Someone ought to reverse engineer those cache sticks. Anybody want to scan or photo one for me so I can look into making clones of them?

You'd simply have to look for a PCChips Cache Stick, good luck, I doubt there are many out there. This was like their middle finger response to people criticizing them for putting plastic things on and claiming they were cache.

HighTreason wrote:

Someone ought to reverse engineer those cache sticks. Anybody want to scan one for me so I can look into making clones of them.

You'd simply have to look for a PCChips Cache Stick, good luck, I doubt there are many out there. This was like their middle finger response to people criticizing them for putting plastic things on and claiming they were cache.

Yeah, you're right. I've been reading more, and found they're very rare. Someone on ebay wants $150 for one 🤣 But it looks like the board will run just fine without cache, so I'll go that route.

Yeah, it will run without it. Just make sure you turn "External Cache" off in the BIOS, otherwise it might have stability issues.

Its PCChips... he might run into stability issues regardless.

RIP my old Cyrix MII 333 I had way back in the mid-late 90s. =(

Hey! I'll have you know PCChips were the fore-runners of great systems like that MSI Celeron that was mentioned yesterday. They had a reputation, you could always rely on PCChips products (to crash) 😁

I wonder if there's anyone who hasn't had a PCChips board at some point. I thought I had none in service anymore, only to discover my Pine PT-7502 is a re-badged PCChips board (Apparently M920, though that comes in as an ATX Board so who knows). Surprisingly, it's the fastest Triton II board I have, so whether Pine (Essentially XFX today) did something to it or they were just a board that was actually good I do not know. Maybe the information is incorrect, but it does resemble something PCChips would make and the Pine model number is just a plastic transfer with gold writing slapped on the edge of the board in a lopsided manner.

Given some of what I've seen, I even wonder if they made their own boards, bought designs from other companies or there were common reference designs used by more than one vendor as there are quite a few boards that were identical to ones from other companies, some of them good, such as the Aquarius MB-4DUVC VLB board I have which came from a bout 10 companies in different colors with different numbers and varying component quality, they all had holes for the different components each company had. It was identical to the JK-042A it replaced aside from it's build quality being better, the JK labelled one had the VX.X silkscreen like PCChips ones. Was actually a good board, the only thing that killed it was the barrel battery bursting and getting between they layers of the PCB so I couldn't clean it out, and it still took ten years to eat anything important.

I have heard the M919 is actually a decent board though, apparently they do work and support a wide array of CPUs. I think they are one of the few boards that run the Cx5x86 reasonably well, but I've never owned one so can't say for sure.

They were VERY hit or miss. I remember seeing a review praising one of their Socket7 (or SS7, I don't remember) boards... only for the site to publish another article a few months later warning that a new revision of the board (same model, but V2.0, looking identical other than the revision number) came out, and it sucked, and that 7 out of 10 (or something like that) they tested had a stability issues.

Iam not sure, but this would be an 486 cache stick:

http://www.ebay.com/itm/120481786035?_trksid= … K%3AMEBIDX%3AIT

Coast sticks using a different kind of memory..

I think they called Quad package memory.

https://ancientelectronics.files.wordpress.co … /win31cache.jpg

I think it's not worth it, the board being cheaper than the cache module.

Thanks for that link, Robin. I'm not sure on that stick, either. Perhaps someone else can chime in?

In any case, I'll give the board a go without cache. It's supposed to be able to support 128MB ram, so I suppose I'll max it out and hope performance is decent (for a 486 of course).

AllUrBaseRBelong2Us wrote:

Thanks for that link, Robin. I'm not sure on that stick, either. Perhaps someone else can chime in?

In any case, I'll give the board a go without cache. It's supposed to be able to support 128MB ram, so I suppose I'll max it out and hope performance is decent (for a 486 of course).

No, it won't be decent for a 486. On the bright side, your board does not have fake chips, it just does not have the chips that are usually fake at all.

I have an m919 board and its actually a half decent board. It certainly has its quirks but once you figure it out its not bad. I run a Cyrix 5x86 120mhz in mine and its pretty speedy. I defiantly recommend the 256K cache stick if you can find one. I paid $45 for mine maybe 2 years ago off the vintage pc forums. Took me awhile to figure out though my board only recognized it after I swapped the edo ram out for fpm. The cache sticks have printed on them something like "for use in m919 board only"

https://ancientelectronics.files.wordpress.co … /win31cache.jpg

I doubt they will be cheap.

The board looks like it would be relatively easy to copy... Anyone dare me to try? I'm extremely tempted to get an M919 and attempt this feat.

I'm guessing nobody is willing to throw one in a scanner without it's paper labels though, so I'll only have that blurry photo and one with labels to go on, meaning I'm potentially going to fry an M919 down the line - ah well.

Edit: Given that those are just normal 32K Static RAM chips, it would also be theoretically possible to alter the design to take normal cache chips I think. Pinout looks the same, but voltage is different. Could always add a header for a floppy / molex connector or something I guess.

Edit 2: Yeah, that's a four-layer board, but the two middle layers are basically a full sheet of copper, one is Ground, one is 3.3V DC. I can skip those then and make the board cheaper... May end up using a few bodge wires here and there though. I can almost entirely omit those middle layers from the copying process as I will be using 5V chips in my copy, meaning that the chips can't be connected to the power pins in the slot anyway. Unclear as yet whether they used the power or ground plane for signalling. One has to wonder if some of the bridged pins are actually acting as jumpers to set the amount of cache... Perhaps if one could figure that out, it might be possible to add a 128K or 512K cache too.

My only concern would be whether 5V chips would work properly with LVTTL signals.

If your serious I could always take some better images of it but id have to dig that PC out.

I'm serious that I want to attempt reverse engineering, but as I cannot promise my attempts will be successful - and getting my own M919 to test any board I fabricate would be some time away - I don't want to have you drag a machine out if it's an inconvenience. The main problem I face is that I cannot see some of the traces as they are obscured by ICs. A large chunk of them are easy as they connect in parallel anyway, but the tag ram and the IO pins are currently unknown.

I'll just work with what I have for now. I might even look for a module myself at a later time so I can take a multimeter to it and figure out where the traces go for certain.

A 512k cache stick would be nice. It should support it. I think I remember reading some documentation that a 512k stick was supposed to be produced but never was. There's no excuse for the fake cache on some boards (like mine) but I think that's where it gets most of its bad reputation. At least you have the option of the cache stick, despite its relative rarity. It does have its quirks like it said. it can be picky about RAM and mine will not detect any L2 cache if EDO is installed. also at 40mhz bus the PCI slots get underclocked to 27mhz or some nonsense. There is a way around this. I jumpered my turbo button to the 33mhz jumper so if I boot the system at 33mhz bus then hit the turbo button to kick into 40mhz bus to get around this though not the PCI slots are out of spec but thus far I haven't had any issues. Its also the only board I know of that supports some of the Cyrix 5x86 advanced features in BIOS.

Here's mine. I think it's a M919 motherboard. Sorry I can't remove the stickers but hope the photos and the scan from the manual should give you some sort of clue.

Interesting. that certainly appears to be a m919 board but I've never seen a cache stick for them that didn't have the warning printed on the back. maybe the sticks with the warning are later productions.

Apparently it claims it supports a upto a 1024KB cache module! I very much doubt they ever got released!
Here's the spec from my manual

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.