Goodmaneuver

My NAS formats the drive on placement but yours may be different. Even if possible the partition would have to be at the front of the drive. It is the 28 bit usage within FAT32 & thus inbuilt into the file management system of Win98 that gives us the address limitations. I do not think it will make any difference using different protocols but depends on if it uses a browser to download files but I do not know. Most of the time I install programs straight from the network drive. As time moved on increasing hard drive sizes were able to be detected in motherboards BIOS. A motherboard BIOS update sometimes upped the size limit. Seagate explained very precisely about 137GB limitation with solutions for different operating systems. Search for 137GB.pdf, it explains how the SATA adapter coincides with ATA-6 as all SATA interfaces at the time of introduction 2003 and so is compatible with 48 bit.

When using the NAS as a network drive with Win98, the total NAS disk size would have to be less than 2.2TB. You could use 2 non-raided 2TB drives and then map 2 separate network drives. Mapping sections of a larger drive could involve sectors/blocks that could be outside the range of the 98 driver as the drive fills up. The 2TB drive size limitation & 4.29GB file limitations will still be applicable & is what I encountered with my WinME system.

DirectX install problems, try WinXP KernelEx settings. See here for some ideas for AC97 audio, here What I will add to the next comment by me in that same topic is that the newer files will have to be removed when trying older driver as installer will not over-write newer files from my memory of such. (safe mode should help here)

Some Win98SE users are stating they use it vanilla like Siria, it is this that makes the difference and those users need to be informed of the updates as you have admitted. Believe what you want about the 32GB limit on USB with WinME unaltered. RAID controllers operate with their own BIOS in the majority and no matter what you try and do the limit was 470GB. There were triple channel RAID controllers like 3SI each channel could only be configured up to a maximum of about 470GB. I have used RAID controllers of this era. There is a hardware interface change between IDE to Sata and as I said it works. FAT32 can be formatted to 16TB I think if 64k clusters are used. It is not as simple to say "There is no such limit in the FAT32 filesystem" It was up to the operating system designers. (file system control in clusters). Ifsmgr.vxd is the file manager AFAIK. Since I have mention the IDE to SATA controller adapter. They seem to accelerate the IDE setting speed and if your IDE is a bit suspect then it won't work. Do not get the bidirectional controllers as both directions can not be used at the same time and they do not work as good (may not work). The old Socket 7 board I used I can not find at the moment had a SIS north bridge & had SIS internal video. As I tried to explain by strong; this means a Quantum Bigfoot hard drive has to work on the IDE.

NUSB driver update is not for WinME though and will not work. A 32GB USB stick was not available in 2000 AFAIK and it uses different accessing to that of IDE as Jumper has said. I tested USB sticks and the highest size that works with vanilla WinME is 32GB. Also tested external USB to IDE on drive and yes the same thing 32GB is the limit there. I did try vanilla 98 at one time and it did not work the USB sticks without the NUSB driver update. I have purchased several IDE to Sata adapters and have used it on old Socket7 machine. Part of their advertising was that it overcame the IDE size limit problem. This makes sense as the Sata driver would need to access drives of the current time of manufacture. I used it on Socket7 mother board where access timings and switching strength were strong & slow enough to work a failing Sata drive. The other way is to use USB to IDE on a strong USB 1 port. At around 1998 - 1999 SCSI raid controllers had a maximum 470GB raid size. It was large for that time. I assume vanilla WinME had this limit as well as when I viewed the 500GB Sata drive with Partition manager it did not display the partition quite right. It was the first time I saw an error with size on IDE.

Kex is short for KernelEx Jumper is working on updating its functioning. USP3 is unofficial service pack 3 for Win98SE. Your mother board is modern and would have 48 bit hard drive access from the initial boot so that large drive sizes are supported at boot time. Otherwise, an old machine specification can be looked up to see maximum drive size that can be used. An IDE to Sata adapter overcame these obstacles from my memory of such up to the OSs limit of 470GB at the time (my ME system still is AFAIK for IDE). I do not know exactly what is or what is not compatible about the USP3. The files contained within USP3 are supposed to be compatible with Win98SE without KEX. It is for Jumper to answer this part. The 4th partition on the mobile card, perhaps they were pushing the size of card & possibly were not FAT32 partitions. Some WWW comments say 4th partition should be extended but I do not know.

I would assume you would need NUSB33e.exe (not sure if latest) installed to view the USB drive. This version is for Win98SE only, the following applies to the NUSB33e driver. When buying a new HDD the GUID Partition Table will have to be changed to Master Boot Record. The size limitation for MBR is 2.2TB per partition. Up to 4 primary partitions can be made or use extended partition if more are required. The large USB drives usually have a USB to SATA Adapter inside the box and Logical Block Addressing would be used for the raw sector accessing. They should be formatted on the OS you intend to use them on. A WinME/98 formatted disk will have a 9MB free area at the end of the drive when viewed with Disk Management on modern OS. It is not free really and if you partition it on modern OS without the MB shown free at the end, it will take longer to open the drive on WinME/98. If more than one computer is required to access the drive then a network drive could be considered for convenience. See here https://msfn.org/board/topic/178190-windows-98-and-nas-network-attached-storage/?do=findComment&comment=1160247 Why if used for backup only, the need for one drive? I think your asking how will Win98 go without patching. WinME without patching could access a maximum size of 32GB on USB so any partition larger than this can not be used. I have USB to IDE adapters and large size partition access was fixed with the updated files. I am not well informed about 98.

Wikipedia works but I have had trouble with standard 12.02. Found this does not work without Opera 12.18 files update as well. https://xem.github.io/minix86/manual/intel-x86-and-64-manual-vol1/o_7281d5ea06a5b67a-197.html This comment is not true it was because TLS2 and TLS1.1 was not enabled.

>>That's all well and good, but how is this helpful to the OP? I do not think I explained this good enough. The updated files are so that network drives up to 2TB are reported correctly. The 2 files I think that are required are Nwlink.vxd and Vredir.vxd but experiments can be done as those I mentioned on page 5 are the only VXDs I have updated. Just to be clear, I am running with the original ESDI_506.pdr for WinME which may mean a restriction in IDE and SCSI size but no problems so far, largest IDE I have is 200GB. I have had to scan disk my 300GB SCSI drive and had no problems about doing it. Here is a link that talks about ESDI_506.pdr for Win98 https://msfn.org/board/topic/78592-enable48bitlba-break-the-137gb-barrier/page/2/ ESDI_506 may be for removable discs. I may not be using ESDI_506.pdr as I use a VIA IDE driver version 2003 vs 2.0.950.3020 which uses its own mini port driver ViaDsk.mpd.

It has been a while but I have labeled some nVidia drivers that work or max version that supports certain cards. Test with Dungeon Siege 1 for correct displaying. I am talking Vista or Win7 but could be same with XP. 175.19 FX5200, 81.95 Xp Quadro NVS, 91.36 Xp Quadro4. I would install 162.50 and check with DS if OK. I would not go any higher in driver vs unless it works flawlessly. I would not go any higher than about vs 280 though. Now you have to install a standard graphics card driver first. Then you have to remove the nVidia files from the system and control panel directory otherwise they will not be overwritten and control panel will not work with older driver files and if driver files not replaced then they do not work as intended when going back in driver version. Hope this helps. The AMD 3450 should be OK but be weary of too moderner driver. Try drivers that are more relevant to the release of the card. It probably is RAM settings or too higher link transfer speed (have to find correct wording). The auto settings push the limits too far and can cause hardware failure eventually. Check for hard faults in RAM. What I do not like in particular is that Win10 deliberately decides to not access so called inoperative RAM so that when we fix the RAM by slowing down access times we can still loose our complete operating system if we are not aware of what is happening. I have to get all my data off my drive now as I did not know enough after an update. I think this was the web site I used to fix it https://borncity.com/win/2017/10/06/windows-10-update-error-0xc0000017-no-ramdisk-memory/ But I did a system restore after the update went wrong and hence it would not boot after I fixed the error. If choose to do a system rebuild as setup disc option, it says it keeps personal files but the rest of the disc is wiped. (I have just done the reinstall with selecting to keep personal files & I am pleased to say that it did not wipe my programs as it said it would & all is back to normal again.) The first paragraph remains correct for GeForce 6 series cards. I have not had any trouble with 512MB nVidia graphics cards with WinME, no patches. (Not sure though, if included in MESP1 I could have the patch). I have a 7300 512MB and a 7600 512MB card. I have virtual memory settings of 500MB. Alias a Win98 game will tell you if there is not enough graphics RAM for the settings chosen. It is not hard to go over 128MB. Settlers 6 and Glory of the Roman Empire must use a bit. If not enough virtual memory is employed, Glory will freeze while the info is calculated on the hard drive. >>But i tried Passmarks Memtest86 and its running fine. All tests will need to be done and will take a long time. The OSs use different modes of driving memory and can still get errors. I had to boot up and then lock settings of RAM to manual as best I could in BIOS settings then lower the access timing as low as possible on the Win10 machine to get reliable performance; 1333 down to min 400. >>multithread test by F2 Make sure multi threading is enabled in BIOS as it seems default is disabled. I did not like the Aureal Vortex driver because it was not compatible with something which I do not remember. It is one of the few sound cards that I have that work though and was difficult to get one that was 64 bit PCI slot compatible. I assume you have tried without this card plugged in. I do not own any LGA1151 boards so I am not up to speed on the modern stuff.

Catalyst 4.11 driver was the first for Xxxx cards but I did not think it was a Betta driver, it was distributed with the graphics cards. The link I have given even though it says it is 4.14.01.9150 it is an earlier driver that does not contain any Xxxx series card support, it is the previous driver. I have found an Omega driver from MajorGeeks.com that is for Win9x that contains Radeon mobility FireGL 9000 :- rad_w9x_omega_2637.exe 2nd month 2004 so no x700 series. I think Cov3rt has talked about this elsewhere though it may have been for nVidia, we probably need to talk with them some more.

I have WinME operating system and 98 drivers are not compatible. Yes files greater than 4.3 GB can not be copied. It is a lot less work than reloading a fresh copy of Win98 and all of your programs. Of course no need for KEX for watching DVDs but I needed to discriminate between my builds of the OS. I have used my 2TB USB drive for games which was NTFS. Paragon NTFS for 98 is free and does not need KEX. I am just trying to help with the previous discussions. I do 1:1 copy first because if trying to resize all in one go will result in errors with WinME and any on-line drive copy is asking for trouble which can loose swap file support. I do not do drive copies anymore because the dates of files can change. Paragon Partition Manager 5.6 or just 6 is what it is sometimes called, is more suited to to 9x OS and reports more accurately. For example if you drive copy directly to a NTFS partition, the DOS 8 drive system files (sys c: from command) are located outside the partition most likely as were the NTFS ones. When viewing with 5.6 it does not show any free space but viewing with Disk Manager 8 it showed a bit of free space at the end of the drive. If this partition is resized to make use of this extra size the partition will no longer be bootable. BTW if drive copying over an NTFS partition with DISK Manager 8 then it has to be done twice to be successful. Some of the NTFS sys files must still remain as FarCry 1 full version worked this way where as only demo would work if drive sys c: was within the partition?

I have these updated VXDs only; Ifsmgr; Nwlink; Vnetbios; Vredir; Vserver & Cdvsd. VXDs are not compatible between 98 and ME. I run Paragon NTFS for 98 and a 200GB IDE drive having XP on it is displayed and writable correctly. I have SCSI with 300GB drive Fat32 for games and have 2TB network drive all writable. The USB will need the SYS files updated. These are obtained from the SP for your OS UHCD.sys USBSTOR.sys USBHUB.sys USBPRINT.sys USBDB.sys & USBCCGP.sys, mine 4.90.3000.1, and from updated USB drivers from the manufacturer they should have a suffix of 11 from VIA the latest. You can copy over updated VXD and SYS files while off line or on line which will be effective next boot. Run the update for 48 bit drive access while on line. Then can do a 1:1 sector copy of your drive off line. Then boot up new copy to get MBR updated, do a disk check but it should be OK because it was a 1:1. Then with the drive off line, resize it. When running on line first time do a disc check to update the MBR and show new disc free space. My KEX 4.5.1 build stopped playing a DVD after about a 20 minute interval with PowerDVD6 and KMP, later KEX builds had no problems with IDE DVD access. The USB DVD did not stop. I had updated Cdvsd.vxd but did not make any difference & have fixed this now, I had a 98SE update for ifmgr.vxd instead of 4.90.3000.*. The 98 driver also slowed down access times. Still has problem plays for about 30 minutes.

After I mentioned that the override settings of individual modules did not work I thought the next KEX release did the opposite characteristic. Override individual settings does not seem necessary if the modules are set properly anyway and would cause problems. PowerDVD7 can still pick up external filters for example. It has been working well with BASES setting. I have filters working and they are set up with their KEX mode setting. When I turn WMP KEX disabled it does not load LAV splitter and error confirms. It does not matter what filter it is as long as it needs a unicode setting to work, it will not load. This indicates that the override settings of individual modules happens when KEX is set disabled on the executable. I have an old build set up with 4.5.1 and I have Will Rock and Crime Scene Manhattan working set KEX disabled. This was done to test for individual modules that need KEX. Some old games like Will Rock select system codecs and if all filters are modern they will mostly not work disabled. There is only one file in Ancient Wars Sparta that needs KEX set to 98SE or similar non unicode setting within the game modules; VFS.dll, and Sparta can be set disabled. Remember this is when experimenting with 4.5.1. The thing is WMP tries to load filters when disabled but can't. Other programs are similarly effected that can operate KEX disabled eg MPC 4.9.0. The KEX settings are in a special order in RAM, let me know if this happens for other testers or does not? If a settings check for duplicates is implemented can a invalid setting check be done like typo error. I have made mistake by typing NT6 instead of NT40 for example. But vigilance is all that is required no need for the extra sheet function.

It has always been the case with MSI files. It would be nice to have this feature but internally written in the MSI are the install instructions and internal files can be error checked. It depends on how the MSI install instructions checks the OS version, a temporary registry change may work. SuperOrca from Pantaray can look inside the installer and the install instructions could be altered. If a CAB file needs editing for example is difficult to edit the CRC check or know how to. If the program install is from an EXE that extracts the MSI then when the MSI has to be edited after extraction but many MSIs have reliance on an external executable to start and if this is the case then there will probably be CRC checks on the MSI itself. Why not install to a different OS and get the files and reg settings, if needed, from the other OS and see if the program works. Settlers 6 works this way and Acrobat 9, as examples only, must know also any new system files that are made. But as always, I do not want to get in the way of Jumper or other members answering your question.

The 12.18 files do not serve any added feature over 12.02 unfortunately. I found for a basic build Opera 12.02 may be better with KernelEx 4.5.1. My idea, 4.5.110 commonly known as 4.5.1 worked well with FF3.6.28 but on later versions did not transfer any data. 4.5.2 or 4.5.120 full name worked up to FF9 just. The freezing will be the TCP/IP not transferring, by turning off JavaScript it makes the pages less huge to render and could be coincidental to not freezing. The glyphs do not always show with latter FF, it all adds up to not enough resources thrown into 4.5.2 to make them work IMO. I have fooled around with buffer sizes from within the omni.ja and it is not only the transfer freezing it all has to integrate with the OS properly. YouTube would play for up to 4 minutes approximately before freezing. If on auto-play then it was OK if video changed within the 4 minutes otherwise it froze. Once a buffer is near full a new allocation would have to be made with the video play back and eventually the original buffer will need to be deleted. Or a couple of FIFO buffers might be OK, one for current video another for new video then back to original when next video is selected. Presto 2 engine was brilliant for its time, YouTube still plays but not all videos now, and is still a good email client. FF caterers for many platforms and therefore has a lot of overheads that are not used. I have upped the size of the KernelEx stack size in CurrentControlSet/Control/MprServices, it seemed to be better with CometBird.

Shifted off topic to Ruthan's here

Blender 2.64 is working using Python 3.2 and KEX24.

>> I tried SciTech Display Doctor but the drivers never wanted take over the microsoft driver so i wasnt able at all to use it. The driver file may have been in use and in RAM. If you know the name of the file/s, I think it was OpenGL32.dll, but not sure now, copy SciTech one over in safe mode or from DOS on a startup floppy. If it is not OpenGL32.dll rename OpenGL32.dll to disable it, in safe mode if necessary. If the Microsoft file is 16bit then move it by left clicking on it and drag to an empty folder after a fresh reboot. 16 bit files can be deleted by deleting the folder they are in or they need to be highlighted then press delete key if right clicking cannot delete or rename. >> - Right-click My Computer -> Properties -> Device Manager tab and remove all entries under Display Adapters. This is done in safe mode otherwise unused devices do not show. Sound driver removal like this can destroy a good system if drivers like streaming devices are removed. These are usually named Microsoft but are not always in use so if using HPSysDig.exe then these devices show as unused but they are just not in use at the moment and should not be removed as they do not come back I found.

Did XP show the 2 drivers? If it does then plug in external monitor to exercise that area of hardware. See what the driver is from XP. Try and fill up your hard drive if you only have one, in case it starts to get write errors. @cov3rt might have used an earlier driver as Catalyst 4.11 has already an inf for x700, it just doesn't add up. I have found some DNA drivers https://www.guru3d.com/files-archive found in September 2004. * Added support for: -:RADEON X600 SERIES -:RADEON X800 XT -:RADEON MOBILITY 9800 -:RADEON MOBILITY X300 -:RADEON MOBILITY X600 -:RADEON MOBILITY M28 (Mx8) -:ATI FireGL D1100 -:ATI FireGL V5100 -:ATI FireGL V7100 -:ATI FireGL V7200 I tried this driver and it looks like DNA drivers did not support Win98 no matter how old they are. It has not got M26 either, it has got M22, M24 & M28. Driver version 4.14.01.9150 date 20/10/2004 for Win98 here https://www.helpjet.net/Fs-64531818-59851182-67091790-devices.html It does not have x*** series cards.

Means full version name for 4.5.1 and 4.5.2 for clarity but may have just been confusing. Python 3.2 seems to be OK now. Tested with Blender 2.64a, Python 3.3 must almost work. Later 2.6 Blender versions use Python 3.3. I have just changed my CPU, this may lower the sensitivity to the video blackout problem? (4600+now Opteron 152) My mother board ECS K8T890-A has an AGP port/slot and a PCIe port/slot, the PCI AGP port pushes out frame rates a lot faster than motherboards with normal AGP I found. Even ATIO9XXX.dll can have an error in VLC3 on exiting. libqt_plugin.dll is definitely working now. VLC2 needed a minimum of 1GHz CPU to start up. VLC is progressing though, 3.06/8 has got a default CPU affinity setting of -1 for WebGL. There are a few options, disable opengl output module within the player if necessary or select WGL extension for OpenGL, then the CPU index can be adjusted. A positive CPU index seems to be better when changing videos within the player but worse when exiting. I think my graphics cards may get better. More about Riched20 it works on any setting incorporating Unicode and showed no sign of differing operation in WinWord, it only shows some odd glyphs when opening Wordpad, this is why I needed NT6 setting. It might be a timing issue. Riched20 14.0.7155.5000 can be obtained from BWC's Archive if have not got Office 10.

Yes updating inf would be good if driver worked. Forcing the install is all that is required to test the driver. As you know, update driver, and choose manually the driver to try from within the install list. I think that some laptops used two different GPUs, one for the screen and the other for the VGA output. One was in chip IGP and the other was external graphics like the ATI driver. This would show up in Device Manager I would expect though as two separate display adapters. I wonder if plugging in an external monitor would switch off the laptop screen? Have monitor plugged in before powering up the laptop to see. If it does then this is worth a try to use the VGA output and try the driver.

If VKrnlEx.vxd is linked to a driver then perhaps it will startup earlier in the boot sequence and therefore load KEX before network login & perhaps load in safe mode as well? pshExports what would this mean as a sheet option, does it mean all imports calling a particular module be directed to a different library similar to KnownDlls but be able to act on a module by module basis? VLC3 with KEX23 used standard windows OpengGL and worked. KEX24 was working as I proclaimed, (I may have had a FX3000 installed at that time), but now I get a black screen when trying to play videos and it will not terminate. This build is using NVopenGL when video is loaded in VLC. (GeForce 7900GS). The one build that is working with KEX24 is using windows standard Opengl in VLC with the same hardware setup. NVopenGL is running with the screen savers on this build. I will update progress with this here soon.

Both Sheet Todo & auto module testing for mode selection are great ideas. >>4.5.2016 API support will then be slimmed down to XP and below for stability. (Kexstubs will still be supported.) I would like to see Vista mode still available because MXMUTE.DLL from Maxthon has no implicit load library errors and Vista mode is required. Some in between modes are probably not required but Win2K_MT is required for a NotePad++ plugin. When running 4.5.110, Opera 12.02 & FF3.6 work but CometBird9 does not transfer any TCP/IP data. I was thinking, the change made to 4.5.120 to make FF8 work is it large enough for later version FF browsers? That is, the size of a buffer, what ever the change was. HTTPS pages in particularly made the browsers stall. While not important, making Explorer KEX disabled prevented rundll32.dll loading NVMCTRAY.DLL and an error notifies this but after placing explorer back to DCFG1 or other setting it still does not start up after a reboot if 2 reboots have past while explorer was disabled. It takes a reinstall of the driver to reinstate Media Center and control panel Daemon. Also the Daemon and Media Center do not start up after runtime (while explorer is running) if terminated, a reboot is required and has always been the case but reinstallation to reinstate the media center tray was not required in 4.5.110. >>Possible additions to the driver include rloew’s i686 + my finesse sse exception handling, and Wdmex code. I know very little of these & would like to know more about them.

http://www.opendrivers.com/download/driver-18305.html This link is to the 4.11 Catalyst download page, the direct link must time out, choose the backup server link towards the bottom of page. When you said it may not be good then this driver is OK. I checked the files to that of my Catalyst 4.11 disc and they are the same.