I think what would help for those of us unfamiliar with this board is a photo of the bottom to see if there are traces leading to nowhere. I have one of the old PC Chips M919 boards with the fake SMD cache on it and a slot for a real Cache Module that looks kind of like a COAST slot but is not - the fake SMD cache is obviously fake to anyone like me who makes circuit boards - you can see the traces just loop around.

The only thing that looks fake to me, maybe, is the chips in the Cache sockets, and that's because I've never seen that brand before. Could just be some oddball chinese knockoff brand making Sub-Par chips - you see that a lot when it comes to making electronic stuff using cheap chinese parts off e-bay - like me building guitar pedals - it might be functional, just might be those chips are not performing their best.

creepingnet wrote:

I think what would help for those of us unfamiliar with this board is a photo of the bottom to see if there are traces leading to nowhere.

Good point. Traces seem legit, but I'm certainly no expert 😀

Actually the M919 is the only case I know of fake cache with traces leading to nowhere. If you're really not sure if the mobo supports cache, take a multimeter and check if there is a voltage between where the VCC and GND pins are supposed to be. If there's one, then making fake wires with some getting power would be very strange isn't it ?

If you don't have a desoldering iron, you can simply cut the chips off and then remove each pin using a soldering iron on one side and a pair of pliers on the other side.

I have now two non PCChips boards that have/had fake cache. I added real cache on one of those and it worked without any problems, so did I with my PCChips m915. The last one has a UMC chipset and is quite fast so I will add cache to it as well.

@Deksor, I think you are right matching this to
http://www.win3x.org/uh19/motherboard/show/6238

All jumpers, connectors and whatnot line up. @pano69: can we have a bios dump? Maybe we can find out something for this KDN1401 marking.

So, I finally had some time to hook up this board again to get a BIOS dump (sorry for the delay @evasive), however, during POST it hangs on "Wait....." with POST code A1 which according to this page [1] is "Check for external cache" (or some variant of it).

These is no option in the settings to turn off L2 Cache. I also tried removing the cache jumpers, hoping that that might indicate disabled L2 Cache but no difference. It is weird it has cache jumpers to being with since the cache is soldered onto the board, you'd imagine that the cache config would be hard wired as well (if the cache is indeed real).

Any ideas to make this board boot are welcome.

[1] http://www.bioscentral.com/postcodes/amibios.htm

Edit: settings available. No difference in Cache WS either 0 or 1, same behavior. Also, these settings are different than the ones illustrated in the manual [2]

[2] http://www.win3x.org/uh19/motherboard/show/6238

Interesting. I believe the bios could be modified to unhide the cache setting (because it's probably there and just hidden). Now there might be something else to be wrong with that board, I mean the board should boot without cache installed (why would anyone fake cache and ship a board that cannot work ?).

Maybe the cache settings are incorrect ? (By that I mean that it doesn't match any real cache chip)

Also I just thought of something you could try : since these cache chips are probably fakes, why not trying to piggyback real cache chips over the fake ones ? This should be an easy test without using any soldering iron

Okay, after some more fiddling, I got the board working. Pretty sure I booted it a year ago or so I knew it "could" work. First I tried different memory (30pin SIMMS instead of the 72pin 8MB FPM), no juice but after I changed the CPU from an Intel DX 33 to an AMD 40, voila, it came to life. No idea why the DX33 didn't work, triple checked the jumpers etc...

Anyhow, yep, cache seems to be fake. These are the results of running Ray van Tassle's CACHECHK:

1CACHECHK v4 2/7/96 Copyright (c) 1995 by Ray van Tassle. (-h for help)
2CMOS reports: conv_mem= 640K, ext_mem= 7,7168K, Total RAM= 7,808K
3Clocked at 486 40.1 MHz
4Reading from memory.
5MegaByte#:      --------- Memory Access Block sizes (KB)-----
6      1    2    4    8   16   32   64  128  256  512 1024 2048 4096 <-- KB
70:   26   26   26   27   54   54   54   54   54   54   --   --   --    μs/KB
82:   26   26   26   27   54   54   54   54   54   54   54   54  54     μs/KB
93  4  5  6  7  <--- same as above
10
11Extra tests----
12Wrt  34   34   34   34   34   34   34   34   34   34   34   34   34<-Write mem
13This machine seems to have one cache!? [read]
14  !! cache is   8KB -- 40.6 MB/s  25.8 ns/byte  (199%)          4.0 clks
15  Main memory speed -- 20.4 MB/s  51.4 ns/byte  (100%) [read]   7.9 clks
16  Effective RAM access time (read ) is 205ns (a RAM bank is 4 bytes wide).
17  Effective RAM access time (write) is 130ns (a RAM bank is 4 bytes wide).
18  Clocked at 486 40.1 MHz. Cache ENABLED

Also ran SPEEDTEST for good measure. This is what came that came up with:

After that I thought I run some DOOM benchmarks using Phils CompLab DOSBENCH (highest settings) for fun and since I probably never use this board again for anything useful. These are the results:

1	                    BUS	Realticks	FPS
2Jovian Logic RUBY 451       ISA	13189	    5.663052544
3WD90C00	                    ISA	10111	    7.387004253
4ATI VGA Wonder-16	        ISA	10614	    7.036932353
5ET4000AX	                ISA	7386	    10.11237476
6TIGA VGA+ ET4000AX	        ISA	7388	    10.10963725
7CL-GD5429	                VLB	4559	    16.38297872
8ATI Graphics Wonder	        VLB	4568	    16.35070053
9ET4000/W32I	                VLB	4567	    16.35428071

It runs pretty good on VLB. I assume the CPU is the bottleneck here. I also played a couple of levels of DOOM. It's totally playable on VLB, at least if I was given this board as a kid to play DOOM on I'd be happy enough.

All in all. This board is kinda weird. It seems bottom of the barrel but it has all these power saving options (totally useless in a home setting) so maybe this was more targeted at offices where it would be "on" 24/7. Not sure. The fake cache is weird but all in all it seems to perform OK if you're not a serious 1994 gamer dude.

Oh, this is a dump of the BIOS (which is dated 08/08/93):

evasive wrote on 2021-08-23, 09:40:

All jumpers, connectors and whatnot line up. @pano69: can we have a bios dump? Maybe we can find out something for this KDN1401 marking.

I missed your comment re. KDN1401 marking originally.

The board seems to come in two main flavors of clock generators. Either with KDN-1401/KDN-1402C or the MX8315. The board I posted here is the KDN-1402C variant.

There is a oscillator between the ISA slots, next to it sits a chip marked KDN-1402C.

Just a thought: if your board is like one of the PcChips that had both fake and real cache versions of same exact board you may need the real cache version BIOS if you put in real cache chips..
Finding it may be tricky

Deksor wrote on 2019-07-25, 09:17:

Actually the M919 is the only case I know of fake cache with traces leading to nowhere. If you're really not sure if the mobo supports cache, take a multimeter and check if there is a voltage between where the VCC and GND pins are supposed to be. If there's one, then making fake wires with some getting power would be very strange isn't it ?
If you don't have a desoldering iron, you can simply cut the chips off and then remove each pin using a soldering iron on one side and a pair of pliers on the other side.
I have now two non PCChips boards that have/had fake cache. I added real cache on one of those and it worked without any problems, so did I with my PCChips m915. The last one has a UMC chipset and is quite fast so I will add cache to it as well.

Yep ! One of the PCChips required the real cache version BIOS or did not enable it even after replacing the cache but cannot remember which and cannot find it (and being AMI was a PITA to fix iirc)...

Well for pcchips boards I don't think there's any of that, they seem to detect when cache is present or not and will actually report a different message : "cache enabled" when there's no cache and "xxxkb cache detected" when there's actually cache.
Same happened on non pcchips boards.

I think that this (wrong?) Information comes from here https://www.redhill.net.au/b/b-bad.html

The problem OP has is different. Unless I'm mistaking, they did NOT swap the fake cache for real cache yet. But the board doesn't post

Edit : bios added here https://www.ultimateretro.net/en/motherboards/6238
Thank you !

There is another thread on this board, so adding a link here in case someone else with this board stumbles upon this.

Help identify 486 VLB motherboard