Oh you mean Spectres

I tried out v0.4 and v0.5 on my 486 DX4 120MHz/VLB, and while they both have a slighter lower realtics count than v0.3 the difference is very small (3612 vs 3601). It is very warm today though, compared to when I benchmarked v0.3 😜

HandOfFate wrote on 2020-08-14, 17:11:

I tried out v0.4 and v0.5 on my 486 DX4 120MHz/VLB, and while they both have a slighter lower realtics count than v0.3 the difference is very small (3612 vs 3601). It is very warm today though, compared to when I benchmarked v0.3 😜

I guess you are using the 486 build, can you test the 386 build please?

ViTi95 wrote on 2020-08-15, 22:11:

HandOfFate wrote on 2020-08-14, 17:11:

I tried out v0.4 and v0.5 on my 486 DX4 120MHz/VLB, and while they both have a slighter lower realtics count than v0.3 the difference is very small (3612 vs 3601). It is very warm today though, compared to when I benchmarked v0.3 😜

I guess you are using the 486 build, can you test the 386 build please?

Sure! I'm away for a few days but I'll try it when I'm back.

New video!

This is showing my new visplane renderer, which is much faster but only allows flat colors to be drawed. The old one has been modified to reenable diminished lightning so everyone is happy now 😁. Every visplane is stored in a set of columns, and latter transformed into spans, merged and then drawed as spans. The new renderer just renders the columns with flat colors, avoiding the process of transforming the columns into spans and then drawing with the span functions. Drawing a span in ModeX is much slower than a column, so this is as fast as it could get.

https://www.youtube.com/watch?v=Oi20FDX8V9w

Now I finally got to test fastdoom.

This time the test hardware was a Digital Celebris 560, at least its innards:
- VLSI chipset
- Pentium 60
- 24MiB of FPM RAM
- Onboard S3 Vision 864 - together with the VLSI chipset it yields a (n un)healthy SiS496 PCI level of VGA performance. So, it's ~12k in Landmark.
- Gravis Ultrasound ACE with 1 MiB of RAM (the smallest, lightest GUS to stick in the riser)

Unfortunately FastDoom failed at this last point. SFX were OK but the music lacked most notes/instruments.

As for the speed, it gained like 20% - frame rate went from 38 up to 45 fps. Which means, there are still point in the game where this hardware still can't do constant 35fps. Damn, it's 1993. This hardware was so new you probably couldn't even buy it, let alone afford it. Damn you, John Carmack 😁

Oh, and I got a freeze on map15 of Doom2 with the 486 Ultimate Doom exe.

The difference between the 386 and 486 version was rather small, btw, but the 486 one is still a teeny bit faster.

ViTi95 wrote on 2020-08-15, 22:11:

HandOfFate wrote on 2020-08-14, 17:11:

I tried out v0.4 and v0.5 on my 486 DX4 120MHz/VLB, and while they both have a slighter lower realtics count than v0.3 the difference is very small (3612 vs 3601). It is very warm today though, compared to when I benchmarked v0.3 😜

I guess you are using the 486 build, can you test the 386 build please?

• Fastdoom 0.4 (386 exe): 3607 realtics (37.483 fps)
• Fastdoom 0.5 (386 exe): 3571 realtics (37.861 fps)
• Fastdoom 0.5 (486 exe): 3570 realtics (37.872 fps)

alvaro84 wrote on 2020-08-22, 17:42:

Now I finally got to test fastdoom. […]
Show full quote

Now I finally got to test fastdoom.

This time the test hardware was a Digital Celebris 560, at least its innards:
- VLSI chipset
- Pentium 60
- 24MiB of FPM RAM
- Onboard S3 Vision 864 - together with the VLSI chipset it yields a (n un)healthy SiS496 PCI level of VGA performance. So, it's ~12k in Landmark.
- Gravis Ultrasound ACE with 1 MiB of RAM (the smallest, lightest GUS to stick in the riser)

Unfortunately FastDoom failed at this last point. SFX were OK but the music lacked most notes/instruments.

As for the speed, it gained like 20% - frame rate went from 38 up to 45 fps. Which means, there are still point in the game where this hardware still can't do constant 35fps. Damn, it's 1993. This hardware was so new you probably couldn't even buy it, let alone afford it. Damn you, John Carmack 😁

Oh, and I got a freeze on map15 of Doom2 with the 486 Ultimate Doom exe.

The difference between the 386 and 486 version was rather small, btw, but the 486 one is still a teeny bit faster.

Thanks for testing! GUS support is hard, i don't own a GUS so I rely on PCEM to test these cards. I'm no expert in sound hardware, but i'm learning how the Apogee Sound System works and trying to fix it accordingly. In the future versions I'll try to add native support for OPL2LPT and OPL3LPT, and add more sound cards that are currently supported in the library (but not in DOOM).

The next version will come with a fix for fast systems, it will be much smoother with fast processors (this includes anything that can get more than 35 fps, 486s > 50 MHz).

HandOfFate wrote on 2020-08-22, 18:28:

[…]
Show full quote

ViTi95 wrote on 2020-08-15, 22:11:

HandOfFate wrote on 2020-08-14, 17:11:

I tried out v0.4 and v0.5 on my 486 DX4 120MHz/VLB, and while they both have a slighter lower realtics count than v0.3 the difference is very small (3612 vs 3601). It is very warm today though, compared to when I benchmarked v0.3 😜

I guess you are using the 486 build, can you test the 386 build please?

• Fastdoom 0.4 (386 exe): 3607 realtics (37.483 fps)
• Fastdoom 0.5 (386 exe): 3571 realtics (37.861 fps)
• Fastdoom 0.5 (486 exe): 3570 realtics (37.872 fps)

Also thanks for testing! New version will come with more optimizations and an important fix that will make your 486 much smoother 😁

EDIT:

FastDoom 0.6!!

• Added option "-nomelt" to avoid melting transition while loading a new levels. For 386 pc's where this functionality it's really slow.
• Fixed bug that made the framerate choppy with fast processors
• Lot's of internal optimizations (non-recursive R_RenderBSPNode, many ideas from GBADoom). This version is faster even without lowering image quality
• Flat surfaces now have diminished lightning enabled or not. The option "-flatsurfaces" has it enabled, and the new option "-flattersurfaces" has it disabled. The option "-flattersurfaces" is MUCH faster than the previous one.

Grab it here: https://github.com/viti95/FastDoom/releases/tag/0.6

can you fix the visplanes problem? so sigil wad should be playable on this build

Here are my timedemos for v0.6. I was thinking that my randomly chosen demo (demo3) maybe wasn't the most intensive one, so also tried demos 1 and 2 (Ultimate Doom v1.9).

I also took the opportunity to run the game in real mode without any driver TSRs. That's not my default boot mode so I probably wouldn't play the game that way, but just to see how high the FPS can be pushed 😀

My sound card (ESS AudioDrive 1869F) works without TSRs so the game was still playing audio.

Fastdoom 0.5, 486 exe:
demo1: 1820 realtics (32.884 fps)
demo2: 2350 realtics (34.955 fps)
demo3: 3605 realtics (37.504 fps)

Fastdoom 0.6, 486 exe:
demo1: 1808 realtics (33.102 fps)
demo2: 2335 realtics (35.179 fps)
demo3: 3572 realtics (37.851 fps)

---

Fastdoom 0.5, 386 exe, real mode, no driver TSRs:
demo1: 1746 realtics (34.277 fps)
demo2: 2253 realtics (36.460 fps)
demo3: 3476 realtics (38.896 fps)

Fastdoom 0.6, 386 exe, real mode, no driver TSRs:
demo1: 1723 realtics (34.735 fps)
demo2: 2206 realtics (37.236 fps)
demo2: 3406 realtics (39.695 fps)

---

Fastdoom 0.5, 486 exe, real mode, no driver TSRs:
demo1: 1767 realtics (33.870 fps)
demo2: 2280 realtics (36.028 fps)
demo3: 3508 realtics (38.541 fps)

Fastdoom 0.6, 486 exe, real mode, no driver TSRs:
demo1: 1754 realtics (34.121 fps)
demo2: 2256 realtics (36.411 fps)
demo3: 3462 realtics (39.053 fps)

You can see that demo3 is getting the highest FPS so maybe it was indeed a bit 'light'. But yes, v0.6 is adding about 0.5 FPS to the game in all cases 😀

I'm wondering why the 386 executable is faster. Do you know why?

Well I don't really know why the 386 executables are usually faster than the 486 ones. The only difference between them is that the makefile of the project tells the compiler (Open Watcom 1.9) to optimize for the specified processor. Also the 486 executable usually weighs 8~12 Kb more than the 386 executable.

Regardles the visplane limit, I'm trying to port the MBF unlimited visplane implementation into FastDoom, but it is giving me lot's of headaches.

FastDOOM 0.6 486, Doom 1.9 Shareware, Demo3
Asus VL/I-486SV2GX4, DX2 @ 66MHz, 1MB L2 Cache, 24MB RAM, 1GB CF HDD, ARGUS Prototype Rev. 02 #0
Cirrus Logic 5428 VLB -> 43,023fps
Ati Mach32 VLB -> 42,851fps

I think its faster with 386 Code because of the Cache Size of the 486.

Hm, the current release (386) crashes a lot for me, e.g. when I run -potato or some of the other options. Without options or with -fps is okay though.

EDIT: I correct myself. It just seems I can't load any saved games!

root42 wrote on 2020-08-24, 20:18:

Hm, the current release (386) crashes a lot for me, e.g. when I run -potato or some of the other options. Without options or with -fps is okay though.

EDIT: I correct myself. It just seems I can't load any saved games!

You're right about savegames, they are broken in version 0.6 🙁 I'll add an issue to the GitHub and fix it for the next release. I don't have much time to develop and test all the functionality, and sometimes this happens. Also, what specs have your 386? It's important for me that FastDoom works in all PC's without problems.

shock__ wrote on 2020-08-24, 14:59:

FastDOOM 0.6 486, Doom 1.9 Shareware, Demo3 Asus VL/I-486SV2GX4, DX2 @ 66MHz, 1MB L2 Cache, 24MB RAM, 1GB CF HDD, ARGUS Prototyp […]
Show full quote

FastDOOM 0.6 486, Doom 1.9 Shareware, Demo3
Asus VL/I-486SV2GX4, DX2 @ 66MHz, 1MB L2 Cache, 24MB RAM, 1GB CF HDD, ARGUS Prototype Rev. 02 #0
Cirrus Logic 5428 VLB -> 43,023fps
Ati Mach32 VLB -> 42,851fps

Wow that's an impressive performance. Maybe the 1Mb L2 Cache it's helping a lot, my 486DX@50 isn't that fast even with VLB and tight RAM and cache timings (256Kb).

ViTi95 wrote on 2020-08-24, 21:12:

root42 wrote on 2020-08-24, 20:18:

Hm, the current release (386) crashes a lot for me, e.g. when I run -potato or some of the other options. Without options or with -fps is okay though.

EDIT: I correct myself. It just seems I can't load any saved games!

You're right about savegames, they are broken in version 0.6 🙁 I'll add an issue to the GitHub and fix it for the next release. I don't have much time to develop and test all the functionality, and sometimes this happens. Also, what specs have your 386? It's important for me that FastDoom works in all PC's without problems.

I didn't update my signature in some time. I am currently running an OPTI 486 ISA mainboard, 256KB Cache, 16MB RAM with a Tseng ET4000AX. I get around 12-13FPS usually. Spikes at around 19-20FPS with few objects around.

And what soundcard are you using @root42? Also what exact model of CPU? I don't have a Tseng ET4000AX so i cannot try with that card, maybe other members can confirm if this happens with this specific video card, or it is another problem.

SnarkBarker for the sound card. Intel 486DX33 CPU. As I said the problem seems to be only when loading games. But I will do some more tests some other time. To see which options work. I could play normally quite for a while!

Very interesting - I also did work on Nuke's source and it's such an awesome base.
Props for enhancing the original Doom.

shock__ wrote on 2020-08-24, 14:59:

FastDOOM 0.6 486, Doom 1.9 Shareware, Demo3 Asus VL/I-486SV2GX4, DX2 @ 66MHz, 1MB L2 Cache, 24MB RAM, 1GB CF HDD, ARGUS Prototyp […]
Show full quote

FastDOOM 0.6 486, Doom 1.9 Shareware, Demo3
Asus VL/I-486SV2GX4, DX2 @ 66MHz, 1MB L2 Cache, 24MB RAM, 1GB CF HDD, ARGUS Prototype Rev. 02 #0
Cirrus Logic 5428 VLB -> 43,023fps
Ati Mach32 VLB -> 42,851fps

Those are impressive results. This 486 I'm testing on doesn't have any L2 cache.

If you have the time, could you run it again with cache disabled? I'm curious to see if that would bring the FPS to roughly my machine's level