VOGONS


external l2 cache issue - 486 socket 3 mobo

Topic actions

First post, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie

hello all

after updating the bios in my 486 build the system hangs on updating escd and the keyboard stops working or freezes

i have discovered - it doesn't hang if disable the external l2 cache in bios or take out the chips

the l2 cache worked fine before the bios update which is on an eprom (hoping replacing it with a flash rom to it original bios at some point as my mobo has a jumper for eprom or flash rom)

the mobo is atrend atc 1415

if i am needing to replace l2 cache chips and if so where is good place to find and buy them

the manual say my mobo support any

32kx8 - 128kb
64kx8 - 256kb
128kx8 - 512kb

the tag ram uses
8kx8 - 128k
16kx8 or 32kx8 - 256kb
32kx8 - 512kb

pic of current l2 cache chip below

Last edited by Robhalfordfan on 2020-10-23, 19:11. Edited 2 times in total.

Reply 1 of 27, by SScorpio

User metadata
Rank Member
Rank
Member

When you flash a BIOS, settings can be restored to defaults. The -15 in the name can mean 15ns, go through the BIOS and see if there is anywhere to set that. It also looks like you have 256KB of cache as you have 4x512Kb and the 256Kb may be a parity chip.

Reply 2 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
SScorpio wrote on 2020-10-23, 19:10:

When you flash a BIOS, settings can be restored to defaults. The -15 in the name can mean 15ns, go through the BIOS and see if there is anywhere to set that. It also looks like you have 256KB of cache as you have 4x512Kb and the 256Kb may be a parity chip.

had a look and defaulted the bios and can not see anything about ns timings

here pics of what settings the bios has

Reply 3 of 27, by cyclone3d

User metadata
Rank l33t++
Rank
l33t++

What BIOS version did you come from and go to?

Are you sure the BIOS you upgraded to is not for a different revision of the motherboard?

Are the cache jumpers set properly?

Yamaha modified setupds and drivers
Yamaha XG repository
YMF7x4 Guide
Aopen AW744L II SB-LINK

Reply 4 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
cyclone3d wrote on 2020-10-23, 19:36:

What BIOS version did you come from and go to?

Are you sure the BIOS you upgraded to is not for a different revision of the motherboard?

Are the cache jumpers set properly?

the cache jumper set when i got the mobo original and worked before and haven't changed anything there

i can't remember what version the original bios was but pic below with label on it
file.php?mode=view&id=94757

the version i updated was
file.php?mode=view&id=94758

i didn't find any other revision of my mobo and went to a website i use before when updating my socket 7 penitum build a while and seem to be ok

i have attached pic and the rom file of both bios

original bios - i backed up via awdflash before i buggered the eprom
the updated bios - i used to flash to new eprom via a programmer

Reply 5 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie

been looking for my original bios just in case if my backup is screw in some way and found this and it seen to be closer or was my original bios

http://chukaev.ru54.com/bios_en.htm#z3zzm002: … 532c394f0ca82f0

looking at the ic chips in this link and my l2 cache chips matches, along with chipset, along with via chip (VT8235N) and the odin chip (OEC12C885)

Reply 6 of 27, by Horun

User metadata
Rank l33t++
Rank
l33t++

Couple things: Your cache cannot go bad just because you flashed the BIOS. It looks correct for 256K using 4 chip+TAG.
TRY: Set the L2 Cache Update Scheme to WR-Through like L1 is and see what happens. ( I can see that is the other choice in the new BIOS)
Ok I cannot extract the "original.tmp" from your original bios backup but can from the new one so am guessing it was not backed up properly or got corrupted while saving,
even my Award util can only see the NCR portion properly but can see all of the newer BIOS sections properly.

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 7 of 27, by jakethompson1

User metadata
Rank Oldbie
Rank
Oldbie

Can you remember whether the original BIOS had that "Plug and Play BIOS Extension"?
PnP BIOS and ESCD are intertwined, and I believe ESCD requires a flash BIOS. So that may be related since we thought your board isn't capable of flashing a BIOS.

Reply 8 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
jakethompson1 wrote on 2020-10-24, 01:44:

Can you remember whether the original BIOS had that "Plug and Play BIOS Extension"?
PnP BIOS and ESCD are intertwined, and I believe ESCD requires a flash BIOS. So that may be related since we thought your board isn't capable of flashing a BIOS.

I think it did have plug and play bios extension, same version aswell

And I thought my board wasn't capable since it was an EPROM but I noticed the jumper on the mobo and looking in manual that it can be change to a flash rom and though I can't any more harm to mobo, so why not try

i have been using the machine with l2 cache off and testing it out with dx4 100 and noticed the machine is a little unstable, as the system would kinda not response but not lock up as i could still move the mouse and use the keyboard after using for a little while and had one instant where the machine rebooted itself but only once

Last edited by Robhalfordfan on 2020-10-24, 16:01. Edited 5 times in total.

Reply 9 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
Horun wrote on 2020-10-24, 00:42:
Couple things: Your cache cannot go bad just because you flashed the BIOS. It looks correct for 256K using 4 chip+TAG. TRY: Set […]
Show full quote

Couple things: Your cache cannot go bad just because you flashed the BIOS. It looks correct for 256K using 4 chip+TAG.
TRY: Set the L2 Cache Update Scheme to WR-Through like L1 is and see what happens. ( I can see that is the other choice in the new BIOS)
Ok I cannot extract the "original.tmp" from your original bios backup but can from the new one so am guessing it was not backed up properly or got corrupted while saving,
even my Award util can only see the NCR portion properly but can see all of the newer BIOS sections properly.

I did try changing l2 update scheme to write through and nothing changed, same as before

I thought the original backup was maybe it did get corrupted while backing up

But in reply below that is a link to what I think is my original bios (not my backup)

Reply 10 of 27, by amadeus777999

User metadata
Rank Oldbie
Rank
Oldbie

Return to the old BIOS and try the configuration, no cache + DX4, to see if the instability issue persists.
Also the new / different BIOS may not be fully compatible - I had a similar issue with a P60 board where a "compatible" BIOS version would screw up everything.

Getting new cache chips is hairy as most, even the slow ones, are fakes. On ebay sellers from France sometimes carry original chips.

Reply 11 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
amadeus777999 wrote on 2020-10-25, 11:11:

Return to the old BIOS and try the configuration, no cache + DX4, to see if the instability issue persists.
Also the new / different BIOS may not be fully compatible - I had a similar issue with a P60 board where a "compatible" BIOS version would screw up everything.

Getting new cache chips is hairy as most, even the slow ones, are fakes. On ebay sellers from France sometimes carry original chips.

that is what i am thinking and waiting to try a flash rom instead of eprom and hopefully it will "fix" my issues, fingers crossed

i am aware on fake chips on line and original ones are harder or more expenisve to find/buy as i imagine they are not "made" anymore

Reply 12 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie

my flash rom arrived and flash the bios from the weblink above which is v3.20 and with l2 cache on it gets past updating escd... and starting booting from hdd or a floppy disk and say starting ms-dos and then nothing but again turning off l2 cache altogether works fine

changing the l2 cache to write through or write back makes no different

just rule out the cpu (which intel 486 dx4 100), i use a intel dx2 66 which works before no issues and does the same as the dx4

both cpu's are write back

prob a stupid question/idea but is there a way to compare the original bios backup with one of the later ones to see what different or got corrupted in original backup

dont know if this would help but the checksum of original bios backup is 061FH - i think - according to award flash program
if anyone has same board and bios eariler then v3.20, able to make a backup and sent to me

Reply 13 of 27, by amadeus777999

User metadata
Rank Oldbie
Rank
Oldbie

Maybe a simple way to compare Bios's is to generate a hash via the crc/sha functionality that comes with 7zip - right click after install and you can generate "hash" keys from files.

Maybe one of your srams is defect... you may need a second set for testing.

Reply 14 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
amadeus777999 wrote on 2020-10-27, 21:29:

Maybe a simple way to compare Bios's is to generate a hash via the crc/sha functionality that comes with 7zip - right click after install and you can generate "hash" keys from files.

Maybe one of your srams is defect... you may need a second set for testing.

i have no idea how to do that and could try since this all new to me and i am learning and nothing wrong with learning something new during current state of the world

maybe one the sram chip is defected - doesn't explain how before the "bios update" that it all worked fine but that was the breaking point

would any sram chips work as long as it cache size and timing are the same

Reply 15 of 27, by debs3759

User metadata
Rank Oldbie
Rank
Oldbie

Any sram chip with the same specs would work.

See my graphics card database at www.gpuzoo.com
Constantly being worked on. Feel free to message me with any corrections or details of cards you would like me to research and add.

Reply 16 of 27, by Horun

User metadata
Rank l33t++
Rank
l33t++
Robhalfordfan wrote on 2020-10-26, 18:13:
my flash rom arrived and flash the bios from the weblink above which is v3.20 and with l2 cache on it gets past updating escd... […]
Show full quote

my flash rom arrived and flash the bios from the weblink above which is v3.20 and with l2 cache on it gets past updating escd... and starting booting from hdd or a floppy disk and say starting ms-dos and then nothing but again turning off l2 cache altogether works fine

changing the l2 cache to write through or write back makes no different

just rule out the cpu (which intel 486 dx4 100), i use a intel dx2 66 which works before no issues and does the same as the dx4

both cpu's are write back

All the 486DX2-66 were write-through except the SX955 SL version was write-back cache. Similar to DX4-100, all were write-through except the SK096 chips. What are the S spec of the cpu's you are trying to use ?

Hate posting a reply and then have to edit it because it made no sense 😁 First computer was an IBM 3270 workstation with CGA monitor. Stuff: https://archive.org/details/@horun

Reply 17 of 27, by jakethompson1

User metadata
Rank Oldbie
Rank
Oldbie
Robhalfordfan wrote on 2020-10-24, 11:25:

I thought the original backup was maybe it did get corrupted while backing up

How did you dump the backup? The good news for troubleshooting is I'm pretty sure it is just a corrupted version of the V3.20 bios you posted. The compressed original.tmp is stored in the first 0xff49 bytes; only a few bytes in that section differ:

$ cmp -l 4umw002.bin ORI_BIOS.ROM | awk '{ printf "%5x %2x %2x\n", $1, strtonum("0" $2), strtonum("0" $3) }'
1 25 0
2aab b1 11
5556 3f 0
...

If I "repair" those bytes using a hex editor in fact it's possible to extract original.tmp from your ORI_BIOS.ROM, and it indeed has the 3.20 version inside.

This doesn't solve the mystery of why you are having trouble with cache though. You definitely cleared your CMOS with the jumper if you had one, and did "Load Power on defaults" and "Load BIOS Defaults" in setup before you started messing around? As Feipoa has in his MB-8433UUD manual there may be hidden CMOS settings you can't see that change between bios versions, and that you can only reset using the "Load defaults" settings.

There's no way you inadvertently bumped off a cache-related jumper when you were messing with the bios chip?

Reply 18 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
debs3759 wrote on 2020-10-28, 00:34:

Any sram chip with the same specs would work.

ok thank you, was not sure if some were compatible and some was not

Reply 19 of 27, by Robhalfordfan

User metadata
Rank Oldbie
Rank
Oldbie
Horun wrote on 2020-10-28, 02:25:
Robhalfordfan wrote on 2020-10-26, 18:13:
my flash rom arrived and flash the bios from the weblink above which is v3.20 and with l2 cache on it gets past updating escd... […]
Show full quote

my flash rom arrived and flash the bios from the weblink above which is v3.20 and with l2 cache on it gets past updating escd... and starting booting from hdd or a floppy disk and say starting ms-dos and then nothing but again turning off l2 cache altogether works fine

changing the l2 cache to write through or write back makes no different

just rule out the cpu (which intel 486 dx4 100), i use a intel dx2 66 which works before no issues and does the same as the dx4

both cpu's are write back

All the 486DX2-66 were write-through except the SX955 SL version was write-back cache. Similar to DX4-100, all were write-through except the SK096 chips. What are the S spec of the cpu's you are trying to use ?

my dx2-66 is sx955
my dx4-100 is sk101