Oaft!... some crazy bus systems here, some I never even knew existed. Good to put that FX3000 bus to bed... had been looking for an answer to that one for a while now.

Having said, I've practically had my own learning experiences lately with proprietary bus systems with an EISA 386 NEC system I recently picked up. Not only does it have ISA/EISA... but it comes with an NEC proprietary RAM bus slot / connector thingy, and another for general I/O, and another extra long extended 8-bit ISA connector for (powering?) a storage I/O card. When opening her up, I couldn't believe all the bus action happening... all in the one box.

sweet, Gigabyte GA-486US rev3 has June 1992 datecodes.

I also found May 1992 "Swan 486/33 DB, The blinding speed of Swan Direct Bus" with S3 86c911. Apparently CPU and graphic chip were on a single, separate PCB daughterboard.
https://books.google.pl/books?id=jcrsPJb76HAC … irect+local+bus
in June they already offered Swan 486/50 DB running S3 86c924 at 50MHz!
https://books.google.pl/books?id=uummfdS7yF8C … local+bus+50mhz

PC Mag 27 Oct 1992 https://books.google.pl/books?id=IuaYd-eFaFoC … %20bus&pg=PA138
LB writeup reviews CSS Preferred 486/33GA with "EISA-like slot to connect the video adapter to the local bus" running 86C911 MaxGraphics/32. They also mentioned testing against Orchid "486 Superboard" system with "Fahrenheit 1280°D" suggesting those two arent the same despite using EISA connectors.
Dell also had a proprietary ET4000 based Local bus variant, apparently one of the first systems with Linear Frame Buffer addressing working, resulting in 2x Windows speed score vs other tested ET4000/S3 911 cards, but even ISA Mach32 was ~2x faster than that 😮, and prototype Mach32 VLB delivered fastest score recorded to date.

My absolutely favorite in that PCMag issue is a commercial for MATROX Impression 1024 card with strategically placed price sticker covering up left half the card to hide the fact its an ISA design, and not the touted VLB ;]

I think I remember reading that Orchid was one of the founding members of OPTi Local bus. That's probably what the Orchid board uses.

386_junkie wrote:

Having said, I've practically had my own learning experiences lately with proprietary bus systems with an EISA 386 NEC system I recently picked up. Not only does it have ISA/EISA... but it comes with an NEC proprietary RAM bus slot / connector thingy, and another for general I/O, and another extra long extended 8-bit ISA connector for (powering?) a storage I/O card. When opening her up, I couldn't believe all the bus action happening... all in the one box.

Just like some proprietary 386 Epson systems, this looks really NEC PC-98-ish. On the PC-98 and compatibles the motherboard is a multi-piece modular affair that uses the same kind of connectors for stuff like extra processor boards, memory expansion cards, etc. Even though third-party cards usually connect to the C-bus expansion ports (PC-98 ISA equivalent) and the system can also handle that kind of boards via that bus.

Possible to use EISA card in ISA slot and vice versa?

Uh, ISA cards can work in EISA slots...but not the other way around. I haven't tried it, but something would probably get smoked.
What I'm curious about is if OPTi local bus, and other local buses that also used the EISA slot could accept regular ISA cards.

Thanks! I didn't know that ISA bus speed is super low 😢

http://www.vcfed.org/forum/showthread.php?564 … d-transfer-rate

RiP wrote:

Thanks! I didn't know that ISA bus speed is super low 😢

http://www.vcfed.org/forum/showthread.php?564 … d-transfer-rate

Thank you for the link! It's a very interesting read. 😀

Yes, ISA is quite slow. Especially on Pentiums (ISA-PCI bridge). It's a bit less slow on 386/486 machines (~8Mhz ISA clock by default, extra ISA-CPU bus logic required), though.
ISA or rather "AT bus" performs best on a late 286, I think, that's what the bus was made for, after all (ISA signals are based on 286 bus signals). 😉

Some 16-Bit cards (VGA, ethernet, EMS boards etc) can handle up to 12MHz or 16MHz. Depending on the cards and age (check datasheet/manual first).
Before the term "ISA" and its spec were set in stone, some expensive "AT bus" systems ran between 12MHz and 25MHz (the latter with wait states likely).
So with a quality 286/386SX chipset -and tolerant ISA/AT bus cards-, th 16-Bit bus can provide twice the bandwith (under best conditions).

Will move the conversation about my recent purchase here. Any way of telling if this is an EISA or Orchid proprietary local bus card? It has Fahrenheit 1280 / D designation just like the one mentioned in the article above.

Is there a pinout of the proprietary bus system so that I can somehow identify EISA from the other one.

I can confirm what Thermalwrong stated in other thread. I check this with EISA ATI Mach32, ELSA Winner EISA, and OPTi Local Tseng ET4000.
If you would like the S3-911 EISA version instead of yours (i bet that is an OPTi Local one) im your man.

Thanks. Looks like the position of ground signals shows that my card is indeed not an EISA. Likely Opti Local Bus. Or was it Orchid local bus? Hard to find any conclusive name of the technology. Assume these two are the same thing...

I am proud of myself finding it out before putting this thing into an EISA board 😀

Now how likely is that I came across the Orchid Superboard to run this baby in? Was it a single board or were there several boards?

I think that Orchid Superboard has OPTi Local bus connectors. You can find that board if you buy that Poly 486-33L 😀

What about the 386SX boards ISA performs better since the address and data path is same as 286? (16bit datapath, 24bit address space).

Cheers,

On an otherwise ISA 486 with integrated graphics, is there a benchmark that can tease out what bus the graphics connect to the CPU over?

You don't need the Orchid motherboard to run this card. There are other 386 and 486 boards out there that have OLB slots too, and they don't necessarily need to have an OPTi chipset either.

This board doesn't have an OLB slot soldered...

...but I own one and it has the solder pads if you want to add it (you have to remove an ISA slot). Sometimes they are already soldered in. This is a fairly common board, and although this one is a ripoff sometimes you can get them pretty cheap.

For my experiments I purchased this Joindata 486 motherboard with a proprietary Joindata Localbus standard. And from a fellow Vogoner ?

+ Joindata localbus combi VGA + I/O card

Apparently this motherboard has not one but two local bus standards. I still wonder how is that even possible.

So in theory the top slot could take my Orchid local bus card I posted earlier. I still investigating if the board and the card are the same standard before getting enough courage to slot it in.

can you post picture of the motherboard back? the way ISA slots are all connected together doesnt leave space for processor bus tracks, must be on the back going to those 245s
similar one on wikipedia, looks like they quickly gave up on this slot idea
https://upload.wikimedia.org/wikipedia/common … oard_at_CHM.jpg

Quickly traced the bus with a multimeter.

Looks like the upper row (A-D) are ISA signals. So whatever slot it is, it looks like it is ISA compatible just like EISA is.

The lower rows (E-H) signal seem to be going mostly towards the 82C493 chip (system controller) rather than the CPU.

Will try to find out a better wiring. This could also be an EISA as it seems to be that the E1 pin isn't grounded as per the post above...

Could the top slot be used for a "memory expansion card"? There are some boards from the early 90s that use the "proprietary" slots for this reason, and this one is very close to the SIMM slots.

Yes it is likely. I traced signals and it looks like at least on the main part of the "EISA" connector there are complete set of MD signals some RAS strobes and RAM even data parity signals. So it looks like this is a combination of ISA slot and memory expansion of some sort. It is a silly design as if you populate the lower SIMM slot, ISA cards with taller chips close to the connector won't fit:

So my Orchid card is still without home. At least I can play with the Joindata propietary local bus...