VOGONS


My Ultimate VL/EISA 486

Topic actions

Reply 60 of 91, by feipoa

User metadata
Rank l33t++
Rank
l33t++

It would be nice to see these results plotted in some bar charts.

Ultimate 486 Benchmark | Ultimate 686 Benchmark | Cyrix 5x86 Enhancements | 486 Overkill Graphics | Worlds Fastest 486

Reply 61 of 91, by matze79

User metadata
Rank l33t
Rank
l33t

It would be very interesting to see some 32bit EISA VGA Cards compared to VLB.

https://dosreloaded.de - The German Retro DOS PC Community
https://www.retroianer.de - under constructing since ever

Co2 - for a endless Summer

Reply 62 of 91, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

That would be a pretty difficult comparison, because VGA cards for EISA bus are usually antiquated. I think a Mach32 or S3 928 VRAM comparison could be possible though. Other VGA chips that are *probably* on EISA bus are S3 911/924, ET4000AX, Compaq Qvision and something from IIT (I think).

The only EISA VGA card I presently own is the Compaq QVision 1024. I am not aware of a VLB equivalent though. I used to have a 928 based card, but I traded it away several years ago. I remember that card was not very fast though.

"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 63 of 91, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

There is something interesting I noticed about using Cyrix 5x86 processors on the SuperEISA (Tyan S1437 and other 5V only boards) is that I cannot get the system to post when the Cyrix is jumpered for 2X or 3X mode. However, when I set the jumper on my VRM to the "2.5X" setting which was originally intended to be used by the IntelDX4, I can then get the system to boot and then switch the 5x86 into 3X mode using a utility like CTCHIP. The 2.5X setting apparently leads to the BREQ pin, and I'm not sure if there are any negative long term affects by doing this.

More interestingly is that this "2.5X" trick does not seem to work with the Cyrix 5x86 chips that have 3X/4X multipliers. I tried it with a 5x86-100/4X, and could not get any sign of life at all. This is rather unfortunate, because it means I likely cannot upgrade to 133MHz version.

Has anyone else experienced this problem? I am wondering if maybe certain CPU interposers have a workaround for this. It seems to be a timing problem.

"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 64 of 91, by j^aws

User metadata
Rank Oldbie
Rank
Oldbie

^^ I don't have the same board, but on my Acer J3 VL/ EISA system with the Socket 3 cartridge, I managed to get the Cyrix 5x86 running with a simple interposer at 100 MHz using 3x multi. And IIRC, with Write-Back cache, too. Also, booted an Intel POD83 at 100 MHz using 2.5x multi.

Maybe your issue is relating to chipset/ BIOS?

Reply 65 of 91, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

Hmm, I guess you are correct that the J3 should support write back L1 cache, because it would have to in order to property support the Pentium CPU cards. I have the J3, but not a 486 CPU card. It would be nice for testing, but I don't think I would actually use it since it's much more interesting to have a dual Pentium with VL and EISA busses.

I suspect the issue is more likely related to the chipset, as I have also tried an AWARD BIOS on this board. But, it's more than likely just a matter of adding an extra resistor or capacitor somewhere.

"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 66 of 91, by Disruptor

User metadata
Rank Member
Rank
Member

I'm using a U320 seagate 10K 36G hdd on my EISA 2740W controller.
When configuring the BIOS with Compaq's ECU, there are speed options up to 10.0 MB/s only.
HDD support is up to 8G with BIOS. Speed is 8.0 MB/s in DOS with BIOS 2.11 driver.
After loading ASPI7DOS.SYS 1.42 I get 16.8 MB/s and full 36G support (but the disk is still bored).
However, when I try the stock Win95C miniport driver, I get 6.8 MB/s only.
Using Win98SE I'm struggeling with 5.2 MB/s
There is no difference when loading ASPI7DOS.SYS in Win9x' CONFIG.SYS

486AE100.png
Filename
486AE100.png
File size
9.27 KiB
Views
771 views
File comment
100MHz Am486DX4-100NV8T @50x2 L1=WT, L2=WB
File license
Fair use/fair dealing exception
486AE150.png
Filename
486AE150.png
File size
9.42 KiB
Views
771 views
File comment
150MHz Am5x86-133 @50x3 L1=WT, L2=WB
File license
Fair use/fair dealing exception

Board: ABIT 486 EISA-AE4 running at 50 MHz FSB -- mod: added a socket for dirty tag ram, including dirty tag sram
DRAM: 8x4 MB = 32 MB 30 pin SIMM
CPU: AMD 486 DX4 NV8T, running at 50x2 on a 3.3 V adapter (L1 WB not possible due to board limitation, so using a WT CPU)
Timings (stable): 50 MHz, DRAM: slower & Write CAS 1T, Cache: Write Cycle 2T & Burst Read Cycle 1T, Bus: CPUCLK/6
SCSI: Adaptec 2740W (BIOS 2.11)
IDE: no-name ISA IDE with multi-io
Graphics: Compaq QVision EISA (seems to be faster than my ELSA Winner 1000)
Network: 3Com Etherlink 10/100 3c597
Sound: SB16 ISA
HDD: Seagate ST336607LW

The added dirty tag ram did not show much difference in L2 WB performance on 33 MHz but on 50 MHz.
Tag rams are 20 ns, cache rams are 15 ns.

3c597 windows 9x performance is 14.4 MBit/s SMB1 (file cache) and 12.2 MBit/s (non-cached).

What plain HDD speeds have you noted at Win9x?
Do you have any idea where I can get a newer or faster Adaptec Win9x miniport driver?

Last edited by Disruptor on 2018-08-25, 14:16. Edited 1 time in total.

Reply 68 of 91, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I have not tried 9x with my AHA 2742W. I have only ran NT4 and Windows 3.11. I haven't noticed problems in either. This adapter is natively supported by NT4. Does it have similar support in 9x? I know with Windows 3.11, you need to install a special driver from adaptec in order to take full advantage of the adapter. Does such a utility exist for 9x?

Also, in the BIOS setup it does not specifically list 20MB/sec support like BIOSes on newer adaptec cards. Instead, the 20MB/sec is enabled by checking the box for enabling the "wide" option.

BTW, v2.11 may not be the latest BIOS release. I have two of these cards, and I think my newer one came with a 3.x BIOS...I'll have to check.

Also, I highly recommend dual booting with NT4 and windows 3.x, because 9x sucks!

"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 69 of 91, by NJRoadfan

User metadata
Rank Oldbie
Rank
Oldbie

Adaptec offered drivers for Windows 9x for the 2740W, its the same drivers (and SCSI chip, the AIC-777x) as the VL-Bus 2840VL. Oddly these files are missing from Adaptec's support site. The full SCSISet diskettes and the EZ-SCSI suite had drivers for DOS/Win 3.1 (including the .386 driver for 32-bit disk access) and Windows 9x/NT.

If you need the latest BIOS, its still on Adaptec's website to burn onto a EPROM.

Reply 70 of 91, by Disruptor

User metadata
Rank Member
Rank
Member

Yes, the 274x does have native support in Windows 95 and Windows 98.
Perhaps I will put my GUS classic 2.2 into this computer. Since the drivers do not work under 98, I'll stay with Win95C then.

I can find a v2.10 BIOS on the Adaptec website.
My adapter has v2.11 already.

Because the mainboard BIOS does not configure the chipset properly, I've to run a debug script in
autoexec.bat

debug < wb.txt

wb.txt

o c18 61
o c1c ee
q

When I install Windows NT, I will have to patch the MBR to get full motherboard performance.

What's the most recent version of EZ-SCSI or SCSISet? Can I download this from Adaptec?
Can you send me a copy of your newer BIOS please?

Reply 71 of 91, by BitWrangler

User metadata
Rank Member
Rank
Member
Anonymous Coward wrote:

There is something interesting I noticed about using Cyrix 5x86 processors on the SuperEISA (Tyan S1437 and other 5V only boards) is that I cannot get the system to post when the Cyrix is jumpered for 2X or 3X mode. However, when I set the jumper on my VRM to the "2.5X" setting which was originally intended to be used by the IntelDX4, I can then get the system to boot and then switch the 5x86 into 3X mode using a utility like CTCHIP. The 2.5X setting apparently leads to the BREQ pin, and I'm not sure if there are any negative long term affects by doing this.

More interestingly is that this "2.5X" trick does not seem to work with the Cyrix 5x86 chips that have 3X/4X multipliers. I tried it with a 5x86-100/4X, and could not get any sign of life at all. This is rather unfortunate, because it means I likely cannot upgrade to 133MHz version.

Has anyone else experienced this problem? I am wondering if maybe certain CPU interposers have a workaround for this. It seems to be a timing problem.

The hacky way around it is to get the pinout diagram and figure out which pins to insulate by "taping" or coating in nail polish or similar, then wiring up active high or active lows from multi signal pins to Vcc and Gnds on the base of the pins with thin strands of wire wrapped around.

Techniques similar to "vid pinning" mods, or Bsel and multi mods on S370

Edit: Oh, another trick is to find some thin wire, that when you strip some insulation off, if the insulation is thin enough you can use that sleeve to insulate a pin, only works with ZIF sockets though, where the hole is a little bigger and there's clearance in the contacts until the lever is clamped.

Basement full of ancient PC stuff, starting to go through it. Most recently toyed with DOS era stuff 15 years ago, so memory might be rusty. So what's this BitWrangler guy's deal ??? >>> Taming the pile, specs to target?

Reply 72 of 91, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

I don't think insulating a pin in this case is going to work. There seems to be some kind of BIOS or timing incompatibility with the Cyrix 5x86 on this board.

On my POST card, my BIOS gets stuck at code 05 with the 5x86 installed.

Soft reset/power-on determined. Going to enable ROM. i.e. disable shadow
RAM/Cache if any. Going to enable ROM.i.e. disable shadow RAM/cache if
any.Chipset initialization over,DMA and interrupt controller disabled. CMOS
pending interrupt disabled.

"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 73 of 91, by isasoundcardsfan

User metadata
Rank Newbie
Rank
Newbie

Hello Anonymous Coward, i got the same mb but without a bios. Tried to find in the net some .rom but unsuccessfully. Could you please upload a dump from yours if it's possible ?

Reply 74 of 91, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

486 EISA VL (EV-3605) - Help with BIOS!

"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 75 of 91, by aries-mu

User metadata
Rank Member
Rank
Member
Anonymous Coward wrote:

I've been working on this system for what seems like forever. I've made a few small changes recently. Here's what I've got so far...

BEAUTIFUL!!!!

I wonder how do those EISA cards perform! It is true that EISA is a 32 bit bus, but it is also true that it's limited to 8 MHz. ISA can get to 16 MHz, so, compared to the latter, EISA is really just twice as fast.
Any experience, real-life stressing software, or benchmark result you might have run which you might want to share?

Thanks

They said therefore to him: Who are you?
Jesus said to them: The beginning, who also speak unto you

Reply 76 of 91, by Anonymous Coward

User metadata
Rank l33t
Rank
l33t

In practice, ISA is rarely ran at 16MHz. That's not even in spec. ISA is supposed to run at 8MHz, just like EISA. You can overclock a little, but if you push it beyond 10MHz you're asking for trouble. For the record, EISA can also be overclocked a little. I like to run mine at 10MHz, and haven't noticed any problems.

The reason I like EISA is not just the extra speed, but also the fact that it has 32 address lines and proper bus mastering. It's probably the only fully matured 32-bit solution for 486 boards.

EISA is really only practical for network cards and SCSI controllers. There are a few VGA cards kicking around, but they're somewhat older than your typical VLB card. If you want to run EISA on a 486 you should really get a hybrid one with VLB slots anyway.

I do have benchmarks around somewhere. I thought they were in this thread...I should really add benchmarks from DOOM and 3dbench though.

"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 77 of 91, by Intel486dx33

User metadata
Rank l33t
Rank
l33t

Unique, Rare and totally Awesome.

Great job.

I have an IBM PS/2 with EISA slots but the battery burst and leaked all over the motherboard so now the motherboard needs repair.

Reply 78 of 91, by bbuchholtz

User metadata
Rank Newbie
Rank
Newbie

Beautiful computer 😘

I've been on a 486 kick, and recently built a couple of computers (VLB and PCI). I thought I'd build an EISA-based machine. It just so happens I recently acquired a Nice SuperEISA board.

I'll be the first to admit that I'm not super familiar with EISA. Growing-up, I was more of a VLB guy. But, so far I got my board to boot with an AMD 5x86 and 256MB of memory. I've also been testing with an Intel 486 DX2 66.

Here's where I have problems... I get video just fine from my 16-bit ISA Tseng Labs ET4000/W32I 2MB. However, I can't get any of my VLB video cards to work (one of which is a Hercules dynamite VL Pro). How'd you get VLB video to work with this board?

Thanks!

-Brian

Reply 79 of 91, by bbuchholtz

User metadata
Rank Newbie
Rank
Newbie

Well, I spoke too soon. I figured out my problem 😎

I saw in the picture that your board's jumper 31 was enabled. This was not enabled on my board. This was the magic my VLB video card needed!

-Brian