Jump to content
MSFN is made available via donations, subscriptions and advertising revenue. The use of ad-blocking software hurts the site. Please disable ad-blocking software or set an exception for MSFN. ×

Running Windows 2000 on modern motherboards - USB issues


Recommended Posts

4 hours ago, bluebolt said:

re: my sound card
Are you referring to the integrated audio chipset (Realtek ALC889)?  As with the LAN, if you're trying to make the Gigabyte driver work, your better bet may be the Realtek website drivers.  I use R.274.  I notice they also have an older one (A.4.06), which I have never tried.

Softpedia appears to have some earlier Realtek audio driver versions for W2k Pro that you could try:

https://drivers.softpedia.com/get/SOUND-CARD/REALTEK/Realtek-HD-Audio-Codecs-Driver-222-for-2000-XP.shtml

Seems you're on a roll, so...keep rolling!

Well, yes and no.

Things are working quite well and I once again tried PAE. It detects all memory but again, the sound card only makes a tick sound in place of actual sound. If I try to play a music file, it sort of sounds like a jack hammer. lol

The card is actually a Creative Sound Blaster X-Fi Titanium Fatal1ty Professional Series sound card. I wonder why Realtek would work with PAE but not the Sound Blaster.

This is interesting, however: https://www.bleepingcomputer.com/forums/t/93761/how-can-i-enable-physical-address-extensionpae-feature-in-vista/?p=531767

Link to post
Share on other sites

You may be able to use the last XP drivers but the likelihood of success is very low because XP x86 was never made with large memory addresses (>4 GB) in mind. Server 2003 x86 drivers would be a good bet (because they have to deal with large memory addresses), if they existed, which they don't seem to being a consumer-grade component.

The Realtek drivers probably work better with PAE because their onboard audio is included in workstation (though probably not server) boards where one may consider OSes that do fully support PAE. Or maybe the driver developers were just open-minded.

Edited by win32
Link to post
Share on other sites
1 hour ago, win32 said:

You may be able to use the last XP drivers but the likelihood of success is very low because XP x86 was never made with large memory addresses (>4 GB) in mind. Server 2003 x86 drivers would be a good bet (because they have to deal with large memory addresses), if they existed, which they don't seem to being a consumer-grade component.

The Realtek drivers probably work better with PAE because their onboard audio is included in workstation (though probably not server) boards where one may consider OSes that do fully support PAE. Or maybe the driver developers were just open-minded.

I'll have to research into that. But I ended up having to reinstall because "something" corrupted my last Win2000 install so badly that I couldn't even get the recovery console to find it. So I dunno what happened. But I decided to not install the X-Fi card drivers this time and just go ahead and install the Realtek drivers. This is what I ended up with.

Win2k.thumb.JPG.3e4dc5073aa4e0bd320bff0ec0d017e7.JPG

I left the whole thing in because I do have my dual monitors working and I'm also playing audio right now via WMP11. It IS working.

I am going to look for Server sound card software. I never thought about Server/Consumer grade versions of Windows and support for >4GBs. That's some great thinking!

Now some more interesting follies. I stayed away from some of the later versions of Extended Core because videoprt.sys was flawed and would not allow for dual monitors to work, this was especially true in version 16a. For fun, I downloaded version 16d, which Blackwingcat only released in Japanese for some reason, and opened it up to poke in the files. I took notice that videoprt.sys was slightly newer in this release and language neutral. So I replaced the file included in 16a with the one in 16d and reinstalled the 16a package. Low and behold, dual monitors work. It was that single flawed file that actually was updated but not for an English release that prevents dual monitors from working. Then again, how many others but me here use dual monitors on Windows 2000? :whistle:

Link to post
Share on other sites

I've seen this USB problem mentioned before, and once again, I ask: How does your Windows 2000 even have any references to "Intel C610/X99 series chipset" or "Z68" or "X79" or X58" even? Unless you are using a modified installation source or are running an Intel Chipset INF update after installation these should not exist!

The Intel Chipset INF updates include garbage do-nothing INF files for later Intel USB2 controllers under 2K (and for Intel USB3 controllers under XP). All these files do is name the controller with it's proper Intel designation, and link back to 2K's USB.INF, but use the UHCI (USB1.0) install section instead of loading a proper USB2 (EHCI) driver. They do NOT properly link to a USB2 driver (.SYS file). (Examine the [USB_2K.NT] (2K) install section versus the [USB2.NT] (XP) section in one of these Intel USB files, I used "patusb.inf" for example.) However, since these Intel INF files are dated newer than any existing USB.INF file under 2K, 2K chooses these garbage files by default and complains if you want to use the older-dated standard driver.

13 hours ago, win32 said:

I wonder if deleting the offending chipset USB driver installers (wellsburgusb.inf, cougarpoint.inf etc.) from the installation media would force win2k to install the generic drivers in setup?

Should work.

If anyone really wants these controllers to be given their specific names, then someone will need to add all the Intel USB2 VEN&DEV ID's and their corresponding proper names to 2K's USB.INF, linking them to the proper EHCI driver install section. See NUSB for 98SE's USB2.INF for reference. This will also have the effect of making the updated USB.INF file have a newer date than the Intel ones, which should make 2K use it by default (although it may complain it's not signed, not sure if 2K does this like XP and later).

Edited by LoneCrusader
examine patusb.inf, add/clarify more info
Link to post
Share on other sites

Definitely modified installation sources are included; this is why I referred to:

22 hours ago, bluebolt said:

...one example of the installation procedure, which varies based on motherboard, mouse, the particular OS package...

The point is, in no instance is USB function achieved.  Not with factory-standard / vanilla W2k Pro, nor in any known package.  If anyone has information otherwise, I am wide open to it.

Link to post
Share on other sites

I am totally using a modified installation, which also made me wonder why these files even existed because obviously they'd be completely ahead of their time for vanilla Windows 2000. Unless they benefit *somebody* out there, they should be removed from archives in my opinion. If Windows 2000 detects and sees these products in your system and installs the drivers as usual yet driver functionality isn't working, what good are they to begin with?

Let's just say that I'm using Tomasz86's latest HFSLIP packages, which interestingly enough no longer contain SP5.1 but perhaps everything that's in it supersedes it so there is no longer a need to include it. But regardless, both packages I've used have these DOA drivers included with it, this newest package and the November 2012 update that I still have floating around on my own system. Going about installing the correct drivers was a breeze, even though inconvenient since you have to use all keyboard shortcuts to achieve this.

@LoneCrusader This thread was originally my goodbye to Windows 2000 thread until these two bright individuals led me on the correct path. I really thought this system would not be able to run it and that I'd have to jump ship to Windows XP x64. I still have it installed on another drive but with Windows 2000 literally working perfectly at this point, despite using Realtek Onboard audio which isn't as bad as I thought it would be, I'm able to use everything I need and still use Physical Address Extension.

Now, the question I'd like to bring up since part of a modern motherboard now includes much more than 4GBs of RAM. Is it possible to modify existing drivers, particularly ones like sound cards and Creative's SoundBlaster drivers to access or at least work with PAE and greater amounts of RAM? This is where I really become dumb so I'm just going throw this out. How much different are x64 bit drivers to 32bit drivers? I understand there's an algorithm that goes with 32/64 bits and that's why more RAM is able to be used natively in 64bit operating systems, but is there a reason we can't use that technology to our benefit?

Link to post
Share on other sites

Backporting x64 drivers to x86 will need recompiling. And that won't happen unless the drivers are open-sourced. :(

Fixing up the existing x86 drivers to support large memory addresses would require disassembling by someone who knows what they're doing, if that is even possible at all. (I'm studying in the wrong field)

And Creative is a company known for bad drivers and attempts to quash unofficial driver writers who fixed the shortcomings of their official ones; that business decision was reversed after universal scorn.

WAIT, did I say unofficial driver? For 32bit XP? Yes I did!

https://danielkawakami.blogspot.com/2017/01/sb-x-fi-series-support-pack-40.html

Hopefully they work better with PAE though again, home users running x86 Windows with full PAE are an incredibly small group. Even in the days when Server 2003 as a workstation was a thing, many people had 2 GB or less RAM.

Edited by win32
Link to post
Share on other sites
7 hours ago, win32 said:

WAIT, did I say unofficial driver? For 32bit XP? Yes I did!

https://danielkawakami.blogspot.com/2017/01/sb-x-fi-series-support-pack-40.html

Hopefully they work better with PAE though again, home users running x86 Windows with full PAE are an incredibly small group. Even in the days when Server 2003 as a workstation was a thing, many people had 2 GB or less RAM.

Good try, buddy! But unfortunately, nope. That was what I was using and they still don't work correctly. :(

I think our best bet is Blackwingcat, if he's willing to take on such a task. Depends on whether he knows how to do it or whether he wants to. But for now, I'm gonna make it to the 20th anniversary. :) Let's keep this thread going though, I like this discussion and I think we should try bringing up any other shortcomings that we can think of when trying to run Windows 2000 on modern hardware.

Edit: That's a different one than I grabbed, I think. I'm going to compare the files in this pack and see how they match up to the pack that I tried out. I'm almost scared to try it out because I don't want to mess up my installation that's going so well. lol

Link to post
Share on other sites
22 hours ago, bluebolt said:

The point is, in no instance is USB function achieved.  Not with factory-standard / vanilla W2k Pro, nor in any known package.  If anyone has information otherwise, I am wide open to it.

Let me be sure I'm understanding the problem... If you perform a factory vanilla installation of 2K Pro with SP4 on such a system, USB2 does not work?

I assume these problem systems also have no USB1 controllers showing up in the Device Manager as well?

If the above is the case, then chances are the problem is that the file USBD.SYS is not being copied during the installation of USB2 controllers. This file is not listed in the copy sections for USB2 controllers in USB.INF, but the USB2 drivers are still dependent on it. It is only listed in the copy sections for USB1 controllers, which at the time the INF was written, would also automatically exist on a USB2 system. On newer systems with no USB1 controllers to install, the file is not copied. This issue also affects Windows 9x.

Verify that USBD.SYS exists on the resulting system. If not, then copy it manually to SYSTEM32\DRIVERS, reboot and see if the problem is cured.

If USBD,SYS does exist on the resulting system, then verify that USBHUB20.SYS exists on the system. It seems to be called in both a copy and delete operation for the same USB2 Hub installation routine.

If anyone can verify these conditions one way or another I may be able to sort it out... :unsure:

22 hours ago, Tommy said:

This thread was originally my goodbye to Windows 2000 thread until these two bright individuals led me on the correct path. I really thought this system would not be able to run it and that I'd have to jump ship to Windows XP x64. I still have it installed on another drive but with Windows 2000 literally working perfectly at this point, despite using Realtek Onboard audio which isn't as bad as I thought it would be, I'm able to use everything I need and still use Physical Address Extension.

No problem. :thumbup I'd just like to get to the bottom of this USB issue since I've seen it reported before and I can't understand what the problem is. It should be simple to fix!

  • Like 1
Link to post
Share on other sites
21 hours ago, win32 said:

Backporting x64 drivers to x86 will need recompiling. And that won't happen unless the drivers are open-sourced. :(

Fixing up the existing x86 drivers to support large memory addresses would require disassembling by someone who knows what they're doing, if that is even possible at all. (I'm studying in the wrong field)

And Creative is a company known for bad drivers and attempts to quash unofficial driver writers who fixed the shortcomings of their official ones; that business decision was reversed after universal scorn.

WAIT, did I say unofficial driver? For 32bit XP? Yes I did!

https://danielkawakami.blogspot.com/2017/01/sb-x-fi-series-support-pack-40.html

Hopefully they work better with PAE though again, home users running x86 Windows with full PAE are an incredibly small group. Even in the days when Server 2003 as a workstation was a thing, many people had 2 GB or less RAM.

So I decided to give the latest version a try.

Bricked my install twice with it. The one time I tried the repair and it worked, sort of. But it seems installing that driver causes issues by corrupting the system part of the registry. I have no idea why. But after installing it and then reinstalling the Extended Core because of the XP version of wdmaud.drv and wdmaud.sys causes drivers installed afterwards to just not play. I went and replaced them with regular Windows 2000 files but I really don't want to do a reinstall again just to see if it'll make a difference. I wish it was easier to make a backup of the system hive. Maybe if I just go ahead and boot into recovery console and do it from there....

Link to post
Share on other sites

Another thing that bugs me is the last set of official NVIDIA drivers seems to cause the OS to get stuck in third-stage boot on "preparing network connections" for about a minute or so. I observed this behaviour on two separate machines with a GeForce 210 and a Quadro FX 3800. Turns out that is the NVIDIA display driver helper service starting, which isn't really necessary. Disabled that and startup time dropped from 3:40 to 1:38!

I also decided to investigate my CPU overheating issue, which has been exacerbated by a recurring sequences of heat waves with low temperatures near 25 C (room temp in the low 30s). A fix to the issue was to disable turbo mode, which caused CPU temps to drop by about 10 C under load.

When a newer CPU with turbo implemented through dynamic frequency scaling is used, Windows 2000 will run it constantly at turbo frequency. Once I disabled it, the OS was able to properly make use of enhanced speedstep, idling the CPU at ~1.6 GHz instead of 3.2 GHz.

Edited by win32
  • Like 1
Link to post
Share on other sites
1 hour ago, win32 said:

Another thing that bugs me is the last set of official NVIDIA drivers seems to cause the OS to get stuck in third-stage boot on "preparing network connections" for about a minute or so. I observed this behaviour on two separate machines with a GeForce 210 and a Quadro FX 3800. Turns out that is the NVIDIA display driver helper service starting, which isn't really necessary. Disabled that and startup time dropped from 3:40 to 1:38!

I also decided to investigate my CPU overheating issue, which has been exacerbated by a recurring sequences of heat waves with low temperatures near 25 C (room temp in the low 30s). A fix to the issue was to disable turbo mode, which caused CPU temps to drop by about 10 C under load.

When a newer CPU with turbo implemented through dynamic frequency scaling is used, Windows 2000 will run it constantly at turbo frequency. Once I disabled it, the OS was able to properly make use of enhanced speedstep, idling the CPU at ~1.6 GHz instead of 3.2 GHz.

Thank you for bringing this up! This always irked me too. I always ended up disabling the display driver helper service after setting up dual monitors.

Edit: You WILL need to brute force the INF file in order to work as Blackwingcat does, otherwise it will tell you there is no file that contains any information on your hardware, so it pretends not to see it

However, last official drivers? Which drivers are you using? I'm just curious. As I was poking around tonight, I discovered a gem, a very interesting gem that even Blackwingcat doesn't have on his blog. While he may have unofficial drivers on his site, I found a very recent (um, 9 years ago recent lol :crazy: ) NVIDIA driver that actually DOES officially support Windows 2000.

https://www.nvidia.com/object/winxp-257.21-whql-driver.html
https://www.nvidia.com/object/winxp-258.96-whql-driver.html (Discovered after posting this, true last official NVIDIA Windows 2000 release)

Really? An official 2xx.xx driver from NVIDIA? Why, yes it is! How many people know about this? If you have a Quadro, you're in even better position for a newer driver!

https://www.nvidia.com/download/driverResults.aspx/25619/en-us

Whaaaaaa????? An OFFICIALLY LISTED Windows 2000 driver!!!! However, it only lists Quadros as supported and NOT GeForce. The files are a bit smaller in this release as well but it leaves me wondering if we can break into these drivers and mod them to support newer things.

I think I'm a lot more impressed by the GeForce/Quadro release. And no, there is no need for unofficial kernels either, it works on vanilla Win2k.These gems are hard to find because of the way the driver search works, it really doesn't want you to find these older drivers, nor are they even on the older archive page. So I'd really suggest grabbing it while you can before it so happens to disappear. The best card these drivers can support are a GTX 480 which I don't think is anything at all to sneeze at. And yes, it supports dual monitors out of the box too. :D

new driver.PNG

Release 260.89 seems to be the first Windows XP exclusive driver, but interestingly enough keeps the win2kdualview in its driver file, so *it* may still work with a little bit of brute force.

Edit: The INF file will not work without modding it like Blackwingcat does. Add/remove hardware complains there's no file that contains information on your hardware, like it pretends it doesn't exist. If I wasn't so happy with my current installation, I'd totally try it out. Maybe someone else with a sandbox wants to try it out? Heck, it's so close and supports all the same hardware, you could probably just copy the INF file from 258.96 and edit the header information for the 260.89 release and it would possibly still work? Although it seems to be the first driver pack that has the "new" layout in files/folders.

nvidia.PNG

No joke, no mods, straight off NVIDIA's website. :thumbup

Link to post
Share on other sites

Oddly enough, when I discovered that I was getting an FX 3800 in my workstation, I actually put Windows 2000 as a filter for the driver search on the NVIDIA site and found 261.19 (December 21, 2010), which I posted in the "last versions of software for Windows 2000" thread. The changelog indicates that it is the very last one for Windows 2000.

I used 258.96 with my GeForce 210, and @tomasz86 listed a few later versions that are working to some degree somewhere in the USP 5.2 thread. Initially, I was unaware that official win2k support stretched that far (NVIDIA started listing their drivers as XP-only in about 2007 or so) and used @blackwingcat's driver. I wasn't satisfied because I'd get a blue screen every time I'd try to stand by or shutdown.

Edited by win32
  • Like 1
Link to post
Share on other sites

Glad you knew about them, I never did! I was just poking around tonight and found them. Now I hope anyone who needs them can benefit them now. Good thinking on finding those drivers. I hate their catalogue search. I'd rather just pick what I want and go with it, not having it treat me like an id*** and act like I don't know what I'm downloading. I've been using Forceware 182.50 for at least two years or better because I didn't even know about these. Like you, I had no idea NVIDIA support spanned that far into time. So even a bit beyond the EOL cutoff date.

Forceware 261.19 is Quadro only which I could use but since I bid and won on a GeForce GTX 260, I'd rather keep it open for GeForce instead of closing the window to it and making it Quadro only.

Link to post
Share on other sites

I just saw the edits to your posts and although I just paved over my sandbox, @tomasz86 has tested some of the post-258.96 drivers already:

261.19 works perfectly for me, but it seems that 306.81 may do fine with a couple files copied over from an older release. 306.81 supports up to the GeForce GTX 650 Ti/GTX 680!

edit: then I read some more and heard about potential instability and reliance on unofficial updates. Nothing is for sure unless someone with a 500/600 series GeForce + > 4 GB RAM comes forward

Edited by win32
  • Upvote 1
Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...