VOGONS


First post, by Flakchak

User metadata
Rank Newbie
Rank
Newbie

Hi all,

It seems cache is still the bane of my existence. I have a CH-890C motherboard and despite being over 20 years old, the Dallas Clock Chip is still ticking.

But after running a number of dos benchmarking programs, I can't get any read on how much cache I have on board. I'm hoping that someone has a little knowledge they can lend me. I don't have a manual, nor can I find one online. The motherboard appears capable of 256k, 512, and 1mb of cache.

I'm just wondering, what types of chips would I need with 8 sockets, plus a tag socket to have 256, or 512, or 1mb. I've tried a number of jumper settings, which leads me to believe the cache I have may not be legit. But if anyone can help me out, I'd greatly appreciate it.

Attachments

  • CH890C.jpg
    Filename
    CH890C.jpg
    File size
    1.71 MiB
    Views
    377 views
    File comment
    CH-890C Motherboard
    File license
    Fair use/fair dealing exception

Searching for a Packard Bell 15" CRT Monitor - 1511SL or 1512SL

Reply 2 of 4, by dionb

User metadata
Rank l33t++
Rank
l33t++

That cache looks bona-fide, but even though the chips all have 32kx8 structure, your problem might well be the way they are mismatched. Generally you want 8 the same chips, with a ninth chip for tag.

Here you have 8x 256kb = 256kB cache, plus an extra 32kx8 tag. For 512kB you'd want 8x 512kb, but the 256kb tag should still be enough. I wouldn't expect any performance boost from more cache though. In fact what you have here is - clock for clock - the slowest So5 (or So7) motherboard out there. It is the slug of slugs, mainly because the chipset is literally a 486 chipset kludged onto a Pentium, with the 32b (not 64b) memory controller to go with that. Correspondingly, it's even slower than earlier UMA attempts like SiS 5511+6202 boards. If anything is interesting on this board it's to see how slow it will actually go. As you can see on your pic, there are settings for FSB down to 33MHz. Watch it crawl 😜

But back to cache: why don't you think you have any? What does CACHECHK.EXE tell you?

Reply 3 of 4, by Warlord

User metadata
Rank l33t
Rank
l33t
dionb wrote:

It is the slug of slugs, it's even slower than earlier UMA attempts like SiS 5511+6202 boards. If anything is interesting on this board it's to see how slow it will actually go. As you can see on your pic, there are settings for FSB down to 33MHz. Watch it crawl 😜

maybe thats good for retro gaming. Probably sucks that its a pentium, but eh seems like people like things that go slow.

Reply 4 of 4, by Flakchak

User metadata
Rank Newbie
Rank
Newbie
derSammler wrote:

Jumpers are set to 1 MB, but the cache chips are 256 x 1 each, so 256 kb in total.

Isn't there a hardware summary screen when the system boots? What does it say?

I changed the jumpers to reflect 256 cache. The hardware summary screen at boot mentions nothing of the cache.

dionb wrote:

That cache looks bona-fide, but even though the chips all have 32kx8 structure, your problem might well be the way they are mismatched. Generally you want 8 the same chips, with a ninth chip for tag.

Here you have 8x 256kb = 256kB cache, plus an extra 32kx8 tag. For 512kB you'd want 8x 512kb, but the 256kb tag should still be enough. I wouldn't expect any performance boost from more cache though. In fact what you have here is - clock for clock - the slowest So5 (or So7) motherboard out there. It is the slug of slugs, mainly because the chipset is literally a 486 chipset kludged onto a Pentium, with the 32b (not 64b) memory controller to go with that. Correspondingly, it's even slower than earlier UMA attempts like SiS 5511+6202 boards. If anything is interesting on this board it's to see how slow it will actually go. As you can see on your pic, there are settings for FSB down to 33MHz. Watch it crawl 😜

But back to cache: why don't you think you have any? What does CACHECHK.EXE tell you?

I ran cache check for the mismatched cache pictured above. This is my result:

previous.jpg
Filename
previous.jpg
File size
1.34 MiB
Views
322 views
File license
Fair use/fair dealing exception

I then put in a set of matched chips. The tag chip I left, as using the same chip as the other 8 would cause a hang at boot:

updatechips.jpg
Filename
updatechips.jpg
File size
1.62 MiB
Views
322 views
File license
Fair use/fair dealing exception

And this is my result in Cachechk:

new.jpg
Filename
new.jpg
File size
1.34 MiB
Views
322 views
File license
Fair use/fair dealing exception

Searching for a Packard Bell 15" CRT Monitor - 1511SL or 1512SL