Adding Heretic / Hexen / Strife would require a lot of work, so it's not on my priority list. The same goes for extended WAD support (like Sigil, MBF, Boom, ...).

BTW I've added an entry in the wiki with all the ideas I want to include in FastDoom (this doesn't mean all these ideas will make it into FastDoom 😅):

https://github.com/viti95/FastDoom/wiki/Feature-wish-list

Congrats on v1.0, FastDoom took something I loved and made it significantly better 👍

ViTi95 wrote on 2024-09-11, 21:01:

The same goes for extended WAD support (like Sigil, MBF, Boom, ...).

Doesn't Sigil just require removing the visplane limit?

Sigil has problems with the visplane and drawseg limits. Also it requires some additional changes in order to work properly, just take a look:

https://www.doomworld.com/forum/post/1994132
https://www.doomworld.com/forum/post/1994456

Congrats on version 1.0! I really like fastdoom
I tried older versions on a IBM 386SLC 25mhz and my pentium 233 MMX. I really liked the vesa 640x480 support which i'm using on the mmx.
Draming on...... maybe someone someday comes with an algorithm that is like 100% more efficient and we could get high detail, full screen and 25FPS on a 386 😁

If that happens it would be like 8088MPH ... designed for specific set of hardware and any deviation could break it. Given that there's not a really fixed set of 386 hardware unless you do it for a PS/2 55SX or something, and all motherboards have different quirks, as do VGA implementations, so even specifying CPU, VGA chip and sound chip might only have it working on a percentage of them, then I think it's unlikely. You might find someone who can write it once, but he is unlikely to want to write it 20 different times plus for multiple variations, and then get slammed because only a couple are actually decent speed, and the others don't so so well. But then if you make it modular, use this code if this, use this code if that, and do other things that work in most cases.... well you get what we've got.

There is no way to run Doom full screen high detail on a 386 at 35fps unless someone creates an ISA 2D/3D accelerator 😅

Do you have instrumentation-based profiling working yet?
Carmack, or Abrash when talking about Carmack, mentioned coming up with perfect zero overdraw algorithm for Doom when developing Quake. On the other hand Duke3d is zero overdraw but much slower.

Yep, we have profiling support. It shows how many calls and time spent per function. The only downside is that it requires a Pentium CPU:

1# Build options for profiling (Pentium required)
2#CCOPTS = $(EXTERNOPT) -omaxtnrih -ol+ -oe=32 -zp4 -5r -ei -j -zq -zc -et $(WCCOPTS)

ViTi95 wrote on 2024-09-12, 16:09:

There is no way to run Doom full screen high detail on a 386 at 35fps unless someone creates an ISA 2D/3D accelerator 😅

Someone will produce a 386-1GHz to spite you 🤣

I bet an isa 2d/3d accelerator isn’t too far off. PicoGus is doing amazing things.

appiah4 wrote on 2024-09-13, 06:40:

ViTi95 wrote on 2024-09-12, 16:09:

There is no way to run Doom full screen high detail on a 386 at 35fps unless someone creates an ISA 2D/3D accelerator 😅

Someone will produce a 386-1GHz to spite you 🤣

Didn't IBM release a 386 variant running at 100MHz? Dunno if someone has tested FastDoom on that cpu 🙄

Blue Lightning BL3, got one in a system, I'll get round to it over winter maybe. It's got some 486 instructions and a 16kb cache too, so does about 1.5x 386 IPC or 2/3 to 3/4 486 IPC, so at 100Mhz is about equivalent to 486DX2/66

edit: for integer only that is. Has to rely on best 387s you can pair with it at FSB speed of 33 Mhz

I've uploaded a bugfix version (1.0.1), there was an error in the frameskipping algorithm for 35 fps modes:

https://github.com/viti95/FastDoom/releases/d … tDoom_1.0.1.zip

noshutdown wrote on 2024-09-11, 03:31:

i would like to see a benchmark of them against fastdoom on a 386 machine, using similar graphics settings, to find out how much the slow down is.

Doom8088 and FastDoom on a 386SX at 20 MHz.
Doom8088 runs at an effective resolution of 60x128 and has an average frame rate of 11.705.
FastDoom runs at an effective resolution of 64x128 and has an average frame rate of 15.377.
https://www.youtube.com/watch?v=Rr5ZmlPV3Rw

Frenkel wrote on 2024-09-15, 08:24:

Doom8088 and FastDoom on a 386SX at 20 MHz.
Doom8088 runs at an effective resolution of 60x128 and has an average frame rate of 11.705.

and here running on actual 8088 in IBM 5150 @19:40 https://www.youtube.com/watch?v=WJz977qkqsc at 0.5 fps? demonstrating that albeit cool academic exercise 8088 port is pointless.

rasz_pl wrote on 2024-09-15, 09:30:

and here running on actual 8088 in IBM 5150 @19:40 https://www.youtube.com/watch?v=WJz977qkqsc at 0.5 fps? demonstrating that albeit cool academic exercise 8088 port is pointless.

At this point in development that seems true but maybe there are additional optimizations and
compromises that can be made. I’ve seen posts here claiming that their hotrod 286’s are on par with lowend 386’s, maybe 8088Doom or RealDoom will atleast give 286 owners something to enjoy.

BinaryDemon wrote on 2024-09-15, 10:02:

At this point in development that seems true but maybe there are additional optimizations and
compromises that can be made. I’ve seen posts here claiming that their hotrod 286’s are on par with lowend 386’s, maybe 8088Doom or RealDoom will atleast give 286 owners something to enjoy.

That I'd like to see. This guy got a Harris 286 running at 37MHz and he says in the video description that FD is too slow still to even bother trying to benchmark. He gets Wolfenstein 3D running super smooth though!

You read it wrong, it's running 25Mhz and gives a 37Mhz Landmark score... but underneath he says something about a 33Mhz clocked one.

BitWrangler wrote on 2024-09-15, 18:26:

You read it wrong, it's running 25Mhz and gives a 37Mhz Landmark score... but underneath he says something about a 33Mhz clocked one.

Yeah, he has a follow up video where he takes a 286 to 33MHz. In any case, he wasn't able to get a recordable FPS in FD with that kind of horsepower idk if there's a 286 in the world that can do it. But would be very cool to see if it exists.