VOGONS

Search

Common searches

isa 386c cache apparently not working [now fixed]

Topic actions

First post, by ratfink

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

Continuing a debate starting in the 3d benchmark thread.

I have an Asus ISA386C motherboard with 2mb ram and a 386dx33.

The 3dbench results are well below what others report for this processor, including for the same board. I get 7.2 up to 7.4 compared to what looks like an expected result of somewhere around 12-15.

I am using a Tseng ET4000AX ISA graphics card.

The bios reports 64kb cache memory, and external cache is enabled. There are no jumpers relating to cache - the board has only 4 jumpers and these serve other purposes. The manual confirms this.

cachechk however reports there appears to be no cache.

There are a couple of socketed chips near the cache, which do not appear on the diagrams in the manual. There are the same sort of size as the cache chips and have some stickers on them including the word "64kb" as I recall. They don't seem to want to seat properly - I've had the system apart lately to rebuild it into a different case, and pushing these chips down results in some movement and a crunch or click. Done that twice now, made no difference. Weird they don't appear to be cache chips [as in: the manual says specifically something like "the cache is 16 chips at location X on the board"].

Any ideas?

Last edited by ratfink on 2011-03-22, 13:05. Edited 1 time in total.

Reply 1 of 30, by Amigaz

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

Please let me know if you want me to check any of my settings on my mobo since I have same setup as you.

My retro computer stuff: https://lychee.jjserver.net/#16136303902327

Reply 2 of 30, by ratfink

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

Thanks Amigaz. If you could run cachechk and post the results, that would help confirm whether cachechk is reporting right for this board. If it shows your board has cache then I know for sure that mine has some kind of problem with cache.

Reply 3 of 30, by Amigaz

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie
ratfink wrote:

Thanks Amigaz. If you could run cachechk and post the results, that would help confirm whether cachechk is reporting right for this board. If it shows your board has cache then I know for sure that mine has some kind of problem with cache.

Sure, I'll post my results when I'm back from slavery....I mean work! 😁

My retro computer stuff: https://lychee.jjserver.net/#16136303902327

Reply 5 of 30, by ratfink

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

Ok thanks, guess I need to pull my motherboard out and have a closer look at it. Maybe I should extract those socketed chips and reinsert them to improve the contact, can't quite see how it's laid out in this case but from stason.org the board has the 16 cache chips and then a few tag chips, I reckon these are they. If they are loose or otherwise not making proper contact the cache won't work. 🤣

Reply 6 of 30, by ratfink

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

Clutching at straws here... aka idiot question but... I noticed that the board defaults to "memory parity error check enabled" and I have to disable this or it won't boot. I'm just using some old ram sticks I had that happen to fit.

Could it be that using non-parity ram is causing this problem with the cache?

Having spent lunchtime coaxing a GUS to work I'm ready to believe anything is possible with such old hardware.

Reply 7 of 30, by Mau1wurf1977

Posted on 
User metadata
Rank l33t++
Rank
l33t++

Maybe someone with a 386 could test what performance drop they get once the onboard cache is disabled?

Parity memory would not affect the function of Cache.

Also if you remove all the Cache chips, does the POST message change?

Last edited by Mau1wurf1977 on 2011-02-09, 19:55. Edited 1 time in total.

Reply 9 of 30, by ratfink

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

The cache is soldered on.

Cachechk results are these [copied by hand so some abbrevaitions here]:

Megabyte # ------ Memory Access Block Size (kb) ------
1 2 4 8 <etc>
0: 73 <repeated>
1: 73 <repeated>
2 <----- same as above

Extra tests -----
Wrt 40 <repeated> 41 40 <repeated>
This machine does not seem to have any cache.
Main memory speed - 15.1 MB/s 69.7 ns/byte (100%) [read] 8.7 clks
Effective RAM access time is [read] 139 ns....
Effective RAM access time is [write] 76 ns....
Clocked at 386 32.7mhz

Reply 10 of 30, by Mau1wurf1977

Posted on 
User metadata
Rank l33t++
Rank
l33t++

Oh I see. Maybe Amigaz can disable it in BIOS and see if he gets the same slowdown?

Then at least you would know if it's cache related or not.

My website with reviews, demos, drivers, tutorials and more...
My YouTube channel

Reply 11 of 30, by TheLazy1

Posted on 
User metadata
Rank Member
Rank
Member

Various cachechk runs, settings forgotten:

 CACHECHK V7 11/23/98  Copyright (c) 1995-98 by Ray Van Tassle. (-h for help)
 CMOS reports: conv_mem= 640K, ext_mem= 7,168K, Total RAM= 7,808K
 386 Clocked at 39.0 MHz
 Reading from memory.  
 MegaByte#:      --------- Memory Access Block sizes (KB)-----
       1    2    4    8   16   32   64  128  256  512 1024 2048 4096 <-- KB
 0:   39   39   39   39   39   39   39   39   89   89   --   --   --    us/KB
 1:   39   39   39   39   39   39   39   39   89   89   89   89   88    us/KB
 2  <--- same as above.
 3:   39   39   39   39   39   39   39   39   89   89   89   87   88    us/KB
 4:   39   39   39   39   39   39   39   39   89   89   89   89   89    us/KB
 5  6  7  <--- same as above.

 Extra tests----
 Wrt   33   33   33   33   33   33   33   33   33   34   33   34   34<-Writing
 This machine seems to have one cache!? [reading]
    !! cache is  128KB--  27.9 MB/s  37.6 ns/byte (225%)         
 >>>> If you think you do have L2 cache, you might have FAKE CACHE chips! <<<<
    5.6 clks
    Main memory speed --  12.4 MB/s  84.5 ns/byte (100%) [reading]  12.6 clks
    Effective RAM access time (read ) is 169ns (a RAM bank is 2 bytes wide). 
    Effective RAM access time (write) is  63ns (a RAM bank is 2 bytes wide). 
    386 Clocked at 39.0 MHz.  Cache ENABLED.
 Options: -t0
 CACHECHK V7 11/23/98  Copyright (c) 1995-98 by Ray Van Tassle. (-h for help)
 CMOS reports: conv_mem= 640K, ext_mem= 7,168K, Total RAM= 7,808K
 386 Clocked at 39.0 MHz
 Reading from memory.  
 MegaByte#:      --------- Memory Access Block sizes (KB)-----
       1    2    4    8   16   32   64  128  256  512 1024 2048 4096 <-- KB
 0:   61   61   60   60   60   60   60   60   60   60   --   --   --    us/KB
 1:   61   61   60   60   60   60   60   60   60   60   60   60   60    us/KB
 2:   61   61   60   60   60   60   60   60   60   60   60   60   60    us/KB
 3:   61   61   60   60   60   60   60   60   60   60   60   60   60    us/KB
 4:   60   60   60   60   60   60   60   60   60   60   60   60   60    us/KB
 5  6  7  <--- same as above.

 Extra tests----
 Wrt   34   34   34   34   33   33   33   33   33   33   34   34   34<-Writing
 This machine does not seem to have any cache.
  Hmmm. It looks like megabyte #1 isn't being cached!
  Hmmm. It looks like megabyte #2 isn't being cached!
  Hmmm. It looks like megabyte #3 isn't being cached!
  Hmmm. It looks like megabyte #4 isn't being cached!
  Hmmm. It looks like megabyte #5 isn't being cached!
  Hmmm. It looks like megabyte #6 isn't being cached!
  Hmmm. It looks like megabyte #7 isn't being cached!
  Hmmm. It looks like RAM at megabyte #1 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #2 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #3 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #4 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #5 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #6 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #7 is REALLY slow!
    Main memory speed --  18.2 MB/s  57.5 ns/byte (100%) [reading]   8.6 clks
    Effective RAM access time (read ) is 115ns (a RAM bank is 2 bytes wide). 
    Effective RAM access time (write) is  64ns (a RAM bank is 2 bytes wide). 
    386 Clocked at 39.0 MHz.  Cache ENABLED.
 Options: -t0
CACHECHK V7 11/23/98  Copyright (c) 1995-98 by Ray Van Tassle. (-h for help)
 CMOS reports: conv_mem= 640K, ext_mem= 7,168K, Total RAM= 7,808K
 386 Clocked at 39.0 MHz
 Reading from memory.  
 MegaByte#:      --------- Memory Access Block sizes (KB)-----
       1    2    4    8   16   32   64  128  256  512 1024 2048 4096 <-- KB
 0:   61   61   60   60   60   60   60   60   60   60   --   --   --    us/KB
 1:   61   61   60   60   60   60   60   60   60   60   60   60   60    us/KB
 2  3  <--- same as above.
 4:   60   60   60   60   60   60   60   60   60   60   60   60   60    us/KB
 5  6  7  <--- same as above.

 Extra tests----
 Wrt   34   34   34   33   34   33   33   33   33   34   34   34   34<-Writing
 This machine does not seem to have any cache.
  Hmmm. It looks like megabyte #1 isn't being cached!
  Hmmm. It looks like megabyte #2 isn't being cached!
  Hmmm. It looks like megabyte #3 isn't being cached!
  Hmmm. It looks like megabyte #4 isn't being cached!
  Hmmm. It looks like megabyte #5 isn't being cached!
  Hmmm. It looks like megabyte #6 isn't being cached!
  Hmmm. It looks like megabyte #7 isn't being cached!
  Hmmm. It looks like RAM at megabyte #1 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #2 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #3 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #4 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #5 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #6 is REALLY slow!
  Hmmm. It looks like RAM at megabyte #7 is REALLY slow!
    Main memory speed --  18.2 MB/s  57.5 ns/byte (100%) [reading]   8.6 clks
    Effective RAM access time (read ) is 115ns (a RAM bank is 2 bytes wide). 
    Effective RAM access time (write) is  64ns (a RAM bank is 2 bytes wide). 
    386 Clocked at 39.0 MHz.  Cache ENABLED.
 Options: -t0

I'm pretty sure I disabled cache for at least one of those.

Reply 12 of 30, by Tetrium

Posted on 
User metadata
Rank l33t++
Rank
l33t++

Are you sure the cache is real btw?

Whats missing in your collections?
My retro rigs (old topic)
Interesting Vogons threads (links to Vogonswiki)
Report spammers here!

Reply 14 of 30, by Tetrium

Posted on 
User metadata
Rank l33t++
Rank
l33t++
ratfink wrote:

It's looking like not, but this is an Asus board - I thought it was only PC chips who did that sort of thing?

Kinda true, but I remember (though maybe not related at all!!) reading once or twice about the Asus P2B (my first motherboard) that there were some counterfeit ones!
Perhaps this one is counterfeit also??

The cache soldered on sounds like PC-Chips though.

Iirc there were also DFI boards counterfeited at one time.

Reply 15 of 30, by TheLazy1

Posted on 
User metadata
Rank Member
Rank
Member

Check and see if the traces from the cache go anywhere.
Option two would be to hack up your own memory using nothing but 10ns sram. 😀

Reply 16 of 30, by ratfink

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

The traces... well some of them don't look at though they quite reach the cache chips, but maybe I don't know what I'm looking at.

I'm just assuming it's an Asus, it's got ISA-386C on the board, it looks like what's in the manual and on stason.org. The manual doesn't say Asus anywhere but I guess that's how things were in 1990. I saw those pages about fake p2b's so I guess anything is possible.

Either way, the performance and cachechk readings indicate cache doesn't work, it's not socketed so I can't test the effect of swapping other chips. The bios says it does work, so there is something suspect here. There's also a speed setting in the bios that does nothing I can detect - it always boots at 33mhz and claims cache is enabled. It does save other settings so it's not a bios battery problem.

I'm going the leave it at that. I wasn't intending to run anything all that intensive on it. 🤣

Reply 17 of 30, by ratfink

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie

Problem solved!

Was just about to take the board out when I noticed the turbo pins. Now I have jumpered these, the cache works [and is reported by cachechk] and 3dbench gives me a more sensible score of 12.9. 😁

Reply 18 of 30, by iulianv

Posted on 
User metadata
Rank Member
Rank
Member

I assume you are using an external battery with this board, right? I also have an ISA-386C (with socketed cache though, not soldered) and need to get (or improvise) an external battery for it, but I'm not sure about what voltage it has to provide...

Thanks.

Reply 19 of 30, by Amigaz

Posted on 
User metadata
Rank Oldbie
Rank
Oldbie
ratfink wrote:

Problem solved!

Was just about to take the board out when I noticed the turbo pins. Now I have jumpered these, the cache works [and is reported by cachechk] and 3dbench gives me a more sensible score of 12.9. 😁

Groovy! 😁

My retro computer stuff: https://lychee.jjserver.net/#16136303902327

Go to top of page Go to top of page
Back to General Old Hardware
Close window
Close window