Very interesting, but way too old unfortunately ...

red-ray wrote on 2020-05-12, 22:00:

https://usermanual.wiki/Document/UMCDataBook1986.1234633335

This databook is for the older UMC stuff.We are talking here about the UMC chipset from the middle of the 90's.

Still there are the ctchip config files that provide interesting data. Not a complete datasheet, but it seems that most populated registers are described (if this is what you are looking for).

wiretap wrote on 2019-08-27, 20:03:

They are design books from the mid 1990's.. Like 2 dozen of them, each 500+ pages. I don't know about scanning them because they still might be under NDA or patent/proprietary information. I have them at work, but don't have any way to scan that many pages anyhow. I'll have to try to find them, in our huge library of microprocessor books.

I'm looking for that elusive datasheet too, also the UM8498/8496 among others. Please if you can get a hold of it, i'd be willing to scan it. Spent ages looking everywhere for that datasheet, from my searching it is not on the net.

===
FWIW, I have extracted the major details out of the file on bitsavers for the other UMC 486 chipsets, perhaps there are a few others:
http://66.113.161.23/~mR_Slug/chipset/chipset … 1=33131,2#29352

Haven't been able to find it.. we moved offices to a different building, and our engineering library of books was moved into multiple cabinets across 3 offices (each office is like 25k square feet, 2 floors).. I'll come across them sooner or later, I just don't have the time to look through the several hundred locked cabinets they could be in.

CuPid wrote on 2020-05-13, 09:18:

Still there are the ctchip config files that provide interesting data. Not a complete datasheet, but it seems that most populated registers are described (if this is what you are looking for).

Those are useful, but what would also be really nice are the pinouts.

Anonymous Coward wrote on 2020-05-13, 12:42:

CuPid wrote on 2020-05-13, 09:18:

Still there are the ctchip config files that provide interesting data. Not a complete datasheet, but it seems that most populated registers are described (if this is what you are looking for).

Those are useful, but what would also be really nice are the pinouts.

I remember I have got a schematic manual for UMC based laptop Mitac 5023. I'll upload it here.

Here it is

Roman555 wrote on 2020-05-14, 11:20:

Anonymous Coward wrote on 2020-05-13, 12:42:

CuPid wrote on 2020-05-13, 09:18:

Still there are the ctchip config files that provide interesting data. Not a complete datasheet, but it seems that most populated registers are described (if this is what you are looking for).

Those are useful, but what would also be really nice are the pinouts.

I remember I have got a schematic manual for UMC based laptop Mitac 5023. I'll upload it here.

Here it is

Thanks,
interesting is, that even it's for the Socket 5 Pentium system, the PCI/ISA Bridge is still UM8886, so atleats we have the pinout 😀
For me the most important are also the registers besides pinout, but I believe they are partly the same with it's predecessor.

Amazing. Thanks!

This is nice, thänk you ~

red-ray wrote on 2020-05-12, 22:00:

https://usermanual.wiki/Document/UMCDataBook1986.1234633335

Well, actually that's a copy of the Bitsavers (the source of the scan) databook I linked to earlier.

wiretap wrote on 2020-05-13, 12:13:

Haven't been able to find it.. we moved offices to a different building, and our engineering library of books was moved into multiple cabinets across 3 offices (each office is like 25k square feet, 2 floors).. I'll come across them sooner or later, I just don't have the time to look through the several hundred locked cabinets they could be in.

I'm a little worried you may be our only hope. Fingers crossed that someday you have the time to locate it and then figure out how to navigate the NDA/trade secrets morass you mentioned earlier. Not a good situation to be sure, but here's hoping. 😀

In the meantime we'll hope that another databook shows up elsewhere out of the blue, but Bitsavers hasn't made any new acquisitions on that front recently either...

Yes. UMC stuff are particularly hard to find.

I'd be also interested in early ALI chipsets. The same story....

mpe wrote on 2020-05-15, 09:34:

Yes. UMC stuff are particularly hard to find.

I'd be also interested in early ALI chipsets. The same story....

I suppose Mitac 4024 , Mitac 4023 (Chaplet Halikan NBD-486) based on ALI M1219. Is it early enough?
There were a service manual for Mitac 4024 but now it's dificult to find.
http://web.archive.org/web/20010122102100/htt … /service/Tr.htm

Roman555 wrote on 2020-05-14, 11:20:

Anonymous Coward wrote on 2020-05-13, 12:42:

CuPid wrote on 2020-05-13, 09:18:

Still there are the ctchip config files that provide interesting data. Not a complete datasheet, but it seems that most populated registers are described (if this is what you are looking for).

Those are useful, but what would also be really nice are the pinouts.

I remember I have got a schematic manual for UMC based laptop Mitac 5023. I'll upload it here.

Here it is

tiseno100 cut out the UMC UM8886N pinouts pages from these schematics, I've attached it separately here.

Great ! those may be very useful! All together makes a decent UM8886 datasheet....
Archive org has the 1990-91 UMC databook but it is not new enough....
(note: I re-added it in May because the original archive from a few years ago had issues...https://archive.org/details/umc-computer-ics- … -book-1990-1991)....

Horun wrote on 2021-08-27, 23:51:

Great ! those may be very useful! All together makes a decent UM8886 datasheet....
Archive org has the 1990-91 UMC databook but it is not new enough....
(note: I re-added it in May because the original archive from a few years ago had issues...https://archive.org/details/umc-computer-ics- … -book-1990-1991)....

Eh, still not enough for a full decent datasheet, but it's getting there. Combine that with info from PCem or 86box or their forks that they've reverse-engineered about the chipset and we may be talking...

WRT your re-addition... Cool. Definitely still something a little weird with this 1990-1991 databook scan, ex. pg. 38: https://archive.org/details/umc-computer-ics- … ge/n37/mode/2up
Feels like it may have once been hosted by Bitsavers but later rejected. I mean, who else would scan a full UMC databook? Sparingly few people out there in Internet-land doing that type of work.
[EDIT] Looks like I am wrong though, I checked Bitsavers on the Wayback Machine back to 2011-ish and it was never there...

BTW, the Mitac schematics previously added to this thread also contain pinouts and their descriptions for UMC UM8891 (UM8891N) and UM8892 (UM8892N).

I found that doc long ago and re-found it in May after reading it was removed from Archive .org while digging thru other archives for ITE docs.
Unfortunately I did not save the location where I found it but think it was from a University linked thru one of UMC's subsideraries.
All I saved with it was a text with this:
"United Microelectronics Corporation, Taiwan. http://www.umc.com.tw/ http://www.umc.com (after 2000) http://www.umcgroup.com/
note: 1996: Spun off its IC design units to establish MediaTek, Novatek, ITE Technology, https://www.ite.com.tw/ Faraday Technology, AMIC Technology, and Davicom
"ITE’s strategy of developing partnerships with industry leaders led to its acquisition of the Computer Products Division of Taiwan-based United Microelectronics Corporation (UMC) in 1996."
And a few archive.org links to nothing special but there could have been other UMC docs at that site too:(

Horun wrote on 2021-08-28, 21:05:

I found that doc long ago and re-found it in May after reading it was removed from Archive .org while digging thru other archives for ITE docs.
Unfortunately I did not save the location where I found it

I figured out where it originated, more or less. I shouldn't really link to either site though...

Stiletto wrote on 2021-08-30, 03:23:

Horun wrote on 2021-08-28, 21:05:

I found that doc long ago and re-found it in May after reading it was removed from Archive .org while digging thru other archives for ITE docs.
Unfortunately I did not save the location where I found it

I figured out where it originated, more or less. I shouldn't really link to either site though...

Yeah if it has three unrelated numbers as part of the file name then maybe that is where I re-found it ;p

I took a time to map most of the pins of UM8881F ETA to its peripheries.

A recheck for CPU control signals and specially for cache would be appreciated (my board ATC-1415 has only one cache bank so I could not find control signals for the second one, also it has only 2 SIMM slots). Some CPU signals are routed through jumpers, so I could miss few.

There is also possibility of a separate cache dirty bit (8+1 configuration), but I didn't see any board (on retroweb) with separated dirty and tag chips. It could be routed to cache on stick used by pcchips boards.

Exact signal mapping for tag cache data (especially dirty bit) could be probably measured by disabling CPU cache and reading/writing a memory address pattern. But from hardware point it doesn't matter which bit is routed to which data pin of SRAM chip.

Pins 21 and 22 seems to be connected backwards to TAG cache on my board.

If you use 486-/PODP-Pinout and Differences as 486 pinout source, careful I think A16 and A13 are swapped there.

ad mitac schematics: Funny how UMC later reused the same architecture for UM891N, but as pentium 1 has 64bit data bus, they had to exclude the data pins to a new chip 😁 (the rest of the signals seems to be almost identical). Gonna update the names in my datasheet to match the namespace used in mitac schematics later.