Palette change stalls are a problem that affects the whole nVidia line-up. It's not just DirectX or MOO2 or Riva128. The DOSbox forum is full of old threads with nVidia users complaining about slowness coming from fullscreen surface output. WinQuake has had stalls on picking up items.

Yes, the whole nvidia compatibility recommendation is a headscratcher to me.

From what I remember, the Windows version of Master of Orion 2 is buggy and unfinished.

Play the DOS version instead. It works fine even from within the Win9x DOS prompt.

leileilol wrote on 2025-03-29, 09:22:

Palette change stalls are a problem that affects the whole nVidia line-up. It's not just DirectX or MOO2 or Riva128. The DOSbox forum is full of old threads with nVidia users complaining about slowness coming from fullscreen surface output. WinQuake has had stalls on picking up items.

Yes, the whole nvidia compatibility recommendation is a headscratcher to me.

well, the DOS version behaves like this even on cards that don't show the transitions in the windows version. that's why i think this is actually intended.

Joseph_Joestar wrote on 2025-03-29, 09:31:

From what I remember, the Windows version of Master of Orion 2 is buggy and unfinished.

Play the DOS version instead. It works fine even from within the Win9x DOS prompt.

unfinished? i wouldn't say that, but buggy - definitely. v1.31 under DOS has all the exact same classic gameplay bugs, though.

reason i don't use the DOS version is that the mouse control feels horrible - probably forced acceleration from an internal driver that ignores everything else. the sound seems a bit more muffled, too. i've not had any luck running the 1.50 fan patch as it's only made for online releases, not CD installs, so have to mix and match a lot of files.

i've even tested the DOS version in a DOSBox window and the input lag was atrocious, not sure if i misconfigured something or this is what people really put up with. it was 1.31 and not 1.50 however.

auron wrote on 2025-03-29, 14:45:

unfinished? i wouldn't say that, but buggy - definitely. v1.31 under DOS has all the exact same classic gameplay bugs, though.

I don't remember the exact details, but there were reports that the Windows version might be missing some features from the 1.31 patch for the DOS release.

reason i don't use the DOS version is that the mouse control feels horrible - probably forced acceleration from an internal driver that ignores everything else. the sound seems a bit more muffled, too.

I haven't experienced any mouse pointer issues, but I'm usually playing MOO2 under the Win9x DOS prompt, with mouse acceleration disabled in Control Panel and using a modern Logitech G403 HERO USB mouse. As for sound being muffled, I only noticed that when selecting SBPro in setup. With SB16 selected, everything is much clearer.

i've even tested the DOS version in a DOSBox window and the input lag was atrocious, not sure if i misconfigured something or this is what people really put up with. it was 1.31 and not 1.50 however.

I don't use the fan patch since it makes a lot of balance changes in addition to fixing actual bugs. At least that was the case a few years back when I last checked the change logs.

Joseph_Joestar wrote on 2025-03-29, 16:06:

I don't remember the exact details, but there were reports that the Windows version might be missing some features from the 1.31 patch for the DOS release.

don't believe that is the case, the reason that people gave for using the DOS version was always higher stability under netplay. the official readme sort of treats the DOS version as a backup solution if not having a video card that supports directx.

the most annyoing bug is the "mark block stack size exceeded" crash, which is supposed to be a memory leak that accumulates from using the races screen - even the win95 version is very limited in terms of memory that it will allocate, regardless of RAM present, so the game will inevitably crash eventually. this bug is fixed in one of the 1.50 releases, and due to it being mentioned there, i'm assuming it's present in the 1.31 DOS version as well.

auron wrote on 2025-03-29, 16:29:

don't believe that is the case, the reason that people gave for using the DOS version was always higher stability under netplay. the official readme sort of treats the DOS version as a backup solution if not having a video card that supports directx.

The PC Gaming Wiki page states that the Windows version was unoptimized and laggy, especially in multiplayer. But I think that I saw player reports about it being worse than the DOS version in other places as well.

the most annyoing bug is the "mark block stack size exceeded" crash

I never had that issue, and I used to play a lot of MOO2 back in the day. That was with the DOS version patched to 1.31.

played two full games using the DOS version patched to 1.31, first thing to note is i was a bit imprecise in describing the mouse issue, as it does actually react to disabling acceleration, the problem is that it doesn't react to setting mouse sensitivity (mouse.com 8.20). the end result is practically the same, though - way too high sensitivity when using an optical mouse. however, i have found a workaround for this: using logitech's 6.50 driver, default sensitivity is much lower, making the game playable. the downside to this is that the driver refuses to load under win9x, saying that a mouse driver is already loaded, so this works only under real-mode DOS.

as expected i did get the aforementioned crash eventually, same message just on the DOS command prompt instead of a windows popup. didn't write down at what turn but well over 200 turns. according to moo2mod, frequently using the "report" function in the races screen to check other players' tech and if they are spying on you is what triggers this, it only happens on fairly long sessions without exiting the game, but it does happen eventually. this is a well-known bug, being mentioned on newsgroups posts back in 1997: https://groups.google.com/g/comp.sys.ibm.pc.g … c/c/zLJYwhoS5qY

otherwise, the DOS version has a slightly different feel, as one can tell the renderer works differently to the directdraw one - for instance, when certain screens are loading in there is a bit of interference, and on the council screen you can briefly see everything loaded in by one by one when looking closely, which is not the case in the windows version. in general, the DOS version seems ever so slightly less responsive than the windows one on the same hardware, and i've noticed a battle music loop issue on a rare occasion. the most major difference i've noticed is an issue with quick autocombat (hotkey Z) - unlike the windows version, this doesn't quickly skip over boarding/raiding prompts. this is also mentioned as being fixed in the 1.50 fan patch. overall, if just wanting to play the original 1.31 in single player on a win95 machine, i don't see any reason to go for the DOS version over the windows one.

auron wrote on 2025-03-31, 14:27:

otherwise, the DOS version has a slightly different feel, as one can tell the renderer works differently to the directdraw one - for instance, when certain screens are loading in there is a bit of interference, and on the council screen you can briefly see everything loaded in by one by one when looking closely, which is not the case in the windows version.

Assuming you were playing in pure DOS, and not from within the Windows prompt, did you have SMARTDRV.EXE loaded? I found that some later DOS games (released from 1995 onward) tend to work better with it enabled.

Graphics card and CPU speed might also be relevant, as I don't remember having any loading issues on my Pentium MMX 200 with a Matrox Millennium II.

p200mmx with a riva128 (elsa victory erazor) and matrox millennium, didn't notice much difference between those. the "interference" i was talking about is sort of an interlace pattern, visible when switching to the tech screen. but it's only visible for a split second, so a lot of people wouldn't even notice it.

smartdrv 16384 /x was loaded, which seemed to reduce loading times compared to not having it, but even with it loaded there were some odd cases of HDD related pauses, most noticeably when getting new tech - sometimes that would cause a 2-3 second load with the monotonous HDD crunching sound typical for DOS, similar to install/unpack operations, which i'm assuming is just due to how inefficient HDD access there is compared to 32-bit file/disk access under windows 95. the windows version of the game doesn't have that issue, running from the same harddrive.

the large smartdrive size also has almost no benefit compared to a smaller size, which i think even microsoft mentioned in some old KB article, and i wonder why exactly that is.

incidentally, when i run the moo2 dos version under windows 95 and click through the opening, it crashes with some red debug text visible. i recall this also happening when testing this before on another card than the riva128, so that seems to be a common problem. it did work when instead waiting for the opening menu animation to finish, however.

i also tested the homm2 windows version, which has similar fading transitions, but it didn't show any difference with drivers. tested with the gold disc with a 1998 date, not sure if this uses wing or directdraw though.