VOGONS


Tyan S1562 Dual Pentium MMX succes

Topic actions

Reply 20 of 36, by chinny22

User metadata
Rank l33t++
Rank
l33t++

Duel CPU, Awesome
Duel MMX, Awesome and rare
Getto Fix, Now has character as well 😀

Reply 22 of 36, by luckybob

User metadata
Rank l33t
Rank
l33t

Short answer; No.

Long answer; kinda.

unless you know exactly what parts on the motherboard need changed. Intel dual cpu setups require a special chip installed on the motherboard, that chip also has to be installed and properly configured via resistors on the board.

There is another thread on this forum where someone is attempting the same, but with a Pentium Pro.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 24 of 36, by luckybob

User metadata
Rank l33t
Rank
l33t

I didn't think you could use MMX chips on these boards, so my whole world is turned upside down.

I know the "official" jumper settings stop at 66mhz. Going further would likely depend on undocumented settings of the individual signal gen.

It is a mistake to think you can solve any major problems just with potatoes.

Reply 25 of 36, by feipoa

User metadata
Rank l33t++
Rank
l33t++

What is the part number on the PLL? Spec sheet might show a frequency table.

I also didn't think p233 MMX chips would work on these boards, but after reading the thread it sounds like the reliability is uncertain.

Plan your life wisely, you'll be dead before you know it.

Reply 26 of 36, by flupke11

User metadata
Rank Oldbie
Rank
Oldbie

The PLL is an ICS9159C-02 if I can read correctly from a picture I took a few months back.

I'll take it out of storage and run a few tests on it this week.

One interesting thing to check is the EDO/FP issue on the 8 slots.

Reply 27 of 36, by feipoa

User metadata
Rank l33t++
Rank
l33t++

Datasheet attached. Looks like 66.6 MHz is the max you can go.

Concerning EDO and FPM on my dual P233 MMX board (NEC Proserva V Plus), I actually noticed the cachechk memory read throughput decreased when using EDO compared to FPM, that is 79.2 MB/s vs. 81.9 MB/s. Not sure why this is or if it is a common result among these dual HX boards. I'd be interested to know what your benchmarks show for EDO vs. FPM. Those values above are using a P200 though. I haven't yet test the EDO again with the P233 MMX chip installed.

Attachments

  • Filename
    ICS9159C-02C.pdf
    File size
    311.88 KiB
    Downloads
    63 downloads
    File license
    Fair use/fair dealing exception

Plan your life wisely, you'll be dead before you know it.

Reply 29 of 36, by feipoa

User metadata
Rank l33t++
Rank
l33t++

That's quite the collection. Is that EISA/PCI one an NX chipset?

I don't really understand the difference between the S1562/3/4 boards.

Do any of your boards support a 75 MHz FSB?

Plan your life wisely, you'll be dead before you know it.

Reply 30 of 36, by pentiumspeed

User metadata
Rank l33t
Rank
l33t

Tyan brings bad memory back in the day, I helped guy purchase his nice setup using dual pentium tyan board, 420MB WD and installed win98se, new memory. Corrupts the win98 and crash.

When we returned, seller is gone. 🙁

I was not sure how good Tyan is ever since and I had not seen any Tyan boards personally or owned one.

Cheers,

Great Northern aka Canada.

Reply 32 of 36, by flupke11

User metadata
Rank Oldbie
Rank
Oldbie

In order to keep things tidy, I'll update this thread with some info and an issue involving the cacheable area.

I'm configuring this system as an all purpose 90's machine (DOS to WIn2K)***. It currently has 4*32 EDO installed (Bank 0 and 1), my COAST module has one tag RAM, the second is installed as the manual demands and all jumper settings are correct to unlock the full cacheable range of the chipset, i.e. 512 MB.

I've replaced the DALLAS, which luckily was socketed. The system happily detects a 32GB CF card on the Startech adapter.

When selecting 512MB instead of 64MB cacheable range in the BIOS settings (Chipset Features screen), I get errors all around: speedsys hangs when it starts to check the cache speed, Win98 hangs at install as well as Win2K.

The manual does not deal with that option, I suppose it's a later addition. The manual also was not aware that the S1562D can run 2 P233MMX's, so that's not much of a help.

Any ideas in how to unlock the proper cacheable range?

*** I know it's useless to have a dual system when using DOS, but I need to drastically reduce the systems I hoard have, so I'd rather keep the duallies.

P.s. the difference between the S5162 and later iterations (S1563-S1564) is at least apparent in one area: no more COAST. With the fiddling I had to do to keep it properly seated, I understand they got rid of it.

Reply 33 of 36, by flupke11

User metadata
Rank Oldbie
Rank
Oldbie

Second update: CTCM7 test:

512 in BIOS, 2CPU: error at Cacheable Area L1: "A20-Error via Himem Prog stopped"
512 in BIOS, 1CPU: idem
64 in BIOS, 2CPU: no errors in CTCM7, cacheable area correctly identified as 64MB.

Reply 34 of 36, by SilverHawk

User metadata
Rank Newbie
Rank
Newbie

Very nice! I regret the day I got rid of my Tyan Tomcat 1564D. It was hard to come by when new and the one I received had a faulty onboard IDE controller. I threw in an IDE controller card and used it for quite awhile. I believe I still have the original manual somewhere.

Reply 35 of 36, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
luckybob wrote on 2020-03-02, 22:37:

I didn't think you could use MMX chips on these boards, so my whole world is turned upside down.

I know the "official" jumper settings stop at 66mhz. Going further would likely depend on undocumented settings of the individual signal gen.

If the fsb jumpers are three two position jumpers your chances increase exponentially of finding undocumented settings. Loop up the data sheet for the clock gen.

On mine I found undocumented 83mhz setting that the clock gen supports that I plan on trying (different single socket board)

But on that note, I actually just ordered a Tyan S1563 Tomcat Dual Socket 7 AT Motherboard

And am pleased to see that dual mmx 233 is possible

I am hoping for more though.

I was hoping for dual tillamook but I hear they totally removed the connection to the smp pins on those chips, so unless the board has it’s own smp controller which I don’t think it does I don’t think it will work except single chip.

I also would like to run dual k6 but again not expecting this to work at all. 🙁

Though if I get ahold of a super old socket 5 or any that has it’s own smp controller, from what I read those two scenarios may become possible but of course a power leap interposer would be required on both sockets on a socket 5 or socket 7 with wrong voltage options.

Many boards, their lowest voltage settings are realized with all voltage jumpers removed but this isn’t always the case. Toms hardware did one oldie tuning or something rather was the name where they had to install a bunch of jumpers to lower the voltage.

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)

Reply 36 of 36, by Sphere478

User metadata
Rank l33t++
Rank
l33t++
flupke11 wrote on 2019-09-08, 17:26:
Update, fixed the issue with the COAST module more or less by a thread: […]
Show full quote

Update, fixed the issue with the COAST module more or less by a thread:

Webcam-1567962779.png

CPU-0 gets 219 on Integer and 135 on Floating
CPU-1 gets 209 on Integer and 136 on Floating

System seems to behave now with the cache all bound up.

I think your memory issues may have been the way your ram was installed in slots 1,2 and 5,6

As apposed to 1,2 and 3,4

Sphere's PCB projects.
-
Sphere’s socket 5/7 cpu collection.
-
SUCCESSFUL K6-2+ to K6-3+ Full Cache Enable Mod
-
Tyan S1564S to S1564D single to dual processor conversion (also s1563 and s1562)