Okay, so the Turbochip seems to be like all the other AM5x86-P75 drop-in solutions from people like Powerleap or Evergreen with the exception that it has no jumpers. If you're crafty like I was you can still add jumpers.
I don't need to tell the regulator to do anything (as it seemed to work in the type 2 complex) and supposedly you need the chip set in Wtite-Through cache mode which the Turbochip has set by default so again I don't need jumpers for that. The only thing that was not what I wanted was the multiplier which we discussed earlier. After reattaching the pin I had broken off earlier for testing the trace between R17 and a via was cut and routed through a jumper so changing the multiplier didn't involve breaking the pin off. With it shorted I was in x4 mode. Without it, x3.

This should work and everything says it WILL work (there's even benchmarks!), but it doesn't. 😕

Probably you haven't seen my numerous postings about how some older systems (particularly IBMs) won't work with the am5x86 in 4X mode. Apparently the CPU uses different IDs depending on how the multiplier is set. Some BIOSes won't boot with an ID they don't recognise. Short of a BIOS update, there are no easy workarounds....unless you can set the adapter to 3X mode an wedge a 1.33X PLL in there somewhere.

Not running because it doesn't like the CPUID? There's a new one for me.

Well I have half of that accomplished with the jumper but I wouldn't have a clue where to source the PLL. I think this in the end might turn into a job for sourcing an Interposer which is rather annoying. I'll have to give your posts a glance but thanks for the heads up.

Edited: I might of sourced a Trinity Works Power Stacker. Supposedly that will work here.

I currently only know of one interposer that uses a PLL, and it's 2X not 1.33X. I'd really like to build such an interposer so we could run Cyrix 5x86 chips at 133 though....

Why do you think the trinity works adapter will work better than the kingston?

Well, again unfortunately I'm running off what I've read. The only time I've SEEN an interposer in a model 95 was with Irongate and he was using an ADZ (but it was on the Type 3 complex so it at least made that variant of the 5x86 work). Everyone else has said along the lines of "yeah, I used this and it worked". The only solid lead I have on the Trinity working is in a Q/A:

> I was all set to install a power stacker 133/586 on my dx50 in the 9595-omt. I for some reason decided to RTFM while having a cuppa. they say the powerstacker will not work with dx50. what cpu can I use to juice up the dx50.?

Jim Shorney
Pay no attention to the manual. Set the multiplier to x3, the cache jumper (if present) to write-through, and plug it in. It will probably work fine.

Won't lie though. Still leaves a lot of open variables and I'm hoping if that still fails I can still use the interposer with one of my other chips. At least the one on Amazon isn't that badly priced.

Are you using this on an original DX-50 motherboard? Those early boards had a lot of stability issues trying to run beyond 33mhz FSB. A later 486 PCI board with UMC chipset can operate with a 50mhz FSB without the problems of the earlier boards.

CelGen is using a PS/2 model 95. A planar swap is out of the question. Pretty sure stability won't be an issue because it's an over engineered IBM.

According to that post on the Trinity Works Q/A the adapter is running in 3X mode. Isn't 4X mode what you wanted?

I have to be honest, squeezing 200MHz out of an am5x86 is pretty unrealistic. You're going to fry the L1 cache.

If you can't get the chip working at all, you can always try the Trinity Works adapter with a Cyrix 5x86-100 in 2x50 mode. They're officially rated to handle that.

According to that post on the Trinity Works Q/A the adapter is running in 3X mode. Isn't 4X mode what you wanted?

Nope. It's a known fact that in my case at least 200mhz is not possible. 150mhz is my ceiling.

You need a lot of luck and the blessings of several major deities to get a 5x86 to run at 200mhz. It just doesn't happen all that often that you get the perfect motherboard and the perfect CPU at the same time.

Second that... especially the blessings of Gods of the cold north wind, icy water pumps and nitro 😀

Some time ago I played with my Am5x86-133. It works 100% at 4x40 = 160Mhz. It even posts at 4x50 = 200MHz. But after the HDD detection it stops - it won't boot. I tried different settings in BIOS but nothing helped. Even changing voltage from 3.3 to 4V.

Now I'm looking at the motherboard manual - CHAINTECH 486SPM - and I'm wondering about the PCI bus clock. Current settings are PCI Clock = system clock, if I'm not mistaking. And there is a setting to halve this - jumper 21. Would this help?

It is easy to figure out - if you can boot at 3 x 50 mhz then it is NOT the high PCI clock.
As we have already found out long ago, you need an ultra-stable board AND a hand-picked processor AND proper cooling (nitro preferred)

Jolaes76 wrote:

It is easy to figure out - if you can boot at 3 x 50 mhz then it is NOT the high PCI clock.
As we have already found out long ago, you need an ultra-stable board AND a hand-picked processor AND proper cooling (nitro preferred)

If you have to use exotic cooling methods to make it stable then it isn't worth it. You want a system that runs stable with the case closed, not one that is an open motherboard with a tank mounted on the CPU that you have to keep refilling with liquid nitrogen every time it boils off. That gets to be tedious and expensive after a while.

you are right - but he could have his 5 min. fame here and on youtube, you know... For some people, that is something at the end of day 😀
We should not discourage him (as long as he does not experiment with 133 mhz Cyrixes on a Biostar MB8433-UUD board or other artefact 😀

Well at least in Phil's Ultimate VGA Benchmark Database Project 160MHz is the ceiling at the moment. We don't have any 50 x 3 or 50 x 4 entries.

Am5x86-P75 40 MHz 4x 160 MHz 3dbench2: 96.3 PCPBENCH/VGAMODE: 24.5 Doom: 55.66 Quake: 15.1 Biostar 8433UUD-A VER:2 Replaced RTC to DALLAS, replaced BIOS chip with flash UMC UM8886BF Modified UUD1212 BIOS from Feipoa 2-1-1-1 Cache WS, DRAM read: 0WS, DRAM write: 0WS 256KB 15ns 8MB FPM RAM 60ns PCI S3 Trio64V+ 2MB MS-DOS 6.22 / No memory managers Mau1wurf1977 VOGONS, OCAU, dosforum, Vintage Computer Forum

Link to project: Phil's Ultimate VGA Benchmark Database Project

50MHz is too much for PCI. I don' t even feel confident running PCI at 40MHz. Use a divider if your boards supports one (3/2 is best for 50MHz).

180MHz on the AMD chip is doable if your board supports 60MHz FSB, but not recommended as the L2 cache can't really handle it and 486 boards get too noisy to do it reliably.

I guess my combo (MB+CPU) won't do 200MHz reliably. I have relaxed all the settings in BIOS and even had to disable CPU cache to boot at 200MHz... but this is a no go. It will stay happy at 160MHz with all the bells and whistles.

Did anyone else try a 60Mhz FSB on a 486 Board ? I got an ADW running at 3x60Mhz stable on GA 5486AL with ET6000 VGA, the PCI clocks at 30Mhz so the system has a very good memory and PCI troughput

As a bit of a content BuMP I managed to secure an Interposer (PGA-DX4/2) for not too much money and discovered I had a 133ADW hidden away in another box which should now work, however I've unintentionally screwed up and discovered that the interposer has an extra pin at position D4. I cannot remember if this was a pin added for orientation keying like on Socket 1 or if it was a clock signal pin like on socket 2 or 3, or if I've forgotten something and I'm forgetting that on socket 2 and 3 the outermost pins are not used. 😕