LoneCrusader

Thanks for the reply. That might do the trick. For that matter even a later version might work; 3DMark03 is the last for 98SE according to Wikipedia. Now to find time to run tests...

Fair enough. Since you did talk to him, ask him whether or not the driver he was given was the 82.69 driver or a different version. If any version later than 82.69 exists for 9x, it needs to be collected and preserved.

What do you need him to tell you that is not already available here? You can use my INF's for ICH6/7/8/9 and you already know about the NVidia 82.69 driver. Whether or not the 512MB card works or not depends on many factors, but all 7xxx cards can work with the driver provided other conditions are favorable. There are some idiosyncrasies with getting the cards to work, but many of them are directly motherboard related and therefore system and configuration dependent.

Does anyone know of a Video Benchmarking program that will run under Windows 9x and has the ability to test Video RAM above 256MB?

Are your Intel processors Socket 478 or Socket 775? If you have good Socket 478 processors, I would recommend boards based on the Intel 865/875 chipsets, because virtually all of these come with full complements of Windows 98SE drivers and are known to be very stable and powerful. Examples: Intel D875PBZ Intel D865PERL MSI 875P Neo MSI 865P Neo2 (and other similar models) DFI Lanparty 875 (various models) SOYO P4-I875P Dragon 2 (these are my favorite but rather hard to find and prone to capacitor plague) If you have Socket 775's, there are a small handful of boards based on the 865/875 chipsets as well, but they are rare and usually only support an 800MHz Bus chip (i.e. no Core2 or later). In order to use a later Socket 775 chip you may need to look at 9xx or later chipset boards. Right now I am experimenting with some of these. Examples: MSI 865PE Neo3-F DFI Lanparty 875P-T MSI 965P Neo (still experimenting) MSI 975X Platinum (the one I have has strange issues, waiting for another one to rule out hardware probs) Gigabyte GA-965P-DQ6 (shows promise so far, still experimenting) Gigabyte GA-G1975X (ordered one of these, still waiting on it) If you want to go with AMD the only board I am familiar with that doesn't have some oddities is the Gigabyte GA-K8NS Ultra-939. I know everyone's experiences and opinions are different, but I have never seen a board made by Asus that was worth anything. I also prefer Award BIOS. Some developments have come up with regard to 512MB video cards. It looks as if only certain models are actually reporting 512MB and able to access it. Others may be limited to 256MB regardless of their actual onboard RAM. rloew can elaborate further on this. Since we're on this subject, does anyone know of a video benchmarking utility that runs on Windows 98 and can be used to test Video RAM above 256MB? EDIT: Moved my question to a new topic.

Are you looking to build a system just for 98, or are you looking to build a multiboot system that can be used for other things on later operating systems? Makes a lot of difference as to hardware choice, because if you want one strictly for 98, then I would recommend say, board x. But if you want a "higher performance" system for XP as well on the same machine, I would recommend board y. I'm in the process of experimenting with Windows 9x on post-9x supported boards, but unfortunately I don't have a lot of success to report so far. Whatever you do, do not attempt to use an Intel-branded board later than the D875PBZ. Some third-party boards based on Intel 9xx and later chipsets have possibilites, but not those built directly by Intel. Also, Intel or AMD, or does it matter? I don't have much experience with AMD since the good old K6-2 Super Socket 7 days.

Temperamental is certainly a good description. Now we know 512MB video cards can work under 9x. But apparently it depends critically on the hardware, so they are far from being a sure bet. I set out on a mission to get to the root of the issue a couple of months back, but I wonder just how deep does the rabbit hole go? I'm also seriously questioning the value of the tradeoff required in order to use a 512MB card. For example, I can use a 256MB card and have 3712MB or 3584MB of RAM available to Windows when running rloew's patch + 4GB of RAM. When I switch to a 512MB card, the available memory in Windows drops to 3328MB or 3072MB, depending on the board. So, in order to gain 256MB of video memory, one must sacrifice 512MB of system memory. Unless a solution to the aperture setting differences mystery (or a way to control BIOS memory allocation for the cards) is found to alleviate this, this almost makes it not worth the trouble. I'd hate to see what a 1GB or larger card would do to the system memory... For a long time I was unable to get any 512MB card to work at all under 98SE. Then one day they started working if I set 256MB AGP Apertures. Anything less and they will fail, contradictory to what loblo and others have observed on their own machines. The slightest change to any part of the Windows configuration can cause the card to stop working, and it's a nightmare to get it going again. I've had this happen on two machines, one of which I still haven't managed to cure. Then we enter the realm of PCI-Express cards. Since PCI-E boards don't have AGP Aperture settings, it gets even more convoluted. I have one machine where neither a 256MB or 512MB 7950GT PCI-E will work under 98SE, but they both work under 95. The most interesting discovery at this point is that the 82.69 driver (and presumably 82.16 as well since 81.98 does) works with Windows 95, and Windows 95 will use the 512MB cards with only a 64MB AGP Aperture setting, when 98SE will not.

rloew's patch has an /A option for handling certain older graphics drivers, but it apparently has no effect on systems using the latest version of the nVidia or ATI drivers (reported this and discussed with rloew). I do recommend the RAM patch regardless though. In my case any AGP Aperture setting less than 256MB will not work at all for any 512MB card that I have tried under 98SE (7200GS/7600GS/7950GT). There doesn't seem to be much logic to this, it apparently varies significantly by hardware. The real mind bender is that Windows 95 will work with the 512MB cards, with a 64MB AGP Aperture set.

The AGP Aperture and the size of the card are not the same thing. I wouldn't expect you to change the AGP Aperture to 512MB as most systems don't even allow this. I meant that in order to use a 512MB card, you will probably need to set a 256MB AGP Aperture in the BIOS. If it works like this, then you can experiment with smaller aperture settings.

What size is the AGP Aperture setting in the BIOS? Try setting it to 256MB and then remove the card in Safe Mode and reinstall the driver. 512MB video cards under 9x are very, very temperamental.

From all the info I've found it's almost impossible to mod an Intel BIOS in any meaningful way. Intel provides some tools for modifying the default settings for options that they have graciously already provided, but nothing whatsoever that would allow "pro-legacy" modifications. One tool I used to examine the BIOS files seemed to indicate that Intel BIOS'es are modified AMI BIOS'es. Another step in the wrong direction... I have yet to see an AMI BIOS that was comparable to AWARD. From just about all of the experiments I have run on post-9x hardware in the past two years, and from a handful of threads about similar hardware I've seen posted here, the lesson learned and the motto to be remembered is "Thou shalt not attempt to use an Intel-branded motherboard later than the D875PBZ with Windows 9x." Third-party boards based on Intel chipsets are sometimes OK, but the Intel-branded Intel-BIOS ones are junk from a 9x standpoint.

I have a custom INF set that provides 9x INF's for ICH10 chipsets, I just never got around to releasing it. It's an update of the "Slipstreamable" Intel INFs that are posted here somewhere. (The original set only goes to ICH9.) Currently it should wipe out all unknown devices except those associated with Intel Active Management/Management Engine. The bad thing is I have run several tests on an Intel DP43TF motherboard (ICH10) and it is an absolute nightmare to get anything working properly. Most of this appears to be due to the Intel BIOS (which is garbage and leaves the user virtually no control options). I'll wager that a third-party ICH10 board with an AWARD BIOS would function much better. There are many problems, but one specifically - I was unable to load any USB2 driver on this board, it will always cause Windows 98SE to hang during boot. This happens with or without using "SETUP /p i" to kill ACPI, and happens whether one uses the Microsoft Win2K USB2 stack or the OrangeWare USB2 stack. I wouldn't even make an attempt on such a project without rloew's RAM and SATA patches though.

You're welcome. I'm glad it's working for you and you're getting some enjoyment out of it. Good luck!

Are you certain that your USB HDD is formatted with the FAT32 file system and not NTFS? If it is formatted as NTFS, then Windows 9x cannot read it and will not assign it a drive letter. Connect your USB HDD to your Windows 7 machine and check the hard drive properties. I'm not familiar with what the exact steps would be under Windows 7, but we need to know what file system the drive is using. A couple of third-party drivers for reading NTFS under 9x exist, but it would be risky to use these if you happen to have any files larger than 4GB stored on your HDD, since 9x cannot support files greater than 4GB without other modifications. If your USB HDD is formatted NTFS, then you would need to back up your data and reformat it as FAT32; or, if you don't want to do that my advice would be to grab a cheap Flash Drive and format it as FAT32 in order to transfer the files.

Good progress so far. Sounds like you probably had a USB.INF that was provided with the Intel Chipset driver from the factory. What is the error given in the Device Manager? Did you reboot again after the USB controllers and hubs were detected and before connecting your USB HDD? If not, this may have led to the error. I may need to make a note about this in the instructions... You will probably need to manually remove the USB HDD from the Device Manager before connecting it again. If the drive is not connected you will need to do this from Safe Mode, as installed devices not currently connected will not show up in Normal Mode. Once you have done this, it should work. If not, please post any error messages that are displayed in the Device Manager.

The unknown devices get listed under Other devices, my mistake for not being more clear but you figured it out. Since you uninstalled USBSUPP and XUSBSUPP and then subsequently removed the devices and deleted the INF's, then Windows 95 should not have been able to identify the controller specifically. Somehow an INF for the USB devices still exists. This is what is used to identify the controller and load the other drivers. It sounds like the INF still exists, but the actual .SYS driver files are missing or not where they are expected to be in order to copy. Did you maybe install an Intel chipset driver at some point that provided an INF for USB? Once you clear the rogue INF you should be able to Cancel all of the driver requests during boot, as 95 will not be able to identify the controllers. Then reinstall XUSBSUPP. USBSTR95 was an unfinished project by another member here, and it was part of my inspiration for this project. I wanted to see it complete, but its author vanished and took his knowledge and source code with him. Problem is, I'm no programmer, so I had to get my friend rloew to basically recreate the entire package from scratch. Then I repackged it integrated with the USB Supplement. The original USBSTR95 works for a small handful of devices, mostly Flash Drives, but most USB Hard Drives do not work with it. XUSBSUPP provides all of the necessary drivers for USB controllers and USB storage. Adding any other package such as the original USBSUPP or a package from Intel, etc, will only complicate things further. Another reason I created it. You're welcome. XUSBSUPP should provide a solution for you. Just remember that Windows 95 cannot read NTFS partitions or use files >4GB without other third party programs. I'm sure you can get it sorted out. Try to find the rogue INF file and eliminate it. If that doesn't work, I will try to provide some more thorough instructions.

Sounds like uninstalling the existing USBSUPP is not removing all of the USB related files, particularly USB.INF. This is why the USB devices get redetected every time the machine restarts. Uninstall USBSUPP (you've already done this) Boot into Safe Mode and remove all USB devices and Unknown Devices. While still in Safe Mode go to the WINDOWS\INF folder and WINDOWS\INF\OTHER folder and delete USB.INF if it exists. In the \OTHER folder it may have a different name such as "MicrosoftUSB.INF" (INF's are sometimes copied and recopied here each time a device is installed) - just be sure you eliminate any and all copies before you reboot. This should clear out the problem and the USB devices should not be automatically reinstalled on each boot. When you reboot you should be able to simply cancel the driver install requests and get to the Desktop in order to install XUSBSUPP. Let me know if this works or if you have any more problems.

You're welcome! Glad it helped you! You may want to check out rloew's homepage for other bits of software that may help you continue using Windows 95/98 machines with your manufacturing hardware.

Be sure to note the difference I pointed out between the third-party add-on SATA controllers that have 9x drivers as with the ICH4 boards, versus the ICH5 and up boards that have the SATA controllers integrated as part of the chipset. For the later ones you will have to run in Legacy PATA mode always or have rloew's SATA patch in order for them to work with 9x. I was confused for a few minutes by the folders inside the chipset INF packages myself. The folders you see inside are only containers for ".CAT" files, which I have yet to figure out any useful purpose for. Windows 95 does not use these, hence no Windows 95 folder. When you actually run the SETUP program it should unpack INFs to the Win95Cur and Win95Old folders you mentioned. The 845 chipset should have it's own properly named INF, but there may also exist an "ALMADOR.INF" and "BROOKDAL.INF" - these were some early code names for ICH4 development and contain the same data as the finished ICH4 INF if I remember correctly.

As far as I know there is no real difference between the variations of ICH4, although I have never tried to run 95 on a laptop, only desktop systems. Search for versions 2.90.009 and 3.20.1008 of the Intel Chipset Software Installation Utility. One or both of these I know supports 95 and contains the ICH4 INF's; I just can't remember which one I used. I've never tried to use SATA drives with a PATA IDE board. I've never had a need to do so, since I have a large stockpile of PATA drives. I know these boards have the necessary settings to place the drives in Legacy PATA mode, so what you're looking for should be possible. Some early onboard SATA controllers came with 9x compatible drivers until the SATA controllers became integrated into the chipset. See the Intel D845PEBT2 board for reference.

Technically no. The last Intel chipset with official 95 support was the 845 ICH4 Pentium 4 chipset. I have several 845 chipset based boards that have working 95 drivers for every single device except the USB2.0 conrollers. If you're just looking for chipset support, it is simple enough to create INFs (or modify those for later OS'es) to properly name all of the devices. The problem comes when other devices on the board no longer get along well with Windows 9x, whether it be integrated Audio or Network cards, or more obscure non-configurable things like the ACPI controller. I have successfully used Windows 95 without any such problems on an Intel 875 chipset board. Not all drivers were available, but there were no resource conflicts or other such annoyances.

Tested and verified this works with Windows 95C and the 81.98 driver. Not tested other versions yet. Windows 95 requires both Messages 4 and 32 like Windows ME in order to Shut Down properly. Message 32 by itself will prevent SCANDISK being triggered by having to manually power off or reset.

Glad you got it working! Although this project and its solutions probably should be in its own thread and not be directly in-line with this thread, which is really meant to be a "list" rather than a troubleshooting experiment. @dencorso Feel like thread surgery? haha

There's no need to get defensive. I know that the root cause is unknown, I was just making a point that I believe the root cause to be the same regardless of 9x version. To my knowledge this has not been disproved, so it is still a valid conjecture. I also know that rloew considers the code safe or he would not have posted it here to begin with. I disagree though that the missing screen is a "minor inconvenience" - yes, it is a minor inconvenience in the fact that it is not important or necessary that it be displayed, but the point is that I believe the absence of it is a clue that the problem lies deeper.

A "universal to 9x" solution may never be found, but I have a feeling that the root cause of the Shutdown bug is the same regardless of whether the driver is running on 98SE or ME or whatever. The current solution works around the problem on 98SE, but not for ME, and 95/98FE are untested. I may be wrong but I think the bit I noted about the "Windows is shutting down" screen not getting displayed is relevant, because I always see this screen on a normal shutdown, even if only for a split second. The 98SE solution allows the machine to achieve a clean shutdown/power off, but I believe something is being "bypassed" to do so.