The first thing I’ve done is flash the latest jan bios to the board for the S1564D using my chip programmer. http://web.inter.nl.net/hcc/J.Steunebrink/k6plus.htm

Pretty easy stuff,

Next I ordered a new RTC from monotech

Now I need to install the APIC

Luckily, this board doesn’t really seem to be missing anything other than the APIC, the socket and possibly a jumper or two

I used a hot air gun and some of this cool soldering paste

First I dabbed a little solder on the pads of the apic then lined it up on the pads and pushed down and went around it with the hot air gun two pics because it looks weird in the first, but it’s all good

Next I have to add the single/dual cpu selector jumper

Not really sure of the purpose of this jumper because my dual works fine with it in the dual position even with a single processor.

Might be responsible for switching on the use of the apic vs using the internal apic in the processor which is probably why it works on a single. Idk I’m just guessing.

Might be kinda neat to see if the external APIC or internal performs faster in single chip mode with a K6

Here you can see the new jumper all installed and ready to go.

(Black one)

Well that’s enough for tonight, unless one of you keen eyed electronic wizards sees a difference I missed I think the only thing left is the socket I even fixed the missing I/O ground tab on the kb port

I’m still waiting on the rtc and a atx to AT adapter so no big hurry.

I may even just leave it at this stage for a while.
I was hoping to find a exact match for the socket that’s in it but the AMP one that I ordered is a little different. I kinda want it to look original. So if anyone has a busted tyan laying around lemme know.

The board so far as it knows is a dual now, just no socket, I don’t have a pair of 233s for it, just 200s might just run a k6 3+ 400 in it for a while and compare to my dual 233

Power leap and cyrix just because it looks cool.

neat! i think you might need to set some 0-ohm smd resistors near the apic chip.

i think i have one of these boards, but i have NO idea where it is atm.

More is always better! this is good work 😀

luckybob wrote on 2021-12-23, 08:33:

neat! i think you might need to set some 0-ohm smd resistors near the apic chip.

i think i have one of these boards, but i have NO idea where it is atm.

I’m not seeing a difference between the two boards in that area now

chinny22 wrote on 2021-12-23, 11:08:

More is always better! this is good work 😀

Thx!

While I’m at it 512k is the most cache this board can take right?

Does this board use dual 8k tag chips?

And if anyone has any ideas for transplanting a faster clock gen into it I’m all ears.
New clock gen for tyan s1564d (Research)

Well, I got bored

Went ahead and just used the AMP socket.

I did notice this: (thx lucky bob) but haven’t done anything about it yet

It seems to post just fine but the RTC is being an issue.

I tried to start windows setup but it needs sata drivers, screw it. Gonna throw ubuntu on it and see what happens

🧐 It appears that I have two cpus at my disposal

Interesting to see that the single socket variant wasn't more crippled than the dual socket board. It would of been a good value back in the day. The dual socket board would of been a rip off for the consumer if you really needed dual socket back then. Since the missing components really isn't expensive.

Sphere478 wrote on 2021-12-23, 15:05:

I did notice this: but haven’t done anything about it yet67170689-57EE-4055-8E22-20B3BD17B66A.jpeg92BF016D-3AC8-4C16-91BF-D41CC4DA1D55.jpeg

This looks to me more like a pcb revision and less like a revision related to dual or single socket.

Next step is solder on more L2 Cache?

Warlord wrote on 2021-12-23, 22:08:

Interesting to see that the single socket variant wasn't more crippled than the dual socket board. It would of been a good value […]
Show full quote

Interesting to see that the single socket variant wasn't more crippled than the dual socket board. It would of been a good value back in the day. The dual socket board would of been a rip off for the consumer if you really needed dual socket back then. Since the missing components really isn't expensive.

Sphere478 wrote on 2021-12-23, 15:05:

I did notice this: but haven’t done anything about it yet67170689-57EE-4055-8E22-20B3BD17B66A.jpeg92BF016D-3AC8-4C16-91BF-D41CC4DA1D55.jpeg

This looks to me more like a pcb revision and less like a revision related to dual or single socket.

Next step is solder on more L2 Cache?

I’m not sure…🤔 I don’t even know if that even applies to the APIC at all…

But it seems stable.

Well, that’s what I’m trying to figure out…

From what I can research it suggests that this board uses two 8 bit tag chips (so even more than your typical board which is cool) but everything I see says hx is max 512k cache I need to talk to someone around here that knows more about cache and see what they think I would love to upgrade to more cache!

I’ll run some more tests eventually but probably going to re focus on my p5a for a while.

Memtest is saying 1 core.. 🤔 (edit: memtest problem not board problem)

Trying windows xp install

luckybob wrote on 2021-12-23, 08:33:

neat! i think you might need to set some 0-ohm smd resistors near the apic chip.

i think i have one of these boards, but i have NO idea where it is atm.

It’s seeming like you are correct.

Just make em exactly like the dual right?🤔💁‍♂️

Do you have the Intel datasheet for the 430HX? I've attached it just in case not.

Skimming the info they seem clear that 512KB is the limit for cache size, but it would take someone with more knowledge to say whether any "cheating" is possible.
I do wonder if having dual CPUs changes the rules any, but probably not since if they had separate cache per CPU then there'd be another coherency issue to deal with.

Page 21 describes programming a register that controls the cache setup. Secondary Cache Size is a 2 bit value with options of No Cache, 256KB, 512KB, and "Reserved". I wonder if there's any chance the "Reserved" setting would do anything interesting.
Also see page 34-35, where they talk in more detail about cache sizes and the Tag RAM requirements per cacheable range.
Unfortunately no hint of going beyond 512KB and 512MB cacheable range.

I'm guessing the BIOS normally programs the setup for this stuff so somebody who has experience with BIOSes might know more about it.

XP says two cores…?

Thoughts?

Anyone with more knowledge on the apic have any ideas if those resistors might be why memtest doesn’t detect two cores? (Edit: memtest was the problem, not the board)

What are they for? Hesitant to modify them without more understanding dual processor seems to be working in linux, post, and windows XP

I just finished a project and i saw my dual tyan board and it was easy to access!

My board https://imgur.com/a/qEd0d1S

My board is a s1562, but the numbers on the resistors are easy to read. As I recall, the chips are all identical.

512KB of cache combined with 2 TAG chips (8 bit each) are sufficient to cache all 512 MB of RAM.
So it would be pointless to extend L2 cache any further beyond 512 KB as long as the system does not correctly and fully address more than 512 MB of RAM.

majestyk wrote on 2021-12-24, 08:25:

512KB of cache combined with 2 TAG chips (8 bit each) are sufficient to cache all 512 MB of RAM.
So it would be pointless to extend L2 cache any further beyond 512 KB as long as the system does not correctly and fully address more than 512 MB of RAM.

Okay so that begs the question, if I add more cache can I install 128mb sticks in my remaining 4 simm slots?

luckybob wrote on 2021-12-24, 07:32:

I just finished a project and i saw my dual tyan board and it was easy to access!

My board https://imgur.com/a/qEd0d1S

My board is a s1562, but the numbers on the resistors are easy to read. As I recall, the chips are all identical.

I have two of these boards one is a dual from tyan and this one was a single.

It’s interesting your board seems to have yet a third configuration for those resistors..

Sent you a pm btw