Sorry, I tend to forget what I've posted and where.

Pointing out hardware combination failures is also a good idea. Thanks.

I'm reviving this thread to ask if anybody has found more motherboards supporting the Ti486SXL2 in the PGA168 format.

Is the pinout (and behaviour) different from that of the Cyrix 486S (the so called 'FasCache')? I've a couple of motherboards supporting it (including the interposer with the Cx487 FPU) and one of them also has a VRM for 3.3 operation.

Yes, I'm aware of the effort by some talented individuals to create an interposer to use this CPU on 386 motherboards, but I'm asking because I'm curious about other possible alternative. Aside from some PC cards made for the Acorn Archimedes, I've never seen a 486SXL/2 employed anywere.

see here: Need help to improve the support of 486/586/686 class CPUs in CPU-Z - I posted a photo of the board with ti486sxl@pga168 there.
Ti486sxl does not work with interposer with the Cx487 FPU. However, the exp4349 board does not work with the Cyrix fpu in its socket with any CPU. ( 386dx, dlc...).
I recommend excellent videos about exp4349: https://www.youtube.com/watch?v=04IlhB1Snzg
Boards without stoket for fpu and 386 only work up to 33Mhz FSB.
The motherboard whose photo I posted had all the sockets and worked at 40Mhz (worked at 50Mhz ? - I don't know, I haven't checked).
I think the EXP4349 was installed in a series of computers for German schools - boards without stokets were shipped with an AM486dx2-66 processor

Correct me if I'm wrong but the EXP4349 is capable only of supporting 5V CPU (therefore, not the SXL2), right? At any rate, besides this Data Expert EXP 4349and the Macronics LX30WB, are there any other MBs, capable of supporting the SXL2 out of the box?

sxl-40 and sxl2-50 work on 5v only sxl2-66 is marked as 3.6v.
all sxl processors (even sxl-40) have clock doubling.
they all start in x1 mode and doubling must be turned on.
it is possible that they are all the same - made in the same process.
exp4349 has free space for a 3.6v power supply, it is not just mounted. see part 3 of the video - this feature is added there.

I saw photos of another motherboard with 486sxl where the bios knew the processor type but it was pga132, there was no socket 168.
I saved the photos but I didn't take them so I shouldn't post them publicly.

I think it is realistic to find only the DataExpert board (with socket 168)

MSxyz wrote on 2024-07-12, 06:40:

Is the pinout (and behaviour) different from that of the Cyrix 486S (the so called 'FasCache')? I've a couple of motherboards supporting it (including the interposer with the Cx487 FPU) and one of them also has a VRM for 3.3 operation.

For the 168-pin package of the 486SXL, TI basically made no efforts whatsoever to adapt this chip to interface with an actual 486 socket. The CPU core used in the 132 and 168 pin versions of the SXL are identical. It's still using the 386 bus, and the extra 36 pins are not used. Maybe the package was used purely for cosmetic reasons, to add to the illusion that you have a real 486 and not some 386-486 hybrid chip?

I believe the Cyrix 486S is a little different. Cyrix made some modifications to the CPU core, the most notable being the extra 1kb of cache which is now capable of writeback mode, and I think it kind of works halfway on a real 486 bus, at least I think the pinout is close to that of an i486SX. I remember feipoa once tried to adapt the 486S to work in a board with no official support, but I can't remember the final outcome. Not sure why Cyrix didn't go for a proper 486SX pinout. Maybe it had to do with the FPU dingus.

The 168-pin package of the TI486SXL is 99% identical to common i486 & am486 CPUs.

For the TI486SXL, besides the extra pins for communicating to the 387 FPU, the NMI pin is only in the same position as the i486SX (different position to i486DX and am486), and the CLK2 pin accepts a doubled clock frequency, whereas real 486's accept CLK / non-doubled.

I recently bought an Acer V5 motherboard (386 SIS rabbit 310/320/330) which has a real 486 upgrade socket. It accepts and runs a i486dx4-100, but not a TI486SXL-50.

With both real 486 and TI486 CPUs available in 168-pin, and being so similar to implement, who would build for the TI486SXL and not future proof the design...

I've tried today to insert a SXL2-50 into the 486 socket of a BC3486 motherboard and configured the switches to accept a 486SX but nothing happens. No beeps, no image, no other signs of life. It's as if there was no CPU inserted. The chips does get slightly warm to the touch. A shame, because this board can do a wide range of clocks both for 386 and 486.