brad1982_5 wrote:

Yes I think I know what you mean. The cache sticks usually state "FOR M919 USE ONLY" Why my one doesn't I don't know. It does show as having 256KB cache memory after post but I have never ran any diagnostics to check it any further. I got the board at a car boot sale many years ago so I don't know it's full history.

use cachechk http://www.minuszerodegrees.net/software/software.htm

it will tell you if its being used.

as I said before mine reported 256k being detected at POST but it actually wasn't using it. I needed to swap EDO RAM for FPM and then it worked fine.

"Future will support pipeline burst SRAM module."

Gotta love chinese/taiwanese manufacturers. Other than that I think this is quite impressive for a 486 board

Cheers for the CACHECHK info. 😀 I will get this board back in a case some day to test it. It also has EDO ram installed at the moment of 4x 32MB sticks but I do have 16MB FPM modules that I can try.

brad1982_5 wrote:

Cheers for the CACHECHK info. 😀 I will get this board back in a case some day to test it. It also has EDO ram installed at the moment of 4x 32MB sticks but I do have 16MB FPM modules that I can try.

Yep indeed 😀
The rest of the manual is very nicely done and is easy to understand.

Thanks, those photos give me another angle to look at things from.

I have made up my mind here. Once I'm done with my current 486 project, I'll try tracking down a module myself so I can do a proper job of this. Reverse engineer it and then fabricate the clone before tracking down an M919. I don't know that I'll ever get other cache sizes to work though, but I'm relatively confident that I'll be able to produce a working 'copy' of the original board. Assuming all goes well, I'd try and sell a few off and then throw the Gerber files on the web for everyone so they could just fab their own. Annoyingly the only fab I have access to is at it's limit with those vias, meaning the board size is increasing.

I know for a fact that PBurst is out of the question. Hehe, damn marketing.

Am I the only one who noticed the solder pads are for PIPELINE BURST? On a 486? 😮

edit:OK I'm not the only one 😁

OK, so given the difficulty of finding cache, I'm not even going to consider it for now. That said, I'll just pick up EDO RAM for the board, and should a cache module become available down the line, I'll look for FPM at that time.

A few other things I'd like some direction on:

1) I have a 486DX-50 processor (not DX2). As I understand it, things get unstable with 40 and 50MHz bus speeds, so I plan to clock the processor to 33MHz. Will I have any problem with that?

2) A PCI video card seems the logical way to go, but I'm not sure which models to look for if I want to retain driver compatibility with old OS's such as DOS 6.22 and Win95. Any suggestions on the video card? There is one free VLB slot, so a VLB video card is a possibility, I suppose.

3) For a hard drive, I'm thinking of using a 4.3GB uw SCSI drive I have, and an adaptec 2940u2w that I also have. Both are just in a drawer at the moment. As best as I can remember, the 2940u2w shouldn't have any trouble in a DOS environment, but I'm not 100% sure. Can anyone chime in on that? If I don't use the SCSI, I'll need to pick up an older IDE drive and use the motherboard's onboard controller. Does anyone have any specific IDE drive suggestions? I'm guessing anything under 8GB should be fine. Or do I need to stay under 512MB?

4) For RAM, if I recall, 72 pin simms will run fine in pairs, no? Was it only 30 pin simms that needed quartets? My tentative plan is to get a couple 32MB EDO Simms. Does 64MB sound like a reasonable amount of memory?

Thanks

AllUrBaseRBelong2Us wrote:

1) I have a 486DX-50 processor (not DX2). As I understand it, things get unstable with 40 and 50MHz bus speeds, so I plan to clock the processor to 33MHz. Will I have any problem with that?

Things that get unstable at 40-50 MHz are usually the L2 cache (M919 has none, problem solved 🤣) and PCI. AFAIK the M919 engages an 1:2 PCI divider at 40 and 50 MHz, so there shouldn't be any issues.

AllUrBaseRBelong2Us wrote:

2) A PCI video card seems the logical way to go, but I'm not sure which models to look for if I want to retain driver compatibility with old OS's such as DOS 6.22 and Win95. Any suggestions on the video card? There is one free VLB slot, so a VLB video card is a possibility, I suppose.

S3 Trio or Virge PCI cards are cheap, easy to obtain and compatible with everything. Comparable VLB cards are rare and much more expensive.

AllUrBaseRBelong2Us wrote:

4) For RAM, if I recall, 72 pin simms will run fine in pairs, no? Was it only 30 pin simms that needed quartets? My tentative plan is to get a couple 32MB EDO Simms. Does 64MB sound like a reasonable amount of memory?

For 486 one 72-pin SIMM would be sufficient. Pentiums need at least two because of 64-bit RAM access.

I would not be so sure about EDO in a 486. These tend not to work even in late boards. The manual can claim that they do, but it's hardly true.

AFAIK the M919 engages an 1:2 PCI divider at 40

1:2 divider at 40? Ugh.
20MHz is really slow for PCI. Many cards handle 40 just fine

Apparently this doesn't happen if you switch to 40MHz while the system is running... Not that I'd recommend that.

I believe the PCI bus ends up running at 27mhz but don't quote me on that. its still slow. at 40mhz with a Cyrix 5x86 and a Ark logic 2000MT PCI card I was still getting beat with some benchmarks with a 33mhz bus AMD cpu at 133mhz

if you have a turbo switch you can set the jumper to the 40mhz jumper. you boot up at 33mhz and after POST hit the switch at go to 40mhz. Ive been doing it without any issues for a few months now.

EDO works fine in this board, except for the whole L2 cache issue that may or may not be an issue with all or just my board.

I second the S3 Trio/Virge suggestion

I'll probably go with an S3 Virge then. I used to have one of those cards back in the day. It should be nostalgic to get another.

So given the PCI bus automatically clocks down on the m919, I guess I'm safe to try to run the 486DX-50 at 50MHz? I still need to get some parts before I'll be ready to give it a shot. Nice to know I only need 1 simm on a 486.

Still curious about hard drive choice if anyone wants to chime in.

Why not get a 66mhz dx2? Or a 100mhz dx4 or amd 133mhz 5x86 for that matter. Their cheap enough and you'll get to run a pci card at full speed.

For the hard drive id just use whatever is size compatible with the bios and works on hand. Otherwise if you are going to drop money on storage a new CF card and adapter seems to be a popular choice here.

AllUrBaseRBelong2Us, the UM8881/8886BF chipset does not require RAM installed in pairs. Singles are fine. The 2940U2W will work on the m919 in NT4 and W9x. It will not work at all in W2K on the M919, but will work fine on the MB-8433UUD (another board based on the same chipset). EDO RAM will work on this motherboard, however it is not as stable as FPM when you are using tight RAM/cache timings at 40MHz+. The M919 does auto enable a 2/3 divider for PCI at 40 MHz but probably not for the VLB slot. I've had trouble getting reliable graphics operation out of the M919's VLB slot though. Your PCI frequency will be at 27 MHz.

Here's an issue I'm having with an M919 v3.4 board if anyone wants to chime in with suggestions. When I use an AMD 5x86-133 on this board, the BIOS option for L1 is greyed out and stuck on Write-through mode. If I use a Cyrix 5x86, the L1 option is not greyed out and lets me select between write-back or write-through. What is going on here and is there a way to circumvent this problem 1) permenently with some kind of hardware mod, 2) with a software driver mod to flip the Am5x86 back in to write-back mode? Note that another M919 v3.4 board I had did not have this problem. Both boards were jumpered identically.

I don't know why i'm afraid I've never had that problem. My board shows the L1 option OK. It's a V3.4B/F with a AMD 5x86. Are the BIOS revisions the same on both boards you have?

Subject change I wonder if the Pipeline Burst SRAM Module's ever got released?

Yes, both BIOS revisions are the same. I even used a different EEPROM BIOS chip and reflashed. There is some other issue that I could not track down.

Pipeline burst SRAM for the M919? I beleive that was a marketing ploy only as the UM8881 chipset is unlikely to support Pipeline burst. Too bad we don't have the specsheet on the UM8881/8886.

feipoa said;

Here's an issue I'm having with an M919 v3.4 board if anyone wants to chime in with suggestions. When I use an AMD 5x86-133 on t […]
Show full quote

Here's an issue I'm having with an M919 v3.4 board if anyone wants to chime in with
suggestions. When I use an AMD 5x86-133 on this board, the BIOS option for L1 is greyed
out and stuck on Write-through mode. If I use a Cyrix 5x86, the L1 option is not greyed out
and lets me select between write-back or write-through.....Note that another M919 v3.4
board I had did not have this problem. Both boards were jumpered identically.

Maybe check the (applicable) jumpers blocs have their metal inserts inside their plastic sleeves,
maybe try firmly pushing sideways the applicable jumper pins whilst the motherboard is going
through its POST (might act to cause contact if there is a dry solder joint on the pins),
maybe go all multimetery and check the levels during POST on the specific jumper pins (& at
the underside of the CPU socket if it can be accessed from the underside of the motherboard).
The applicable pins could be figured out from the 5x86 pdf from here;
http://web.archive.org/web/20020804155237/htt … _docs/19751.pdf
also, thankfully, J. Steunebrink already sussed these pins out here;
http://web.inter.nl.net/hcc/J.Steunebrink/amd5x86.htm

I even used a different EEPROM BIOS chip

Do you mean you pared back both motherboards to a minimum setup and swapped the EEPROMs
between each board? (Did you also perform a 'clear CMOS' routine... is that not applicable here?)

(In regards to this different matter);

The 2940U2W will work on the m919 in NT4 and W9x. It will not work at all in W2K...

Do you mean the motherboard wouldn't recognize the SCSI device, or mean it couldn't
read files accessed via the SCSI device?