I would think this is a good question for the CPU-World forums.

You might be surprised. I've asked the same question in the past on CPU-world concerning another CPU and there wasn't a single response. e.g. http://www.cpu-world.com/forum/viewtopic.php?p=235804

In general, I find the Vogons forum substantially more active, especially when it comes to the use of classic CPUs.

Nobody?

There is a seller in Florida who has dozens of new, old stock TI 486SXL2-66 chips and I can't help but wonder what the target motherboard or system was. The hardware requirements of the SXL2 different somewhat from the i486SX, however, I may test it anyway in a socket 3.

Cant help but is he selling the chips individually?

Yes, at least when I bought them 1-2 years ago he was. I just checked, and it looks like he's still selling them for $16.50 each. https://www.westfloridacomponents.com/IC490PE … ocessor+TI.html

Because of the cheap cost and easy of availability, I thought it would be a good idea to create a PGA132-to-PGA168 adapter, Custom interposer module for TI486SXL2-66 PGA168 to PGA132 - HELP! , this way, pretty much anyone with a 386 can run the the fastest plug-in compatible upgrade, outside of a Transcomputer module or BL3. Unless someone else steps in to design this translation PCB w/VRM, I'll probably pick it up again in the fall. I want to map out the SXL2 PGA132-to-PGA168 (w/out VRM) interposer I have now to ensure I have the pins right. I don't understand why the interposer in hand contains a 7-bit clock counter, 4xNAND gates, and one unknown IC. I'm hoping that these aren't necessary. The QFP144 interposer I have also has some logic circuit, but its a PAL, so there is less value in mapping it out.

I still see 46+ for sale. I don't see an inventory listing anymore. I bought 3 some time ago and the import duties were a little painful to pay.

Kind of sacrilege but I was thinking about buying the CPU and using in a hybrid 386/486 board and calling it a day, pretend its a 386 version but for all practical purpose its the same thing, might be too hard or expensive to find a real 386 version.

edit: IS the voltage much of an issue?

Old skool 486 boards are 5V only and won't work, is 3.3V too little?

Feels sorta lame to put the PGA168 version of the SXL2-66 into a hybrid 386/486 board as those boards also take Intel 486DX CPUs, however I was thinking about doing the same thing before I obtained these PGA132 variants. If you do put it into a 386/486 board, at least you have the added advantage of VLB. VLB + SXL2 is a neat weird combination.

I've only tried the SXL2-66 at 3.45 V, so I cannot comment on 3.3 V. Do you want me to test it out? Do you have a VRM set to 3.3 V?

I was able to run my SXL2-66 at 3.32 V. I ran DOOM timedemo a few times and loaded Win3.11. Opened Word Perfect 6 and loaded google.ca with IE3. Far from conclusive, however I can say that it does not, not work at 3.3 V.

I have a test motherboard which supports the UMC U5SX, i486SX, Cyrix Cx486DX and Cyrix Cx486S. I plan on testing all these CPUs and hopefully the TI 486SXL2 will work with one of those CPU jumper settings. If not, I'll try to modify the board using the guidelines in Appendix D of the TI 486SXL databook.

I know the Cx487S adapter (photo below) should work with the FPU-less Cx486S CPU, however, I'm hoping that it will also work with the SXL2 and U5SX. I suspect it won't work with the i486SX.

I know for a fact it won't work with the 486SX, because others have tried it.

Your best bet for the SXL on a socket3 board is to use the jumper settings for the 486S. Both are essentially using DLC cores with extra cache. However, I believe the 486S supports writeback cache, where the SXL likely does not.

Anonymous Coward wrote:

I know for a fact it won't work with the 486SX, because others have tried it.

Your best bet for the SXL on a socket3 board is to use the jumper settings for the 486S. Both are essentially using DLC cores with extra cache. However, I believe the 486S supports writeback cache, where the SXL likely does not.

I second Anonymous Coward, I also have this Cx487s interposer and so far only the Cx486s worked with it.
And the Cx486s uses writeback cache, but not all boards support it/detect it like that.

The settings for the normal 486sx or the Cx486s should have the biggest chance getting the SXL to work. Biggest problem I foresee is that hardly any bios wil recognize them...

I vote the 168-pin version of the SXL as the weirdest x86 chip ever.

Nvm1 wrote:

I also have this Cx487s interposer and so far only the Cx486s worked with it.

Which CPU settings did you use on the motherboard that you tested the Cx487+Cx486S with? Did you try running an application which requires the FPU? I'm wondering if the CPU jumper settings for the Cx486S have BUSY# and PEREQ connected. Without those, the FPU doesn't work properly. If they are not connected, then it seems that the jumper settings for the Cx486DX might be needed for the Cx487S+Cx486DX. Did you also try the CX486DX CPU settings when using the CX487S+Cx486DX?

Did you try the Cx487S with any other CPU, like the UMC U5S or U5SX?

I'm pretty sure that the i486SX doesn't work with the Cx487S because BUSY#, PEREQ, and one ore two more pins are NC (no connect), so it doesn't have the internal infrastructure to negotiate/hold while the FPU is active.

Anonymous Coward wrote:

I vote the 168-pin version of the SXL as the weirdest x86 chip ever.

It is certainly high up on the list.

feipoa wrote:

Which CPU settings did you use on the motherboard that you tested the Cx487+Cx486S with? Did you try running an application whi […]
Show full quote

Nvm1 wrote:

I also have this Cx487s interposer and so far only the Cx486s worked with it.

Which CPU settings did you use on the motherboard that you tested the Cx487+Cx486S with? Did you try running an application which requires the FPU? I'm wondering if the CPU jumper settings for the Cx486S have BUSY# and PEREQ connected. Without those, the FPU doesn't work properly. If they are not connected, then it seems that the jumper settings for the Cx486DX might be needed for the Cx487S+Cx486DX. Did you also try the CX486DX CPU settings when using the CX487S+Cx486DX?

Did you try the Cx487S with any other CPU, like the UMC U5S or U5SX?

I'm pretty sure that the i486SX doesn't work with the Cx487S because BUSY#, PEREQ, and one ore two more pins are NC (no connect), so it doesn't have the internal infrastructure to negotiate/hold while the FPU is active.

I will have to check my old notes for my exact testing and results, will try to do this next week. It was quite some time ago so I don't want to write things I didn't do 😐
Out of my head I tested it with:
- Intel 486sx-25
- AMD 486SX
- UMC 486SX-40 (U5S)
Settings for the Cx486s+Cx487s where those of the Cx486DX --> otherwise no fpu was detected. This I am sure about.
As soon as I have checked what I did I will write a reply.

Did a massive test of some 80 cpu's between september and march so some things I know for sure. I know the Cyrix Cx486s+Cx487s was a pain to get going on some motherboards...

Oh, great. Looking forward to more info. Sounds like the motherboard jumpers don't need to set the 486S for 2 KB cache if the 486DX jumpers work.

What motherboard did you use?

Will you be providing a benchmark table or spreadsheet for your 80 tested CPUs?

I was able to get the Cx486DX-40 and UMC U5SX running in two different UMC 8881/8886-based motherboards. Unfortunately, when the Cx487S is added to the U5SX, Landmark doesn't show any FPU usages. I tried the Cyrix Cx486S in the same two motherboards and in both cases, I was unable to POST. The system seems to hang right at the point where it should boot from a floppy drive. I tried, both, Cx486S and Cx486DX settings. When adding the Cx487S to the Cx486S, all I get is a black screen after power-on. TI 486SXL2-66 shows a black screen at power-on in both motherboards, even after several CPU jumper configurations. Still black screen with the Cx487S added to it.

The results of these tests were rather depressing. I really hoped to get an FPU added to the UMC U5SX chip. Anyone seen an official PDF datasheet for this CPU?

I suspect the issue with the Cx486S is a chipset issue. I should try a motherboard with a SiS 496/497 chipset or ALi.

feipoa wrote:

I was able to get the Cx486DX-40 and UMC U5SX running in two different UMC 8881/8886-based motherboards. Unfortunately, when the Cx487S is added to the U5SX, Landmark doesn't show any FPU usages. I tried the Cyrix Cx486S in the same two motherboards and in both cases, I was unable to POST. The system seems to hang right at the point where it should boot from a floppy drive. I tried, both, Cx486S and Cx486DX settings. When adding the Cx487S to the Cx486S, all I get is a black screen after power-on. TI 486SXL2-66 shows a black screen at power-on in both motherboards, even after several CPU jumper configurations. Still black screen with the Cx487S added to it.

The results of these tests were rather depressing. I really hoped to get an FPU added to the UMC U5SX chip. Anyone seen an official PDF datasheet for this CPU?

I suspect the issue with the Cx486S is a chipset issue. I should try a motherboard with a SiS 496/497 chipset or ALi.

Well, I just peaked on my old phone and found the pictures from the stacked Cx486s and Cx487s with the picture to show it was detected:

It was a PCChips M912 motherboard. Rev 1.7 from the looks of it

I think I have that motherboard, although mine doesn't have that surprisingly cool looking PC Chips sticker on it. I'll try the PC Chips motherboard to ensure my 486S is OK.