the original batman board is quite rare, i don't even recall it being discussed here beyond people mistaking batman's revenge for it. if it really turns out that these boards can't run any PCI IDE controller that'd be an interesting discovery. you could try some SCSI hardware if you have it, though i should mention that on my batman's revenge board apparently the BIOS was overwritten by the adaptec SCSI BIOS because the write protect jumper was off, bricking the board and requiring a reflash. i'm not sure if that applies to batman as well.

auron wrote on 2023-05-25, 22:04:

the original batman board is quite rare, i don't even recall it being discussed here beyond people mistaking batman's revenge for it. if it really turns out that these boards can't run any PCI IDE controller that'd be an interesting discovery. you could try some SCSI hardware if you have it, though i should mention that on my batman's revenge board apparently the BIOS was overwritten by the adaptec SCSI BIOS because the write protect jumper was off, bricking the board and requiring a reflash. i'm not sure if that applies to batman as well.

That's an interesting perspective. The computer did come with a SCSI controller which I have removed due to the desire to switch to flash storage. With mechanical drives starting to die off at a rapid rate and lacking a SCSI to SD adaptor, I wanted to take the IDE to CF route.

Looking at the motherboard, I am wondering if PCI IDE was originally envisaged and temporarily abandoned in order to get Pentium out of the door for 1993. In November\December '93 this would have been a high end system and the 2GB double height 5.25" SCSI hard drive would have been one hell of a pricey option.

Round by me, most SCSI 2 SD adaptors are designed for the Amiga and I cannot find definitive answers that they would be compatible with the PC. Added to this, I had hoped to add other IDE devices such as a ZIP drive and CD-Writer. My hope with the Tekram was that the ISA connector board would allow the system to communicate with the hard drives, but again, nothing that I have done, no combination of slot or ancillary devices seems to work. If it does transpire that the board simply will not work with PCI IDE then as you said, that provides a lot of interesting hostorical context although it does not help me persoanlly. In the meantime, I think I will add an ISA IDE controller and then see about acquiring a SCSI 2 SD adaptor later.

just to be sure, did you check with PCI.EXE if the IDE controllers properly get an IRQ assigned? these boards predate plug'n'play and could be a bit finnicky at times if you have multiple cards installed.

in any case though, in 93/94 you really had to go SCSI if you wanted optimal performance. batman had an option for an onboard NCR 53C810 PCI SCSI controller so i'm not sure they were thinking about PCI IDE in 1993 quite yet. the batman's revenge and plato boards did come with the RZ1000 PCI IDE controller but it's not busmastering and also had to be gimped later by disabling prefetching. to my knowledge there just wasn't an IDE solution before 430FX/PIIX that could properly offload the CPU like a good SCSI controller does, and an adaptec 2940 series will give you busmastering even in DOS without having to use a driver there.

by the way, according to the manual, even the early batman's revenge revision lacked the onboard RZ1000, relying on a paddle card instead:

Premiere/PCI Baby-AT Technical Product Summary · Page 30 wrote:

IDE PCI IDE was not integrated on the original Premiere/PCI product. Instead, a PCI IDE paddle card was used in PCI slot 3 (J10C […]
Show full quote

IDE
PCI IDE was not integrated on the original Premiere/PCI product. Instead, a PCI IDE paddle card was
used in PCI slot 3 (J10C2, the PCI slot closest to the 3.3V connector) and made use of some of the
reserved pins on the PCI interface, as listed in the following table:

I figured that it would be a good idea to follow-up on my previous posts with an update on the machine. I'm still not 100% there, but a lot closer.

I did indeed verify that there were no IRQ conflicts and came to the conclusion that as this machine has date codes for November 1993, I likely have a very early version of the Mercury chipset that is buggy as anything. Attached is an article from October 1993 detailing how Gateway were one of the few manufacturers that were shipping Pentium systems with a PCI bus the following month, with other manufacturers holding off due to the litany of issues with the PCI chipset. Experience is the better part of valour, and as such I have given up on my quest to install a superior IDE controller card (it is almost certainly the reason Gateway stuck to a single ISA based IDE channel).

The system did come to me with an Adaptec 2940 and a simply enormous 2GB hard drive from 1995. I presume this upgrade was installed during that year, but I did not fancy using a noisy, double height 5.25" hard drive that might fail at any moment. So I gambled a little and bought a ZuluSCSI being marketed for Amigas' at a reasonable price. The gamble has certainly paid off, with a little trial and error resulting in a 16GB SDCard acting as 3 hard drives within the machine (SYSTEM, DOCUMENTS & SETUP).

So why am I not 100% there?

Performance is fine but I have two problems - one being a rather nice problem to have and the other needing a degree of patience to resolve.

The nice problem is what soundcard to use. This is a case of having plenty of excellent choices rather than any issues. Although this is not intended to be a gaming computer, I certainly want the option and to have good sound when I do. I can use a Sound Blaster AWE64 Gold, an unbranded CMI8728 card with Dream MIDI chip onboard, I have a MK8330 with e-wave MIDI add-on on its way from Greece as we speak, and I can also do a little sound card shuffle, so potentially switch to a Sound Blaster 16 or a number of other cards with OPL3.

I have tried the AWE64 previously but I think it is a bit of a heavy option for the system. The CMI8728 with Dream is a lovely card, but with no official Windows 95 drivers and issues with the ones that it does use, I would prefer to use this in a Windows 3.x machine. I am building a 386DX machine very soon and I reckon it would be perfect in there. The MK8330 is a harder choice. I could put it in my 486 which I only pull out from time to time when I get the chance, or install it in this which I will use quite frequently because it is in my KVM stack. Buuuut.... the 486 is a dedicated DOS gaming computer and this is a legacy photo extraction and editing computer with occasional gaming chops. Frankly, having this sort of problem is a delight, as it means I have a lot of choice and truly I am spoiled.

The other issue is networking. I have a dedicated VLAN for my legacy network with network shares available to access program installers, drivers etc. Unfortunately when transferring large files, the computer often freezes. I strongly suspect a software issue, with some indications linking the unofficial service pack 1 as a culprit. I have switched from PCI 10/100 network cards to 10Mb and currently ISA 10Mb network cards, but all have been 3Com, so it could also be a driver issue. I am confident that somewhere I will find a solution, I'm just not there yet.

I will post a further update a little later on, but to conclude, storage issues appear to be resolved through the power of giving up and spending more money. I have more work to do on the computer, but hope very soon to have a functioning early model Pentium to which I can connect my early 90's digital cameras and occasionally play the odd DOS game.

Side note: I would have thought these early network cards don't use FPU calculations, but could the patches mean that they start to? My CPU definitely has the FDIV bug and that would explain the lock-ups.

indeed, the SIO southbridge chip saw a change from 82378IB (also used in the 486 420TX chipset) to 82378ZB. don't think i've been aware of this before but that might be the reason why the batman's revenge board was made in the first place. looks like the ZB revision is what added PCI IRQ steering: PCI IRQ steering on 82378IB (original SIO)...

the only other change i see is this sole mention of the 82378IB in the 82378ZB datasheet:

Special Cycles are considered invalid by the 82378IB and are completely ignored. The 82378ZB responds to a Stop Grant Special Cycle.

i'll add the datasheets in case anybody wants to go through them, can't put up the 82378IB datasheet for file size reasons but it's here: https://datasheet.datasheetarchive.com/origin … SC000149496.pdf

still, shouldn't the batman board have jumpers to deal with PCI IRQs? regarding sound cards, in my book the AWE64 would be very adequate if you treat it as an AWE32, which would have been a contemporary upgrade option for a machine like this in 1994/1995.