I played through the original game back when it first came out on an SX-40. It was mostly fine, though with a little bit of slowdown in some of the more hectic battles.

It was definitely a little slow for WC:2 though.

cool! I will get the board now.

I also just remembered that I have a 286-25 stored somewhere. If the 386 ends up too fast I think the 286 should work.

Some motherboards allow you to select the bus frequency or have options in changing the oscillator crystal. This way you might have an option in clocking the SX40 at 33MHz if it turns out to be too fast. If you have a luxury motherboard (probably not, the SX40 was mostly used in budget systems) you might even have cache options. I used to run (not with Wing Commander, however) a SX16 with 16k cache which ran very nice; the cache helps 😉.

SX40 is roughly as fast as a DX20/25. That game's sweet spot is DX33 IIRC so expect slight slowdown here and there. WC2 will probably not be as smooth though.

appiah4 wrote on 2023-06-21, 10:41:

SX40 is roughly as fast as a DX20/25. That game's sweet spot is DX33 IIRC so expect slight slowdown here and there. WC2 will probably not be as smooth though.

Just wanted to point out that this difference in performance is because most SX motherboards don't have cache, not because of the CPU. If you compare it to an SX with cache they are much more closer.

We had a 386sx-25 with cache and it ran WC1 fine. When I upgraded to a 386DX-40, some stuff ran too fast, including the game loading intro.

There is a newer version made for Windows that won't have the speed issues. Not sure about the sound support though.

See here for the different releases.
http://www.wcnews.com/background/wing-commander-1.shtml

Works in my Made is USA 386sx-33 computer and in my 486dx4-100 computer too.
On the 486 I think I had to disable cache in bios to slow the computer down to 386 speeds.

See posts:
Re: Made in USA 386

konc wrote on 2023-06-21, 12:13:

appiah4 wrote on 2023-06-21, 10:41:

SX40 is roughly as fast as a DX20/25. That game's sweet spot is DX33 IIRC so expect slight slowdown here and there. WC2 will probably not be as smooth though.

Just wanted to point out that this difference in performance is because most SX motherboards don't have cache, not because of the CPU. If you compare it to an SX with cache they are much more closer.

A number of those later tiny baby AT 386SX boards that came with soldered AMD, had 1KB in chipset, not sure how much that helped.

clownwolf wrote on 2023-06-21, 07:21:

cool! I will get the board now.

I also just remembered that I have a 286-25 stored somewhere. If the 386 ends up too fast I think the 286 should work.

A 286-25, if running at 0 wait state, should be just as fast as a 386SX-25, if not ever so slightly faster. I can't quite get any of my 286 boards running at 25mhz 0ws, but at 22mhz 0ws it is 'playable' fast... At least on less complex missions.

Tip: Most classic ATs don't do copy (+map) the VGA BIOS in to RAM (Shadow Memory).
So they have to access/load video routines over ISA bus each time (in addition, there's the comparably high latency of EPROMs: ~60ns RAM vs ~150ns EPROM).

That's why there's historically a RAMBIOS utility supplied with the driver/utility disks that shipped with ISA VGAs.
If you use it, graphics performance might increase a bit.

Running ISA bus at 10 MHz also helps to reduce bottlenecks.
So far, all my old ISA VGAs (Paradise, OAK, Tseng etc) supported that when I tested them in a 10 MHz AT w/ 0 WS.

Edit: I forgot to mention. Wing Commander likes Expanded Memory, apparently.
So it's maybe helpful to enable Chipset EMS, a couple of ATs have that feature mentioned in CMOS Setup.

Edit: Animation is also affected by EMS. More information here:
https://www.wcnews.com/chatzone/threads/wc1-e … mparison.27730/
Wing Commander 1 on DOS machine, why does it say 'no expanded memory detected'?

Edit: Oh, and beware, EMM386 users: 386/486 processors are being slowed-down by V86, so you'll loose performance here.
So having Chipset EMS/UMBs is still relevant. To my knowledge, it were the later generation 486 cores which fixed the issue (maybe due to VME).