Windows 3.x

Post Reply New Topic

What do you guys think of Windows 3.x? Have any of you had any luck in getting these old OS's working on newer systems? Any suggestions? I'm new to pc emulation

If you are new to OS emulation, then start with VirtualBox:

https://www.virtualbox.org/


Why do you want to use Windows 3.x? There are many other operating systems that you could be taking a look at..

Use Dosbox and follow the instructions here

It would by far be the fastest method even though Dosbox is mostly used for gaming, I have ran Windows, Office, compilers and everything in between in Dosbox. The Dosbox Mega Build by H-A-L-9000 is grossly outdated but adds support for things like networking which Dosbox might be lacking. Might be worth playing with the SVN or Mega Build to see if it works for your needs.

Windows 3.x works pretty well in DosBox and PCem, though PCem also requires a copy of MS-DOS (or really any Win3.x compatible DOS) in order to run it, as well as an appropriate system BIOS rom and VGA BIOS rom. PCem is better if you need machine accurate emulation, but it's also more resource intensive. DosBox, on the other hand, is more of an HLE type emulator that does a fuzzier approximation of a DOS PC, and because of this you can get much more performance out of it on lower specs.

It's hard to say which program is easier to set up, because DosBox has been around a LOT longer and has more support available for it, but it also relies heavily on manually editing configuration files. PCem, at least on Windows, has more of a GUI, but it's also a much newer program that doesn't have nearly as much support, and in a way it is actually more technically oriented than DosBox. It's also horribly inefficient with system resources if all you want to do is play a few games.

Now, since someone mentioned Virtualbox, I'll quickly point out that it is aimed MUCH more at running modern OSes, and that its support for older, more obscure OSes is lacking at best. Also, it's not really oriented towards gaming, so graphics and sound support can be really iffy. Also, Virtualbox is not really an emulator in the traditional sense, it's a virtualization environment.

Emulation and virtualization are topics that simultaneously fascinate and infuriate me. One of the reasons I'm infuriated by the emulation community is because a good portion of it has this "good enough" mentality where they're willing to accept mediocre to poor emulation as long as they're able to say slowly boot Windows 95 on their smartphone, or play N64 games on their laptop with glaring graphical errors. But that's kind of going off topic.

Do you understand how emulation works on a programming level?

That good enough mentality brought us a lot of things that would perform too slow otherwise, why don't you write one if you think low-level, cycle-accurate emulation is easy to write.

I recommend starting here, I am working on an emulator now actually based off this dude but I am the worse programmer that lived, still its a fun pet project in boredom.

http://fms.komkon.org/EMUL8/HOWTO.html

I sort of do, but I can't claim to be an expert. I know that it involves basically emulating the instruction set, timings, and quirks of one hardware architecture, in order to run its software on another architecture. The difference between high level emulation and low level usually has to do with how fine grained your emulation algorithms are, like whether certain timings and limitations are imitated or not.

I understand good LLE emulators are hard to write, and that progress is being very slowly made. Not only that, but it also requires much more computing power than an almost-equivalent HLE emulator in a lot of cases. For instance, Snes9x will run most SNES games full speed on a Pentium III or 4 with almost no apparent inaccuracies, but Higan, a cycle-accurate SNES emulator, takes a modern multi gigahertz CPU to accomplish the same task, with nearly perfect accuracy.

Now with ZSNES, it's noted for its ability to run SNES games full speed on extremely modest hardware, with most games being perfectly playable on a Pentium classic. BUT, it is also riddled with all sorts of inaccuracies and shortcuts, many of which were unintentionally put to use by the romhacking community. As a result, there are a good number of SNES romhacks now, mainly for Super Mario World, which won't even work on a real SNES since they take advantage of tricks only present in ZSNES.

Now, don't get me started on the N64 emulation community, that's such a huge messy can of worms... xD At least after years of stagnation, people have finally started to get their act together to write emulators that aren't these hacky messes held together with bubble gum and duct tape.

Virtualization is your friend.

Ok this is actually a good point and I agree, speed hacks should never set the standard but what I don't agree with is their absence completely. I think this comes from a lack of appreciation for how emulators work, people don't understand the consequences to inaccurate code because they want something that just works. Maybe we should only have speed hacks and dirty code as a source code patch so those that'll know how to use it, knows the consequences such code would have in the community. It should be accuracy first and then speed hacks as far as a development outline and it seems all to often its speed hacks first and accuracy last.

I agree that speedhacks are necessary in some scenarios, without them most people wouldn't be able to emulate PS2 or Gamecube games, since it would take an insane amount of computing power to emulate them with cycle accuracy. Hacks of other types are also necessary for doing things like running games at higher resolutions than the original hardware. BUT, I think speedhacks should be something you can toggle on and off, and the core of an emulator should strive to be as accurate to the original hardware as possible.

Accuracy is necessary for preservation, but sometimes hacks are necessary for gameplay. Hacks shouldn't ever be the "standard" though.