I doubt that Windows 9X SATA-Raid drivers are available for many of the newer motherboards.
I also found that in at least one motherboard, using SATA-Raid mode disabled my Adaptec SCSI Card from booting.
I agree that IDE Compatability mode is more complicated. I had to write a Patch and .INF file to make it work properly, but it works on my newest motherboards.
I'm not sure what VIA Driver you are using, but the 2003 Miniport Driver has a bug in it. You wouldn't see it since you are using a WD Drive.

It's my impression that there are win-98 drivers for the SATA controllers that were integrated into many motherboards up to at least 2006 or 2007. Perhaps moreso for SATA-1 vs SATA-2 controllers. The governing factor in win-98 compatibility for motherboards made after 2006 will be more for the chipset (north/south bridge) than anything else. If the SATA controller on any given motherboard does not have a win-98 driver, then chances are it won't also have a driver for north/south bridge anyways.

I still think the Asrock Dual or 4-core VSTA is the best motherboard for building a win-98 system (that's still available for retail sale).

Here are the details of the system I mentioned previously:

Hard drive: Western Digital WD5000KS (500 gb) SATA
SATA raid controller: VIA VT8237A Raid controller (viamraid.mpd, ios.vxd, viamvsd.vxd)

I'd still like to hear your opinion regarding win-98 and the max number of clusters-per-volume it can reliably work with.

I did some preliminary tests.

Windows 98 SCANDISK/DEFRAG is limited to less that ~4,000,000 Clusters (~128GiB for Standard Format)
Windows ME SCANDISK/DEFRAG is limited to less that ~32,000,000 Clusters (~1TiB for Standard Format)
DOS, and Windows 98 appear to handle the design limit of ~256,000,000 Clusters.
DOS SCANDISK can check and repair ~256,000,000 Cluster Drives but will not do a Surface Scan at 182,000,000 Clusters. The probable limit is 128,000,000.
There appears to be a cosmetic flaw in DOS SCANDISK above ~43,000,000 Clusters.

I've done some tests with SCANDKSW (the SCANDISK from Win ME) and reported here that it works OK with 26.4 million clusters (26,389,392 clusters), but throws the infamous "ScanDisk could not continue because your computer does not have enough available memory. If any other programs are running, quit one or more of them, and then try running ScanDisk again." message with 26.6 million clusters (26,588,648 clusters). You are the first, AFAIK, to determine the limits For the DOS SCANDISK. Thanks for the great info! BTW, from the way you wrote it it seems to me that the limit for SCANDSKW and DEFRAG are one and the same, what makes me think it must lie inside DISKMAINT.DLL. Is that right?

Why do some hard drive utility tools (such as Partition Magic) have a limit of about 6.3 million clusters when they prepare a FAT32 volume - regardless of the size of the volume or the cluster size ?

That number (6.3 million, or 6,291,204 to be exact) seems to be the upper limit as far as getting an instant result when performing a DIR command in DOS, and possibly for bringing up quick initial directory listing in win-98.

What do you know about this phenomena of dos or win-98 and the time it takes to perform their first directory listing as a function of the number of clusters on a volume? I think that the computation of free space is involved, but I don't know how or why.

I did some additional tests.

DOS SCANDISK will do a surface scan at 122,000,000.
A new cosmetic error appear during the Surface Scan when displaying Cluster numbers above 99,999,999.
The limitation on Surface scans does not appear to be related to available Extended Memory.



I suspect so. Defrag does a Disk Scan before starting to defrag.



If the Saved Free Space Count is -1, then DOS will recompute the Free Space. This requires reading an entire FAT, so it takes a while.
If the Saved Free Space Count is valid, DOS will use it and complete the Directory listing quickly.
Different Formatters may or may not set the Saved Free Space Count properly.

I don't see much significance in the 6,291,204 value other than it is only 252 less than 00600000 in Hexadecimal.

In my experience, if a system has booted into DOS, and if it has a FAT32 drive or volume with some arbitrarily large number of clusters (perhaps larger than 6.3 million clusters) then the first DIR command that is performed after bootup will not be instantaneous - but every dir command performed afterward will be (until the system is re-started).

I believe that I have also seen this behavior in win-98 - ie if win-98 is started and the drives are browsed, that it will take several minutes to view the file system if a volume with some arbitrarily large number of clusters is present - but this only happens the first time the drives are browsed.

What .mpd are you referring to? Is it this one:http://downloads.viaarena.com/drivers/RAID...alATA_V220E.zip? If so, can you describe the bug in more detail, please? Why does it depend on the HDD manufacturer to manifest itself?
And, yes, I do agree that most of today's boards don't have any SATA drivers. I think, BTW, that the chipset in wsxedcrfv's board is the very last VIA chipset that works with the .mpd I've just pointed to. Also BTW, does your SATA patch to esdi_506.pdr work with the LLXX patched .pdrs? It'd be nice if it did. You've mentioned before your SATA.INF alone suffices for the ECS GeForce6100SM-M2. Is it true for all nForce4 boards or just a special case?

This post has been edited by dencorso: Jul 5 2009, 10:45 PM

The .MPD file I tested was for IDE. It is named VIA_IDE_MPD_V320B.ZIP on the VIAARENA site.

Western Digital, like most Manufacturers did not correctly follow the ATA 48-Bit LBA Standard.
VIA made the same incorrect assumption when writing the Driver. It works for most brands.
Seagate followed the ATA Standard exactly, leading to an incompatability.
Depending on layout, there is a 1 in 8 chance that Windows will specify the last Sector in the 28-Bit LBA range when crossing the boundary.
This access will fail.

I became aware of this problem in 2004 and corrected the Code in Version 2.1 of my High Capacity Disk Patch.
It is documented in the User's Manual for the Full Version of the High Capacity Disk Patch.
I also Patched the VIA Miniport Driver to eliminate this Problem.



I believe my SATA Patch will work with the LLXX Patch, but I haven't tested it.
I only have one nForce4 motherboard so I can't say for sure. My other two Motherboards with integrated SATA needed the Patch.
Using a different version of my SATA.INF with my SATA Patch worked with the JMicron SATA Cards which have no Windows 9x Drivers.

This post has been edited by rloew: Jul 6 2009, 12:45 AM

Just to let everyone know the MSI PC54G2 wireless card works perfectly with 98SE. Though the software kinda sucks.

I don't think I mentioned this before but if anyone is looking for relatively recent laptops that support Windows 98 the Medion MIM 2080 does, and has working drivers for everything. Finding them is difficult and obviously the driver disks don't always have 98 compatible drivers. But, it is possible and I have done it on that machine. it also runs Windows 2000 very nicely. Since it has no internal CD drive, the setup files have to first be copied to the hard drive.