is on windows 98 to play DX games, doesn't say anything about what kernel each uses, or that it's all the same binaries.

I'm familiar with Blackwingcat's work. I've used the Extended Kernel to get both the Quadro 4000 and a GT 640 working with Windows 2000 in the past. I'm taking on building an overkill Windows 2000 machine using only the vanilla kernel (as officially updated by MS) as a personal challenge (I could run my Windows XP drive if I want to use the XP kernel).

I'm considering getting a GeForce GTX 285 for this build because it's supported by the last "certified" Nvidia driver release (197) for Windows 2000, but I wanted to know if anyone had success with the 200, 300, or 400-series cards without using the Extended Kernel first.

I've seen evidence online that folks have gotten the 8-series cards to work, but nothing after them.

I have a high end AGP Radeon with 2k and it works well.

kolderman wrote on 2021-11-20, 18:54:

I have a high end AGP Radeon with 2k and it works well.

I have a Radeon x1900 GT which works quite well with Windows 2000. I'm looking to beat it.

I also have a Radeon HD 4850 that might work with some driver tweaks, but again, I'd like to beat that too.

It's also possible that Nvidia's Release 258 was just a bad driver for the Quadro 4000 given how new it was at the time of release. As I recall, this driver doesn't work well with Windows XP either. It's also possible that the error doesn't affect any of the related GeForce cards.

I'd just rather not buy yet another video card without hearing if anyone else has had success.

Radeon HD 2xxx/3xxx can be installed on vanilla kernel with simple file editing. And with some more shenanigans, everything up to HD 6xxx.

http://blog.livedoor.jp/blackwingcat/archives/571484.html - browse and use at your own risk.

I know that an NVS 290 works on stock kernel with 182.46. Never tried to go any further than that on stock kernel.

A GT 240 works on extended kernel with 197.45 but I never tried 197.45 on stock kernel (though it is said to work...)

bakemono wrote on 2021-11-21, 08:29:

I know that an NVS 290 works on stock kernel with 182.46. Never tried to go any further than that on stock kernel.

A GT 240 works on extended kernel with 197.45 but I never tried 197.45 on stock kernel (though it is said to work...)

It looks like the NVS 290 is an 8-series card, so it make sense it would work with Windows 2000 given that I've seen screenshots of other 8-series cards (such as the Quadro 4600) working in Windows 2000 using the vanilla kernel.

If someone has gotten a GT 240 working, then it stands to reason that a GTX 285 would also work. Have you seen a report of the GT 240 working with the vanilla kernel somewhere online?

fosterwj03 wrote on 2021-11-21, 16:34:

If someone has gotten a GT 240 working, then it stands to reason that a GTX 285 would also work. Have you seen a report of the GT 240 working with the vanilla kernel somewhere online?

Well, there is a blog comment from 2010, in Japanese, from someone who says they used a GTX 260 with no problems.
http://blog.livedoor.jp/blackwingcat/archives/1114373.html
And one here about using the 257.21 driver to run a GT 240
http://blog.livedoor.jp/blackwingcat/archives/1370860.html

bakemono wrote on 2021-11-21, 22:01:

Well, there is a blog comment from 2010, in Japanese, from someone who says they used a GTX 260 with no problems. http://blog.li […]
Show full quote

fosterwj03 wrote on 2021-11-21, 16:34:

If someone has gotten a GT 240 working, then it stands to reason that a GTX 285 would also work. Have you seen a report of the GT 240 working with the vanilla kernel somewhere online?

Well, there is a blog comment from 2010, in Japanese, from someone who says they used a GTX 260 with no problems.
http://blog.livedoor.jp/blackwingcat/archives/1114373.html
And one here about using the 257.21 driver to run a GT 240
http://blog.livedoor.jp/blackwingcat/archives/1370860.html

Google's translation of Japanese to English is always a joy to read. You're right, it looks like someone did run a GT 240 without using the extended kernel. That gives me some hope.

I decided to roll the dice and purchase a used GTX 285. I'll let you all know how it works once I receive it and get it installed in my test rig. It should give 3D applications a huge boost if it works.

I can now report that the GTX 285 does work with the Windows 2000 vanilla kernel (Service Pack 4) using both the GeForce Release 197 and Release 257 drivers. Unfortunately, the nVidia Control Panel app does not work using either driver (it reports a .DLL error just like it does for my Quadro 4000). The driver itself functions correctly giving me all of the resolutions supported by my monitor up to 32-bit color. DirectX 9 also passes all diagnostics. The Display Properties dialog and task bar app function correctly as well.

I can also say that the fan on my BFG GTX 285 OC doesn't sound bad at all even when the GPU got up to 86 degrees C.

Just for fun, I ran some Crysis benchmarks using all "High" settings with anti-aliasing "Off". It's definitely playable at both 1024x768 and 1080p (barely) using these settings. I ran the benchmarks using both the 197 and 257 drivers. Release 257 performed slightly worse, but likely within the margin of error for only one run each (results not shown).

fosterwj03 wrote on 2021-11-20, 15:59:

I've been trouble shooting a Nvidia driver issue in Windows 2000. I'm trying to get a Quadro 4000 to work in Windows 2000 SP4 w […]
Show full quote

I've been trouble shooting a Nvidia driver issue in Windows 2000. I'm trying to get a Quadro 4000 to work in Windows 2000 SP4 with the stock kernel (i.e. not the extended kernel from Windows XP). While the GeForce and Quadro Release 258 drivers (which support the Quadro 4000) will install on Windows 2000, the drivers fail to load at startup and the Nvidia control panel app fails with a series of .DLL system call errors (system calls that work in Windows XP).

Oddly, the same system with the same drivers will load when I have a GeForce 6800 GS in the slot instead of the Quadro 4000. The Nvidia control panel from Release 258 also functions correctly with the 6800 GS.

I can only conclude that the driver set executes different system calls (intended for Windows XP) when it detects the presence of the Quadro 4000. I wonder if this is caused by some sort of requirement for DirectX 10 which Windows 2000 SP4 does not support. I also wonder what other cards encounter this problem given that the drivers support the GeForce 200, 300, and 400-series (all DirectX 10 capable).

Has anyone here gotten a GeForce 200, 300, or 400-series card to work with Windows 2000's stock kernel? If so, how did you get around the issue I described above if you encountered it at all? Thanks.

I installed a GTX 460 on vanilla 2000, with the 258.96 xp geforce driver. Everything works except for the hd audio driver. Even the dualview works and the control panel works. Said system has an i3 9100f, and the only reason I had to install the extended kernel was because of the lack of good web browsers for 2000. Although at some point when I can afford it and someone *finally* ports a firefox based browser to vanilla 2000, I would love to build a vanilla 2000 rig as well, maybe with a 4790k and 2 480s in SLI (I am assuming if the xp driver supports 2 way sli it probably would also work in 2000). Maybe when I make this build I could use it for video editing, and maybe play some Doom 3 on it 😉

That's great! I haven't tried the install from Device Manager trick yet, but I might give that a try soon.

I might also watch for a deal on a GTX 480. If the price is right, I might invest in an upgrade. The 480 would provide somewhere between 15 and 60% performance boosts over the 285 depending on the application.

I purchased a GTX 480. I'll have to see how it works with Win2k and how it performs vs my GTX 285. I'll do some benchmarks on the GTX 285 this weekend for a future comparison.

I tried to reinstall the GeForce drivers using only Device Manager (after uninstalling the previous drivers). I'm afraid that it didn't work. Windows booted with VGA graphics and complained that the drivers were meant for Windows XP. I tried different sets of drivers all with the same result. While I'm using the Advanced Server version of Windows 2000, I doubt that caused the driver install to fail.

I went back to using the Nvidia setup program to install the driver. That restored my graphics.

Small update: I got the GTX 480, but it short circuited on my first power-up. It ended up damaging my motherboard. I switched over to another motherboard, and, while testing my GTX 285 on the new board, I noticed some strange behavior. The CPU no longer performed properly in single-threaded applications, giving me about 60% of the expected performance in Windows 2000. Oddly, multi-threaded performance behaved as expected. I thought that the GTX 480 short had damaged the CPU as well (a Xeon E3-1280 V2).

I just got a replacement CPU (a Xeon E3-1275 V2), but it performed similarly to the E3-1280 V2 in single-threaded workloads. An i5-3570, on the other hand, performs properly with single threaded workloads. I did some trouble shooting, and it turns out I only see this behavior when I have Hyperthreading enabled on the Xeon. The Xeon behaves properly in single-threaded workloads with Hyperthreading disabled. This only seems to happen in Windows 2000 Advanced Server.

I didn't see this before I got the GTX 480. I performed some benchmarks before ordering the GTX 480 so I could have some comparison data, and I got expected performance from the CPU with Hyperthreading enabled (at least I think I had it on at the time). The only other thing I changed in the setup recently was that I installed .Net framework to see if I could get all of the Nvidia control panel features working (it didn't change anything). I doubt that would change the single-threaded behavior, but I don't have time right now to to do a clean install of Windows 2000 to rule out .Net as the culprit.

At least I can conclude that my E3-1280 V2 isn't damaged. It's just a weird situation, though. I'll keep using the E3-1275 V2 since my backup motherboard can use the Xeon's on-board GPU (handy for OS's like BeOS).

I got the CPU performance back. I uninstalled all of the .Net framework system updates, but those didn't seem to have any impact on the CPU. I then uninstalled a Windows 2000 Update Rollup for SP4 that I had installed with .Net. That solved the problem. I can only think that somehow that rollup includes a security feature that puts a fair bit of load on each core, but that wouldn't explain why it only occurred with Hyperthreading. Regardless, I now have my Windows 2000 Advanced Server install back to normal.