The Speedstar 24 offerings on eBay are way over my budget. I do have an S3 801 and at ET4000 with a true colour RAMDAC sitting at AVICC though. I'll test them out when I ship my AVICC holdings to me. I was told that the 801 can do 2D acceleration in Windows, so this might speed up the system quite a bit.

Do you specifically recall testing your Micro-Labs ET4000/w32i on this here VLSI 286 board? I'd be surprised if we had different outcomes. I've attached a photo of my ET4000/w32i card:

I see this screen on initial power on:

Shortly after RAM checks OK, it hangs on this screen (notice the garbled characters):

EIDT: @bitwrangler
It is possible.
Is it the VLSI 100 series based one ?

Try re-testing the Tseng card with the 0WS jumper set to 'Disabled' (bottom two pins). Beneficial as 0WS mode is, it doesn't always work on all old motherboards, and sometimes causes weirdness like that to occur.

Never used the BIOS test tools on any of the VLSI boards. At least don't remember doing so.
What i know is that ET4000/w32i was smooth with any common tests and any apps/games i threw at it.
Can try and see if it hangs with the BIOS test.

Exactly the same model card here.
Good chance the mobo was tested with it, but cannot confirm 100%.
It was 3 years ago.

My et4000ax works perfectly at 0ws on my VLSI 200 series system (20MHz CPU/10MHz ISA)

Eep386, I had already tested with 0 ws and no 0 ws, and with IRQ enabled and IRQ disabled. I also tried without any other cards installed. The outcomes were the same on my VLSI200 286 system. On the other hand, the Microlabs card worked in all 386 boards I tested it on, so...

I'm still waiting on a BIOS from megatronUK from his VLSI board, maybe that will help matters. Unfortunately, though, the Micro-Labs card only supports high resolution (I think 1024x, or 1280x, I forget which) and high colour in interlaced mode, which my LCD doesn't support. I've been told that some cards have better RAMDAC's which can do 60 Hz with the 2 MB, but not mine.

Megatron-uk got his VLSI-200 board from me.
It was identical in functionality to yours.
I doubt BIOS from him will make a difference.
Hope it does.

---
Tomorrow will run the BIOS test with the w32i card and see if it crashes.

---
If I remember correctly that card can do 1024x768x16 non-interleaced.
Will check that too before say too much.

---
@maxtherabbit
Et4kAX is fine for everybody.
The question is why the w32i variant gives Feipoa trouble.

When I look at the website for the card, https://www.microlabs.com/xl2.html , it says 45 ns DRAM. It looks like I upgraded my card with 50 ns DRAM. Could this be causing some issue?

I confirm that Micro-Labs ET4000/W32i passes the VLSI-200 BIOS tests.
The card here looks like this.

Doubt that 5ns difference in speed rating for some of the memory chips you used will make the card unstable, but who knows really.

feipoa wrote on 2021-12-25, 01:53:

I've been told that some cards have better RAMDAC's which can do 60 Hz with the 2 MB, but not mine.

I'd like to know which boards these are.

The 5V FPU and SXLC arrived and I soldered them onto the Evergreen 486 SuperChip. The good news is that the chips are functional, that is, not DOA; the bad news is that the 5V SXLC2-40 suffered the same fate as the 3.6V SXLC2-50. Trying to use the L1 cache of the SXLC CPU causes the system to hang. So, I'm guessing there is some aspect of the Evergreen's onboard logic that is interfering with proper operation of an SXLC.

This leaves the SLC 25/50 or SLC 33 as the fastest options for this interposer. I was informed, unfortunately, that the SLC 25/50 is permanently stuck on 2x mode. This will greatly limit the front-side bus and performance. Does anybody have any data or anecdotal evidence of how far the Cyrix SLC 25/50 (not IBM varient) and Cyrix SLC 33 will overclock reliably? I will have the following scenarios to test once the SLC 25/50 chip arrives.

SLC2-25/50
CRYSTAL - FSB - CPU, where CPU = CRYSTAL/4*2*2
50 - 12.5 - 50
55 - 13.8 - 55
60 - 15 - 60
66 - 16.7 - 66
70 - 17.5 - 70

SLC33
CRYSTAL - FSB - CPU, where CPU = CRYSTAL/4*2
66.6 - 16.7 - 33.3
70 - 17.5 - 35
75 - 18.8 - 37.5
80 - 20 - 40
87.5 - 21.9 - 43.8
90 - 22.5 - 45

I doubt the SLC2 25/50 will be stable past 60 MHz and the SLC33 past 45 MHz. My SLC25 would only do 33 MHz stable, so it is possible 40 Mhz is the upper limit for the SLC33. My guess is that SLC2 25/50 can do 55 MHz reliably with a 13.8 MHz FSB and the SLC33 can do 40 MHz reliably, with a 20 MHz FSB. Which is the faster option?

So you think the SLC 25/50 will have an upper limit of 55MHz just because that's all the TI version can do? I'm more optimistic about the Cyrix. It has far less cache, so it should be easier to overclock. With the Cyrix branded chips in particular, I would also think that the date of manufacturer would influence your results. The SLC2 25/50 was one of the last SLC chips, and they should have all come out pretty late.
Unlike TI, Cyrix actually had a 5V part that ran at 66MHz. Although it's hard to say exactly how many people have overclocked their DRx2/SRx2 25/50 to 66MHz successfully, I can tell you that it is definitely possible. My SRx2 25/50 runs at 33/66 just fine.
Now, I remember that the DRx2 33/66 did not have great yields (at least initially) and it was kind of hard to get for a while...but Cyrix never made an SRx2 33/66, which in my opinion means that the 25/50s were not cherrypicked and should have a better chance of getting to 66 than the DRx2 25/50.
The SLC2 should be basically identical to the SRx2 but without the built in flush logic.

My SRx2 25/50 runs at 33/66 just fine.

Thanks, that is very reassuring. I won't know the datecode on the chip until it arrives. If the SRC2 can run at 66 MHz, then the decision to stick with the SLC2 is obvious. Yes, I was using my SXL2-50 results to guess the overclock potential. Due to the cost of these chips, I probably won't buy more than 1 to experiment with.

My SRx2 25/50 runs at 33/66 just fine.

Did you try taking it north of 66 MHz to see where the road ends?

Nope. Because in the 90s I remember hearing stories of people blowing out the cache trying to overclock to 80MHz.
SRx2 is not exactly cheap anymore, so I count myself lucky that 66MHz is even possible.

I'll keep that in mind. Are, or did you, use a heatsink/fan on yours?

Any data on how far the SLC33 will go and if the Cyrix SLC33 is any better for OC than the TI SLC33 ?

Disappointing.
These cpu upgrade modules are flaky pain in the a.
So, you are confident that no combination of these switches unblocks the L1 cache ?

As mentioned before I am able to run sxl2-50 at up to 70mhz on these vlsi mobos, but the practical limit is around 64mhz.
Not sure this is a valid ref point for sxlc chips.

BTW, did you try to plug the memory card, even unpopulated to see if it ticks at all ?

The TI and Cyrix 486SLC chips should be identical except for the branding. Unless they were manufactured after 1993, I wouldn't expect them to overclock very well. Unless your your chip is exceptional, 40MHz is probably the best you'll be able to pull off.

I tried several combinations of DIP switch settings when I had the 3.6 V SXLC2 installed previously but didn't get it working properly. I didn't try all possible combinations as there are too many.

What memory card are you referring to? The only 286 memory card I own is sitting in my box of parts at AVICC in Oregon. I might have this box shipped up to me late January.

OK, so SLC33 at 40 Mhz or hopefully the SLC2 at 60 or 66 MHz. The SLC at 40 MHz would have a FSB of 20 MHz, while the SLC2 at 66 MHz only 16.7 Mhz, or 15 Mhz FSB if 60 Mhz CPU needed. I guess this makes my procuring 70 ns 256kx1 DIP ICs for the memory card irrelevant. The memory card comes with 100 ns chips, which I assume is fine at 0ws and 16.7 MHz ? By the way, the best price I could find for 70 ns 256kx1 was $0.60/each and at 60 ns, $1.20 each.

Forgot that the memcard was still not with you.
At up to 20mhz most peripherals do just fine.