I have a CF adapter that has both master & slave so I populated master with an unformatted 4GB CF and the slave with a 16GB. Using the network card I redirected boot so it was 81h instead of 80h which made the slave boot fine

I broke down and made an XT-IDE EPROM for my network card (it isn't easy for me since I don't have a proper EPROM burner; just a Realtek network card that can write to EPROMs).

I don't have many ISA cards for use with my Gateway 2000, so I have space for a network card. Universal XT-IDE seems to work well on the system, with no hang-ups (so far) or boot failures from pre-formatted disks and floppies.

I am having some trouble making and booting from just created partitions on new SD cards, so I think I have some more troubleshooting to do.

I'm a bit disappointed because the user flash method would feel more integrated than the network card solution. Oh well.

Here give this a shot. Turns out I needed some extra modules. NOTE: THIS IS ONLY CONFIGURED FOR CF SUPPORT! THERE IS NO POWER MANAGEMENT AT ALL! To fit power management would require slimming some strings down as there is a space constraint of 69 (nice) bytes.

1DEFINES_CUSTOM =  MODULE_STRINGS_COMPRESSED USE_AT USE_286 USE_386 MODULE_8BIT_IDE NO_ATAID_VALIDATION EXTRA_LOOP_UNROLLING_SMALL

I didn't notice the write issue because smartdrv was caching my data. This seems to perform properly on my machine.

sammargh, much appreciated! Your new version of the User Flash seems to behave very much like the full Universal BIOS with my SD cards.

Quick question for you. Have you tried to install an NT-based OS on your Classic R with XT-IDE? I'm getting an Inaccessible Boot Device error on both NT 3.51 and NT 4.0.

Thanks.

I've done some testing, and it seems that the Universal BIOS is about 5 - 10 time faster than your latest User Flash version when I use the motherboards Primary IDE (device set to Master). I wonder if that's because I configured the Universal BIOS for a 32-bit controller. Would the 32-bit controller code fit within the User Flash space?

Also, the User Flash version only allowed me to create an 8GB partition on my 16GB SD card whereas the Universal BIOS allowed me to create a full 16GB partition. I wonder what accounts for that difference?

None of the fixes were put in place for Windows 95 partitioning in module selection so this bios will perform very poorly in Windows 95. With the VGA chipset limiting the board to sub-16MB ram it really isn't a suitable candidate for Windows 95 and is best left to MS-DOS or Windows 3.11.

I'm using 32 MB of RAM on my board. Windows 95 seems pretty responsive (I'm using a POD83 as the processor). Shoot, I've even used NT4.0 on this board (with a drive overlay) which actually seems like a pretty good fit all things considered for this vintage.

I'd like to continue using NT as well since I plan to swap out SD cards on a per operating system basis. NT's ATAPI driver doesn't seem to like XT-IDE on this board.

Well, a fair warning that the VGA chipset Cirrus Logic GD5428 has a known issue where systems that have more than 12MB of ram lose linear frame buffer support in VESA. If you're only using Windows this wouldn't be a major issue, but is with MS-DOS games.

I can honestly say I've never noticed the VESA LFB issue, but it doesn't surprise me (or bother me). I'm one of those weird people who likes to mess with computer parts far more than actually using them.

In this case, I'm trying to max out a Gateway 2000 low profile computer that reminds me of my first new PC which was a Gateway 2000 386SX. This system currently has a POD83, 32MB of RAM, a 10BaseT network card that can host boot EEPROMs, a wavetable sound card with on-board EIDE controller, and a 12x ATAPI CD-ROM drive.

I'd like to use some large SD cards as boot drives, so I'm looking for good solutions. While I can use, and have used, overlay software, it isn't easy or convenient to modify these drives on other computers once the overlay software partitions/formats the drive. I had hoped that XT-IDE would solve the drive geometry problem across the board. It seems to work for well for DOS and Windows 95, but NT's failure is a bit of a bummer. I'm going to test OS/2 next.

I did more testing, and OS/2 Warp works with XT-IDE on this motherboard using both IBM's and Dani's IDE/ATAPI drivers. I'm still surprised NT's drivers fail.

I'd still rather use a tiny version of XT-IDE in the user flash space than the large version on my network card given that I might need to disable the add-on BIOS to use NT. I'll post any progress I make on a custom build once I get to it.

I'm thinking about excluding the 8-bit drive access modules to make room for the 386 instructions and advanced ATA modules for the primary IDE port on the motherboard.

BTW, is the GD5428 LFB bug hardware or software related? Would artificially limiting the maximum RAM to say 12 MB in the DOS XMS driver solve the problem?

I figured out the issue with installing NT using XTIDE on this motherboard. It turns out I needed to have a dummy drive listed as installed on the controller in the motherboard's BIOS (I picked "Type 30"). That allows NT to access the drive using the BIOS interface vice the ATAPI interface. XTIDE still takes over and provides access to more drive space, but the NT AT disk driver has a hard 8GB limit for partitioning. Oh well, that's plenty of space for a 486-class computer.

I also messed around a bit with building a tiny version of XTIDE that could fit in the user flash space. Unfortunately, the Advanced ATA module I'd want to accelerate the VLB controller is just a bit too big. The smallest I could get the bin file was about 4.2k. I might need to stick to the network card as the host for XTIDE at this point.