VOGONS


My Ultimate VL/EISA 486

Topic actions

Reply 80 of 115, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Cool, it's good to see another SuperEISA in action.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 81 of 115, by bbuchholtz

User metadata
Rank Newbie
Rank
Newbie

So, I've finally overcome my learning curve with EISA. I've learned the ways of ECU 😀

I have a question about bus mastering. Would you happen to know how many slots on this board support bus mastering? From my perspective, I'm getting conflicting information around slot #8.

In all of the related ECU CFG files, it would seem that only slots #1 - 6 support bus mastering:

; Configuration file for MCCI EISA system board
; Revision: 1.0
;
; 10/30/92
; Modified: (Changed ID to AMI68B1)

BOARD
ID = "MCC0001"
NAME = "MCCI-486 EISA System Board"
MFR = "MCCI"
CATEGORY = "SYS"
SLOT = EMB(0)
READID = YES

SYSTEM
NONVOLATILE=8192
AMPERAGE = 26500
SLOT(1)=EISA
BUSMASTER=YES
SLOT(2)=EISA
BUSMASTER=YES
SLOT(3)=EISA
BUSMASTER=YES
SLOT(4)=EISA
BUSMASTER=YES
SLOT(5)=EISA
BUSMASTER=YES
SLOT(6)=EISA
BUSMASTER=YES
SLOT(7)=EISA
BUSMASTER=NO
SLOT(8)=EISA
BUSMASTER=NO

However, when I look at the motherboard itself, I get the impression that only slot #7 is slave:

20190709-234956.jpg

Any insight would be greatly appreciated!

-Brian

Reply 82 of 115, by NJRoadfan

User metadata
Rank Oldbie
Rank
Oldbie

I don't think the SiS EISA chipset supports more than 6 slots as busmaster. The reality is most folks don't have that many busmastering EISA cards to begin with so its rarely a problem. Both my Micronics and Tyan (uses SiS chipset too) made the VLB slots non-busmaster since it was likely they would see VLBus cards anyway.

Reply 83 of 115, by bbuchholtz

User metadata
Rank Newbie
Rank
Newbie

Thank you for your reply 😀

That does make sense. And I agree that having 6 bus mastering slots is probably enough.

This was just my wishful thinking. Most full-length cards take advantage of bus mastering. Slots #7 and 8 have plenty of clearance. Unfortunately, the CPU fan and heatsink prevents the use of full-length cards.

-Brian

Reply 84 of 115, by bbuchholtz

User metadata
Rank Newbie
Rank
Newbie

Hey Guys,

Which ECU are you using to manage EISA with this board? So far, I've had the most luck with this one:
http://files.mpoli.fi/unpacked/hardware/misc/ … r/int-eisa.lzh/

However, it's not perfect. In particular, I'm trying to add this SCSI host adapter:
http://108.59.254.117/~mR_Slug/EISA/!CPQ4002.CFG/

I'm getting the impression that I should should be using a Compaq ECU with this peripheral. But so far, Compaq's ECU always seems to hang whenever I try saving the EISA settings.

-Brian

Reply 86 of 115, by bbuchholtz

User metadata
Rank Newbie
Rank
Newbie

Ok, I got the AMI EISA CFU working with my system. At first, I was greeted with this message, after saving EISA settings:
20190715-185223.jpg

Based on my Googling, it would seem that EISA has a limited amount of space for configuration. Long story short, I had too many EISA cards installed. When I pulled some cards out, the error went away and I was able to boot from the HBA.

Onto my second issue... All CFUs fail on parsing this CFG, except for Compaq's:
http://108.59.254.117/~mR_Slug/EISA/!CPQ4002.CFG/

20190715-205709.jpg

At first, I was getting complaints about lines 14 and 15:

 VERSION=1.2AH
REV=0

These seemed benign; so I commented them out. However, line 45 seems more crucial:

 RESOURCE="ADDRESS",COMMON,LOCAL

Unfortunately, Compaq's CFU still hangs on my system...

-Brian

Reply 87 of 115, by NJRoadfan

User metadata
Rank Oldbie
Rank
Oldbie

How many cards/what cards are installed? The Shadow RAM error is likely something else. I know the CMOS setup has various shadowing options.

Reply 88 of 115, by bbuchholtz

User metadata
Rank Newbie
Rank
Newbie

Currently, I do not have any POST errors. Here is my current configuration:

[SLOT1] Adaptec AHA-2740W (EISA bus master)
[SLOT2] (Empty)
[SLOT3] Diamond Stealth64 4MB (VLB)
[SLOT4] Intel EtherExpress PRO/100 (EISA bus master)
[SLOT5] Adaptec AHA-1542CF/1540CF (16-bit ISA)
[SLOT6] Gravis UltraSound Plug'n'Play (16-bit ISA)
[SLOT7] SIIG High Speed IO-1809 16550 Serial (16-bit ISA)
[SLOT8] (Empty)

Are you suggesting I should try playing around with the Shadow RAM settings in the CMOS?

Thank you again for taking a look!

-Brian

Reply 89 of 115, by Intel486dx33

User metadata
Rank l33t
Rank
l33t
Anonymous Coward wrote:

It's hard to say how much it cost me, because I've been working on it for 7 years...but probably something like $400.

I'm running PC-DOS 2000 and WFW311.

This is the Nicest computer I have seen in a while.
But it would cost allot more than $400 to build today with the inflation of used computer parts.
What are the benefits of EISA cards over VLB and ISA ?

Reply 90 of 115, by NJRoadfan

User metadata
Rank Oldbie
Rank
Oldbie
bbuchholtz wrote:
Currently, I do not have any POST errors. Here is my current configuration: […]
Show full quote

Currently, I do not have any POST errors. Here is my current configuration:

[SLOT1] Adaptec AHA-2740W (EISA bus master)
[SLOT2] (Empty)
[SLOT3] Diamond Stealth64 4MB (VLB)
[SLOT4] Intel EtherExpress PRO/100 (EISA bus master)
[SLOT5] Adaptec AHA-1542CF/1540CF (16-bit ISA)
[SLOT6] Gravis UltraSound Plug'n'Play (16-bit ISA)
[SLOT7] SIIG High Speed IO-1809 16550 Serial (16-bit ISA)
[SLOT8] (Empty)

Are you suggesting I should try playing around with the Shadow RAM settings in the CMOS?

Thank you again for taking a look!

-Brian

That isn't too many cards. Personally I set the entire UMB area along with the usual BIOS/Video BIOS to shadowed and then use the latest version of "The Last Byte Memory Manager" to use the area for loading things HIGH. I have the same Adaptec card and never had a problem with it on that motherboard.

Reply 91 of 115, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t
Intel486dx33 wrote:
This is the Nicest computer I have seen in a while. But it would cost allot more than $400 to build today with the inflation of […]
Show full quote
Anonymous Coward wrote:

It's hard to say how much it cost me, because I've been working on it for 7 years...but probably something like $400.

I'm running PC-DOS 2000 and WFW311.

This is the Nicest computer I have seen in a while.
But it would cost allot more than $400 to build today with the inflation of used computer parts.
What are the benefits of EISA cards over VLB and ISA ?

EISA over ISA? Multiple busmasters, 33MB/sec vs 5MB/sec bandwidth, 32-bit addressing (no 16MB limitation).
EISA over VLB? Stability, supports way more cards.

EIsA is the way to go if you want wide SCSI and 100MB/sec network controllers. Admittedly, not such a good choice for graphics but you can get boards with both VLB and EISA slots and have the best of both worlds.

Bringing PCI into the equation changes things a bit, as it's much cheaper and easier to get PCI 486 boards. There are a lot of crappy PCI 486 boards though, whereas most of the EISA 486 boards are quite solid.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 92 of 115, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I don't have any big news to report, but I do have a few small tidbits that may or may not help people trying to run Cyrix 5x86 CPUs on older 5V motherboards.

As I mentioned several years ago, I have a Cyrix 5x86-100/4X that I was unable to get working on any of my 5V-only boards that accepted the 2x/3x masked 5x86 chips (with some trickery). Well, that is until recently when I made an accidental discovery playing with my DTK PKM-0031Y Symphony based ISA-only board. That board uses an AMIBIOS that is similar to the one on my SuperEISA. It exhibits the exact same behaviour as my SuperEISA when handling Cyrix chips. A year or two ago a large stash of MR-BIOS ROM images were found, including images that work with the Symphony chipsets. Since I owned a Symphony board, I figured I'd just give a shot. On top of all the other goodies MR-BIOS normally provides, it also added additional CPU support. While not being new enough to support Cyrix 5x86, it seems it added just enough Cyrix support for the 5x86 chips to boot up properly in my board....including the 5x86-100/4X. So, it seems that the issue is indeed a BIOS problem, not a hardware one.
Unfortunately, MR-BIOS didn't make replacement BIOSes for EISA boards, so that doesn't exactly help specifically with the Nice SuperEISA, but perhaps the existing BIOS can be modded.
There is also an Award BIOS for the SiS EISA chipset, which I believe came from the A.I.R. 486EI, that I tested as working in my board (I think it only saw up to 512kb cache). That one also did not like the Cyrix 5x86, but it seems the Award BIOS is better understood by modders, so there may be hope for that one.

So other than that, the only thing I've really learned is that my 5x86-100/4X can apparently work at 133MHz, at least running the Quake timedemo loop for several hours didn't lead to a crash. That may be a first for a 100/4X chip.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 93 of 115, by BitWrangler

User metadata
Rank l33t++
Rank
l33t++

Yes, award BIOSes above a certain age you can swap CPU modules in and out, give your 486 board PIII support if you want 🤣

Unicorn herding operations are proceeding, but all the totes of hens teeth and barrels of rocking horse poop give them plenty of hiding spots.

Reply 94 of 115, by feipoa

User metadata
Rank l33t++
Rank
l33t++
Anonymous Coward wrote on 2021-05-22, 15:11:

my 5x86-100/4X can apparently work at 133MHz, at least running the Quake timedemo loop for several hours didn't lead to a crash. That may be a first for a 100/4X chip.

Your Cyrix 5x86-100/4x is marked as 100/4x, or is unmarked for speed? If marked for 100/4x, this is an interesting discovery. Which register settings did you have enabled? If BTB and FP_FAST were set to 1, and LSSER set to 0, that is quite the discovery. Are you able to give Windows 9x a try (obviously BTB must be set to 0)?

Plan your life wisely, you'll be dead before you know it.

Reply 95 of 115, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Yes, it has the 100/4X silkscreen markings. It came from a broken evergreen upgrade.
LSSER was 0, and BTB was almost certainly disabled, since the chip is S0R5.

I believe the Award EISA BIOS is version 4.50, which is a bit different than 4.51. Does anyone know if there are tools to alter v4.50?

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 96 of 115, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I've been having some issues with my SuperEISA board for the last 5-6 years. It had an annoying problem of taking about a minute to warm up before POSTing. With the help of a POST card I was able to trace the problem to the keyboard controller. A number of years back I was playing around with the keyboard controller socket while trying to develop a PS/2 mouse adaptor. My KBC socket uses machine pins, and I had been jamming solid core copper wires directly into the holes to connect up a breadboard. As it turns out I wore out the socket, so the KBC was not making good contact. The temporary solution is to sandwich in another machine pin socket. I guess at some point I should probably replace it with a standard one. The reason this board uses a machine pin socket is because this board was an OEM for one of the Everex STEP systems which used a special message display controller which fit under the KBC.

I am also still being driven nuts by the Cyrix 5x86 issue that requires a few inconvenient hacks to get it working. I noticed in the ST5x86 databook that it says this:

"It is recommended that the ADS#, LOCK# and SMI# output pins be connected to pull-up resistors, as indicated in Table 2.2. The external pull-ups guarantee that the signals remain high (inactive) during hold acknowledge states."

I'm going to try making a custom interposer to see if it helps at all.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 97 of 115, by feipoa

User metadata
Rank l33t++
Rank
l33t++

When you say custom, I assume you just mean to take a thin wire and solder it from the Vcc on the socket to ADS#, LOCK#, and SMI# (also on the socket)?

Plan your life wisely, you'll be dead before you know it.

Reply 98 of 115, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Correct. I don't want to solder directly on the VRM, CPU or motherboard.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium

Reply 99 of 115, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I am going to add the CPU bus frequency table for the Tyan S1437 here, because it is not on TH99 (or anywhere else I can find), and it may be of use to somebody else (or myself in the future). Somebody can add it to Ultimate Hardware 2019 if they want"

Jumper | JP26 | JP27 | JP28
16MHz | 1-2 | 1-2 | 1-2
20MHz | 2-3 | 1-2 | 1-2
25MHz | 1-2 | 2-3 | 1-2
33MHz | 1-2 | 1-2 | 2-3
40MHz | 2-3 | 2-3 | 1-2
50MHz | 2-3 | 1-2 | 2-3
??MHz | 2-3 | 2-3 | 2-3 (something really slow, maybe 4MHz)
??MHz | 1-2 | 2-3 | 2-3 (even slower, maybe 2MHz)

The clockgen on this board is AV9107. It also supports 66, 80 and 100MHz, but maybe not all the pins are wired up. You can probably hack it to provide these, but at best only 66MHz would be useful. This clockgen along with a 14.318MHz input crystal sit within the pads of a standard oscillator socket, so I would guess if you removed these components and soldered in a socket you should be able to use external oscillators instead. Not sure if any additional jumpers would need to be set.

"Will the highways on the internets become more few?" -Gee Dubya
V'Ger XT|Upgraded AT|Ultimate 386|Super VL/EISA 486|SMP VL/EISA Pentium