It's this infamous board

More info here:

http://redhill.net.au/b/b-96.html

Don't buy it...it has no L2 cache so it's dog slow
Mine went into the bin during my move 😵

Oh dear! 🙁 😢 I already bought it. However, this photo shows the 2 fake cache chips "soldered out" on the PCB, which means they are not trying to use fake chips. Maybe that cache stick to the left will work OK? Nevertheless, it's another poor purchase decision by me. Agh! PC Chips strikes again!!

Mine looked the same

But it still had no L2 cache, the cache module is a total fake
If I'm not mistaken no variant of these boards had real L2 cache..even though it had soldered cache looking chips 😢

🙁 😢

retro games 100 wrote:

🙁 😢

At least you own a "historic" board now 😀

The board has a very nice layout, wish Asus made a board looking like this back then

will it play crysis >:P hehehe sorry on the bad purchase

I did a very quick test and it POSTs - that's the good news. 😀 Speedsys gives me a score of 17.24, which seems low, but I'm not sure how well a Cyrix DX 40 should perform. (Typical speedsys tests on an Intel DX2-66 give me about 24.something on most of my mobos). Hopefully, I'll get some time later on today to run some more tests. The initial BIOS POST datestamp displays: 05/06/96 for the AMI BIOS, but the "table of hardware information" 5 seconds later says something/'94. Also, the cache is displayed onscreen as 256k, but I reckon that's nonsense. The mobo's system jumpers are set to 40M. (Not 25M or 33M, or 50M). The mhz speed of the chip is displayed as 40mhz, but speedsys says it's 13mhz. Strange. Speedsys also reports nothing at all about L2 cache. I guess that means it couldn't find any, because it's not on the board! Also, on the cache stick it says "For 486 M919 use only".

I found out quite a bit about the board here -

http://motherboards.mbarron.net/models/486pci/m919v1.htm

Edit: the above link is for version 1 boards. The later version boards (eg 3.4) is below -

http://motherboards.mbarron.net/models/486pci/m919v3.htm

The DL jumper manual section contained some interesting info.

I just ran the freely downloadable Cachechk (version 7), by Ray Van Tassle, and it reports that the mobo's cache is fake. I disabled the L2 cache option in the BIOS, then reran speedsys, and it doesn't affect the overall score at all. It's still 17.24.

Edit: When the mobo arrived, it had 3 sticks of RAM in its 4 RAM slots. I removed one stick, to just leave a pair of RAM sticks. These are single sided sticks - the one I removed was double sided. The BIOS POST screen reports that the RAM total is now 8mb. It was 16mb. Running speedsys on the default "safe" cache and memory timing settings gave me 66mb/s memory bandwidth. I adjusted the wait states (read and write) from 2 to 1, then reran speedsys. I then got a score of about 88mb/s. I adjusted the wait states in the BIOS set up area once more, to 0 for both read and write, and now speedsys reports that the memory bandwidth is 107.34mb/s. Wow, that's quite high isn't it? For every speedsys test run, the overall score is always the same: 17.24.

I removed the Cyrix DX 40 CPU, and replaced it with an Intel DX2-66 CPU. (I also adjusted the mobo's system speed from 40 to 33.) This SL enhanced but non-write back cache CPU (SX807) usually gives me a score of about 24.something on most of my mobos. On this thing, I get 23.84, which is surprisingly OK! 😀

Edit: Another surprising minor improvement found: I noticed that the 2 sticks of RAM onboard had "60" markings. I went to the BIOS, and set the DRAM setting from Normal to EDO. That worked! Also, I "maxed out" the cache (hehe) and memory timings (to 0 wait states etc), and reran speedsys. I now get 24.95, which really is fairly respectable, given it's a fake cache PC Chips board.

Memory bandwidth is 100.16 mb/s, and L1 = 61.86, and memory throughput = 39.88.

Edit 2: I did the doom -timedemo demo3 benchmark, and got 26.39 fps, which is not bad.

Good to see you've got the board whipped into shape. I guess this just goes to show: when it comes to 486-class CPUs, Intel is the way to go. Cyrix is useless.

I pulled the cache stick out of the mobo (which was hard to do cos it was really stuck!), and the BIOS POST screen now says "Write Back Cache ON". Yeah, right! 🙄 I run Speedsys:

Memory bandwidth = 70.75
General score = 25.01
L1 = 47.94
Memory throughput = 28.22

I then "max out" the BIOS memory and cache settings, then rerun Speedsys:

Memory bandwidth = 99.37
General score = 25.04
L1 = 61.86
Memory throughput = 39.88

I then pushed the "fake cache" stick back in to the mobo. The BIOS POST screen now reports "256 KB cache memory". 🙄 I run speedsys:

Memory bandwidth = 70.75 = Identical to test #1 above!
General score = 25.01 = Identical to test #1 above!
L1 = 47.95
Memory throughput = 28.22

Conclusion: the 256 kb cache is reported by the BIOS POST screen, but not used by the mobo! Scandalous! 😮

I then "max out" the BIOS memory and cache settings, then rerun Pcpbench, using a 2mb basic Virge card, and get 5.2

Is it possible to use a real COAST cache module on this board?

Anonymous Coward wrote:

Is it possible to use a real COAST cache module on this board?

Please go to this webpage -

http://motherboards.mbarron.net/models/486pci/m919v3.htm

Then DL the .zip file from the "download jumper manual" link, then read the help-tips.htm file. It's got lots of info about this "M919 VIP" board, including plenty of commentary about its cache options. I haven't read it from end to end, so I can't answer your specific question, but it seems this guy is a fan of the board, and maybe he answers your query in this help page. HTH.

Wow, that is a pretty cool find! I have read about that board in a few places, and a friend had one a long time ago. It really wasn't that slow considering the lack of L2 cache. I think it is a neat board to have just for the historical aspects!!!!! I recall that my friend tried an actual COAST module with 256kb on it and the board wouldn't post at all with it installed. Would be interesting to see if your board would work with a COAST module.

Weren't most of the cache modules sold pipeline burst? I'm pretty sure that board would only support asynchronous/synchronous cache. Assuming you could find such a module, it would possibly work.

If you search for: PCchips M919 on ebay.com, you'll find a fairly well known seller who still has 6 of them. Incredibly, he's sold 3 of them already, at inflated prices. It looks like the version he has is the same as mine: V3.4B/F. This seller has thousands of retro items for sale. Very unfortunately, the prices for nearly all of them are too high. One extreme example - he has an exciting mobo for sale, item title:

486 Motherboard MiniEISA-486AS EISA DX4-100 32 MB 256K

...But he wants $1000 for it.

Regarding the M919 again - On this webpage: http://redhill.net.au/b/b-96.html, it says:

"The cache slot is not a COAST at all - it really takes a special asynchronous module (something like DIPPs on a stick) rather than the pipeline burst cache the Pentium boards took."

MiniEISA isn't interesting at all. It's just a crippled version of EISA. It supports the extra address lines of EISA without supporting a lot of its extra features, so cards that use busmaster won't work properly.

Other 486 motherboards used cache modules, did they not? I would guess they should be fairly standard.

People would have noticed their cache modules were fake.

I have the V3.4B/F revision of this board, mine is not the PC Chips model though [I don't think], I forget what company produced the one I have.

This site (very old but useful) has alot of information on this board, it's revisions and tweaks to get the most [speed] and stability out of this board.

And there's also this site as well, and provides information as well as jumper manuals, etc.

Latest BIOS: Here (Direct Link)
UMC [IDE Drivers]: Here (Direct Link)

I hope someone will find this information useful. 😀