Just out of curiousity: did you try one of those 387 emulators (em387, franke387, etc.)?
Yes, they just emulate the fpu in software, but I fuzzily remember that some programs did even run faster with them instead of using their own emulation code.

feipoa wrote on 2024-05-10, 00:58:

The mp3 player I'm using does require an FPU and will not operate in Windows 3.1 without one. It was noted above, if you try to run it without an FPU, you get "WinPlay3 requires a 486 Processor with a build in FPU".

It is only the 32-bit version of the same player from within NT 3.51 which tries to play without an FPU, and it does a horrible job at it. Since the "yo bro, no FPU" error does not appear in NT 3.51, it is likely that NT uses an FPU emulation feature which tricks WinPlay3 into thinking there's a real FPU present. Given the massive gaps between music playback, it is evident that WinPlay3 does require an FPU on this class of hardware and emulation alone is grosely insufficient.

Well, now I'm rabbit-holing with NT3.51 and seeing if it natively emulates a 387 when one isn't present. I can't get Quake to run in that PCem session, but Fractint will just immediately return to the command prompt if the FPU is disabled in the machine configuration in PCem, even if Fractint was previously configured to *not* use the FPU. It's weird! Maybe I'll install NT3.51 on real hardware (I have Ti486SXL-40 and a Cx387-DX, so testing it without/without FPU installed might tell me SOMETHING. Maybe.

Landmark 6.0 just crashes if the FPU is disabled in PCem. Weird.

But I shouldn't derail this thread, so I'll maybe start a new topic sometime...

Finally finished my SXL2-90 build based on the results from this thread.

Decided to go all SCSI (AHA-1520B):
ACARD 7722 with HDD firmware swap + custom 50-pin narrow SCSI PCB to terminate the high byte+ Sandisk 8 GB CF + CF-IDE slot bracket
...
NEC SCSI CD-ROM
...
ZIP 100 SCSI

Yeah, it would have been much easier to just use a SCSI2SD or subsequent clones, but I wanted a system with swap-able CF cards. And my other SXL2-66 system already uses SCSI2SD. Notice how only one CF-IDE slot-bracket is actually connected. The other CF-IDE slot bracket is just a holder for swapping CF cards. Otherwise, the secondary CF card could accidentally get used for another system or lost.

I ran out of slots, so decided to drill out one of the CF-IDE brackets to run the PS/2 mouse port.

Added 3C509B ethernet, serial/parallel, Cirrus Logic GD-5434 video, ES1868 sound w/Dreamblaster X2. Also added a Floppy-Max controller card for the LS-120 superdisk drive. This lets me boot to 1.44 MB and 120 MB floppies directly.

Unfortunately, the CPU location is directly under the HDD caddie. This means I would have to disassemble the whole system to swap the CPU. I decided to leave the 99 cent buck regulator in series with the interposer's linear VRM. If it fails, I can still directly connect the 12 V rail to the interposer. The buck is set to output 6.2 V, the linear is set to output 5.20 V.

The 461 northbridge was running at about 50 C with sustained uptime. This seemed a bit warm to me when compared to other 386 chipsets, thus I added a heatsink. The heatsink stays at under 32 C. Also, the DRAM gets to about 50 C in the centre after sustained uptime (30 minutes). This seems a bit hot for a 386. Not sure if this means anything?

Using the earlier version of the interposer it looks like?

So you have concluded that CF is the best storage option? No speed benefit from other options? Or couldn’t get scsi working?

feipoa wrote on 2024-05-17, 10:58:

Finally finished my SXL2-90 build based on the results from this thread. […]
Show full quote

Finally finished my SXL2-90 build based on the results from this thread.

Symphony_461_SXL2-90_outside.JPG
Symphony_461_SXL2-90_inside1.JPG
Symphony_461_SXL2-90_inside2.JPG
Symphony_461_SXL2-90_inside3.JPG
Symphony_461_SXL2-90_inside_dual_8GB_CF_cards.JPG

Decided to go all SCSI (AHA-1520B):
ACARD 7722 with HDD firmware swap + custom 50-pin narrow SCSI PCB to terminate the high byte+ Sandisk 8 GB CF + CF-IDE slot bracket
...
NEC SCSI CD-ROM
...
ZIP 100 SCSI

Yeah, it would have been much easier to just use a SCSI2SD or subsequent clones, but I wanted a system with swap-able CF cards. And my other SXL2-66 system already uses SCSI2SD. Notice how only one CF-IDE slot-bracket is actually connected. The other CF-IDE slot bracket is just a holder for swapping CF cards. Otherwise, the secondary CF card could accidentally get used for another system or lost.

I ran out of slots, so decided to drill out one of the CF-IDE brackets to run the PS/2 mouse port.

Added 3C509B ethernet, serial/parallel, Cirrus Logic GD-5434 video, ES1868 sound w/Dreamblaster X2. Also added a Floppy-Max controller card for the LS-120 superdisk drive. This lets me boot to 1.44 MB and 120 MB floppies directly.

Unfortunately, the CPU location is directly under the HDD caddie. This means I would have to disassemble the whole system to swap the CPU. I decided to leave the 99 cent buck regulator in series with the interposer's linear VRM. If it fails, I can still directly connect the 12 V rail to the interposer. The buck is set to output 6.2 V, the linear is set to output 5.20 V.

The 461 northbridge was running at about 50 C with sustained uptime. This seemed a bit warm to me when compared to other 386 chipsets, thus I added a heatsink. The heatsink stays at under 32 C. Also, the DRAM gets to about 50 C in the centre after sustained uptime (30 minutes). This seems a bit hot for a 386. Not sure if this means anything?

That's a cool mini tower case

feipoa wrote on 2024-05-17, 10:58:

Finally finished my SXL2-90 build based on the results from this thread.

That's amazing! Love the case badge, too!

Now let's see the benchmarks!

Sphere478 wrote on 2024-05-17, 14:07:

So you have concluded that CF is the best storage option? No speed benefit from other options? Or couldn’t get scsi working?

I am using SCSI through the CF card.

feipoa wrote on 2024-05-17, 21:07:

Sphere478 wrote on 2024-05-17, 14:07:

So you have concluded that CF is the best storage option? No speed benefit from other options? Or couldn’t get scsi working?

I am using SCSI through the CF card.

You mean the inverse of what your wrote, right?

"I am using the CF card through SCSI."

I meant the write 'I'm using SCSI with the CF card', thus eliminating directionality.

Confusion, please elaborate? 😀 pics, if you can.

You found a scsi to CF adapter?

Or you are using SCSI AND CF separately?

Or…?

I guess SCSI to IDE adapters exist, so there is that route…

Sphere478 wrote on 2024-05-18, 04:22:

Confusion, please elaborate? :-) pics, if you can. […]
Show full quote

Confusion, please elaborate? :-) pics, if you can.

You found a scsi to CF adapter?

Or you are using SCSI AND CF separately?

Or…?

I guess SCSI to IDE adapters exist, so there is that route…

Hmm... thought I was pretty clear previously. See:

feipoa wrote on 2024-05-17, 10:58:

Decided to go all SCSI (AHA-1520B):
ACARD 7722 with HDD firmware swap + custom 50-pin narrow SCSI PCB to terminate the high byte+ Sandisk 8 GB CF + CF-IDE slot bracket

Item is partially visible in the 4th photo, top right. Will present again: download/file.php?id=193088&mode=view

BlueSCSI (for Mac & Amiga) hardware-emulate a SCSI hard drive using a Rasperry Pi and SD Card.

Unsure if it's PC compatible. Unless all SCSI hdd's at the time were the same.

https://bluescsi.com/docs/

I've made all the new changes to my interposer and aim to order them in 5 June, and solder/test a ~week after.

Only missing the lines under the two ICS (ripple counter, quad NAND)....Could anyone with the Improve Technologies interposer identify those lines?

74HC4024D Ripple Counter - Is line 6 connected to line 1, anything else? Anything on these lines 1-5, 9, 11, 12?
74VHC0 Quad Nand - Where do these lines go? 3, 6, 8, 12, 13

@Feipoa
Assembled PC looks good indeed.
The scraggly cables scream 90ies - excellent touch.
Back panel appears rusty. Do you plan to do something about it, or leave as is ?

@MikeSG
What was the point of the flexible ribbon mod ?

The ribbon mod is from Improve-It Technologies. Mike is wanting to copy their design (cache invalidation circuit) and integrate it to his own SXL PCB.
…
The SCSI cable is way oversized because the one I wanted to use is just 4” too short. I’ve purchased a 6” SCSI extension cable to remedy that.
…
Yes, only the back of the case had some rust. I’ve removed most of it. I can improve it further with some #00 steel wool, but you will always see the the colour imperfections in the metal, I suspect even if I were to buff the non-rusted sections. Thus I have left it as is for now. Might address it again if I’m feeling like a perfectionist down the road. I’m sure you know this feeling well.

Speaking of which, does anyone know what the coating is that gets placed on the unpainted metal of these 90’s era chassis? If I recall right, I think the coating conductive.

I think they are zinc coated.

You could wax it perhaps to protect it cheaply. If you don’t want to electroplate it. Clear coating is also another option.

There are rust converters (acid) that makes rust black.

Just some ideas.

Sphere478 wrote on 2024-05-19, 17:27:

I think they are zinc coated. […]
Show full quote

I think they are zinc coated.

You could wax it perhaps to protect it cheaply. If you don’t want to electroplate it. Clear coating is also another option.

There are rust converters (acid) that makes rust black.

Just some ideas.

If anyone wants to test if it is indeed zinc, you can take a small piece and put it in hot vinegar. If it bubbles after some time in the solution, that is a fairly definitive test. Any weak acid will do, but vinegar is one acid that is fairly pure that most people have sitting around.

the fastest SXL2 rig to date.
i sense motivation bubbling up around here.
SXL2-100 rolls nice off the tongue, but my only sxl2-66 trips hard past 80mhz.
this will be effort ...

btw, did you narrow specific isa card that caps you at 11mhz bus ?

Zinc? Hmmm. Bolts that I see at the hardware store that are zinc coated tend to be pretty shinny. But the MB tray and chassis tend to have this dull grey coating. Maybe zinc plus another constituent.

The DTK PEM-4030Y was particular about SCSI cards from the onset. I could not, for example, load DOS SCSI drivers with the AHA-1542CP card installed, even with L1 disabled. I haven't run into that before. I was happy that the AHA-1522B worked.

I recall removing all ISA cards except for the GD-5434 + AHA-1522B, and still could not run at 15 MHz ISA bus. the next divisor down is at 11.25 MHz, which worked. Swapping the crystal oscillator from 90 to 80 MHz, I tried 13.3 MHz, but I don't recall the outcome. I think it worked, suggesting my bus limit is somewhere between 13.3x and 14.9x MHz. pshipkov's ability to get 25 MHz ISA bus speeds with SCSI cards astounds me. As we both now have the DTK PEM-4036YB board, I will probably try to replicate some of your success stories.