• Please review our updated Terms and Rules here

Just found out I bought a fake cache 486 board

U

Uniballer

Experienced Member
Joined
Apr 3, 2014
Messages
448
Location
USA
My junk bin had a working 486 VLB board labeled "SIS 486G 3.3/5V Ver: H", with SiS 85C471 chipset and an Award BIOS. I ran NetBSD (0.9?) on it with a AM486DX2/80 CPU when the board was new-ish. It has 9 AA26257AK-15 chips soldered in (8 that don't work for cached data, and 1 that doesn't work for the cache tag data). There is no manufacturer's logo on any of these chips, but the date code is 9518. I confirmed that the cache does not work with the CACHECHK.EXE program. There is no BIOS option to disable the external cache, but it does claim to be able to set it for write-through or write-back. The BIOS ID string is "08/18/95-SIS-85C471B/E/G-2C4I9F32-00". Looking up this ID gives the manufacturer "Full Yes" (whoever they were). Oh yeah, the other problem with this board/BIOS is that the BIOS always reports having 255MB of RAM, which kills Memtest86, later versions of Unix, etc.

Anybody else confess to being a victim of this scam?
 
Last edited:
Back in the day, we sold a bazillion fake cache boards (I worked for a Chinese copy)
I was too swamped to verify what suspected until I gave a 2wk notice.. Then I had lots of time on my hands for 2 weeks.. Sorry you got one of those.
 
Don't feel bad about it. I ran a web server on it. It was fast enough for that, and the board paid for itself many times over. I'm sure I would have been angry about it back then, but I was madder about the Maxtor 4GB SCSI drive that crashed on that box. Now the fake cache seems kind of funny.
 
I had one of those--it was cheap, but reliable. Performance wasn't important. So, not such a bad thing.

I still have a couple of Ampro 8600 P1 boards--they work just fine for their application. I was much more upset with the 90's boards with bad caps.
 
Chuck(G) said:
I had one of those--it was cheap, but reliable. Performance wasn't important. So, not such a bad thing.

I still have a couple of Ampro 8600 P1 boards--they work just fine for their application. I was much more upset with the 90's THROUGH CURRENT boards with bad caps.
Click to expand...
There I fixed it for you ;-)
I've seen so many cap failures right up through modern core2duo's (in HP/CPQ, Dell, and Lenovo systems no less). Pretty well expect to start seeing cap failures in our i3 and i5's soon at this rate :-/

On topic, I had a fake-cache board back in the day too, while slower, they were typically fairly stable :sneaky:
 
Chuck(G) said:
I still have a couple of Ampro 8600 P1 boards--they work just fine for their application. I was much more upset with the 90's boards with bad caps.
Click to expand...

I still have a Shuttle HOT-553 with a Pentium-100 (P54C) with 128 MB 72-pin EDO RAM running 24/7. It runs an environmental control application under FreeBSD 6.4 for my kennel - basically a networked thermostat/hygrostat. It doesn't take much horsepower but has to run all the time.

I have recapped some motherboards and LCD monitors myself. And I'm keeping my eye on the half dozen Dell Optiplex 745's at work that have already had a few caps replaced before I got them (no big problems, yet). Fortunately, all of my i5 stuff seems to be strictly polymer capacitors.
 
Last edited:
RWallmow said:
There I fixed it for you ;-)
I've seen so many cap failures right up through modern core2duo's (in HP/CPQ, Dell, and Lenovo systems no less). Pretty well expect to start seeing cap failures in our i3 and i5's soon at this rate :-/

On topic, I had a fake-cache board back in the day too, while slower, they were typically fairly stable :sneaky:
Click to expand...

Because all those boards likely were lower end. I know for a fact that the Optiplex 745/755 use wet electrolytic caps. The higher end stuff is all solid polymer caps now and as the price has come down, it has appeared in mid-range stuff as well.
 
njroadfan said:
Because all those boards likely were lower end. I know for a fact that the Optiplex 745/755 use wet electrolytic caps. The higher end stuff is all solid polymer caps now and as the price has come down, it has appeared in mid-range stuff as well.
Click to expand...

Even the Optiplex 745 boards have some polymers, especially near the CPU chip (under the cooling assembly) where the heat would cook the wet aluminum electrolytics too fast. I'm sure this is just an adjustment based on experience with the previous generation, and the dropping costs of the polymer caps.
 
njroadfan said:
Because all those boards likely were lower end. I know for a fact that the Optiplex 745/755 use wet electrolytic caps. The higher end stuff is all solid polymer caps now and as the price has come down, it has appeared in mid-range stuff as well.
Click to expand...

No, these were all higher end enterprise machines, not retail crap. I will admit not many Core2's mobo's have succumb to caps (though a few have), but quite a lot up through the pentium 4 era sure did, so the problem sure goes well past the 90's like implied.

We have lost HUNDREDS of lenovo core2 and i3 system power supplies to blown caps though, they did own up to that one and shipped us several cases of revised power supplies (from different vendor, IIRC the duds were ACBel made). EDIT: for clarification, these were SFF systems, and power supplies are external bricks, so just a matter of swapping bricks.
 
Last edited:
There is also some just plain terrible thermal design involved as well. I've worked on an LCD monitor that had electrolytic caps hot-glued to a heatsink...
 
My experience with the fake cache and other cheapo PC Chips 486 / Socket 7 era boards is that they usually work OK, but the motherboard is incredibly thin and can flex enough when inserting cards that even though it might look like the card is properly seated, it is not making good enough contact, and you'll end up with a flakey or non-booting system.

VLB cards are especially problematic due to the large number of contacts across a wide area. You may even need to put in additional props under the motherboard (like rubber feet glued to the bottom of the case) to keep the board from flexing too much when you put in the cards!
 
The onboard L2 cache of the 486 had a very bad price-performance ratio. These boards were sold cheap and I can't say that they were not worthy of their price. Their mistake was probably the way they were doing it, which was an obvious hoax. They could just sell boards without L2 cache and just say that the juice is not worth the squeeze for adding L2 in low-cost systems. They didn't need to hide the fact that they weren't using branded chipsets either. If one wants a top-performing board, he would look elsewhere anyhow.
 
orion24 said:
The onboard L2 cache of the 486 had a very bad price-performance ratio.
Click to expand...
I'm kinda surprised that you think so. My testing with an AM486DX2/80 showed that the L2 cache had a much bigger effect on performance than going from 66 to 80 MHz, almost a factor of 2 as compared to less than a 10% improvement for a 20% clock rate increase. My performance test was compiling and linking a bit of code that takes about 14 minutes to build on a Pentium-100 under FreeBSD 6.4 (as opposed to measuring graphics performance, or filesystem bandwidth, for example). For perspective, that test only takes about two seconds on an i5-2500 if you use all 4 cores.
 
Last edited:
smeezekitty said:
Then why not make boards with real cache?
Click to expand...
cash for cache!

Cache chips are expensive! If they can fool a few buyers into thinking its got cache and paying the price for cache, but not delivering the cache, they can line their pockets with more cash $$ ;-)
 
I saw something on the web about boards apparently designed just for fake cache, but my board appears to be a workable board with fake SRAM chips, key jumpers soldered onto the board (i.e. no headers for you!), and a BIOS that tries to hide the deception by reporting the presence of non-existent cache. Maybe even the incorrect reporting of RAM size was intended to prevent certain diagnostic programs from working right. I think in theory that this board could have real cache installed and enabled but it would mean a lot of work, as well as finding a correct BIOS... Not gonna happen.
 
Uniballer said:
I saw something on the web about boards apparently designed just for fake cache, but my board appears to be a workable board with fake SRAM chips, key jumpers soldered onto the board (i.e. no headers for you!), and a BIOS that tries to hide the deception by reporting the presence of non-existent cache. Maybe even the incorrect reporting of RAM size was intended to prevent certain diagnostic programs from working right. I think in theory that this board could have real cache installed and enabled but it would mean a lot of work, as well as finding a correct BIOS... Not gonna happen.
Click to expand...

PC Chips hacked the BIOS to say "Write Back Cache On" during POST when, in fact, no L2 cache was installed. They also soldered the BIOS chip to the motherboard to prevent people from replacing it with a non-hacked BIOS.

Amusingly, PC Chips' Socket 7 Pentium/6x86/K6 boards still say "Write Back Cache On" when you disable the L2 cache, even though (AFAIK) they all came with real L2 cache onboard. (These boards came with a so-called "VX Pro" chipset that had nothing to do with the real Intel VX chipset.)
 
You must log in or register to reply here.
Back
Top