VOGONS


Reply 1260 of 1280, by theelf

User metadata
Rank Oldbie
Rank
Oldbie
ViTi95 wrote on 2026-04-19, 22:16:

I'll take a look, I'm sure I introduced a bug when I made changes to how the executable's texts are stored 😅

Hi!! how are you? do you think is possible to keep sprites in high resolution while use low mode?

Reply 1261 of 1280, by ViTi95

User metadata
Rank Oldbie
Rank
Oldbie

Hi! It's definitely possible to have different detail levels for sprites and world rendering, but it all comes down to having spare time to develop it, which sadly I don't have right now. Also, keep in mind that some walls are actually rendered as sprites, so this might cause some unexpected behavior.

https://www.youtube.com/@viti95

Reply 1262 of 1280, by ViTi95

User metadata
Rank Oldbie
Rank
Oldbie

New release, FastDoom 1.3.0

Changelog:

- Improved frame interpolation in uncapped framerate modes for smoother rendering
- Fixed frame skipping in uncapped framerate modes when FPS drops below 35
- Added integer scaling support (up to 5×) for backbuffered VESA modes, enabling any VESA resolution in FastDoom

https://github.com/viti95/FastDoom/releases/d … tDoom_1.3.0.zip

https://www.youtube.com/@viti95

Reply 1263 of 1280, by AncapDude

User metadata
Rank Member
Rank
Member

"Removed network gaming support" 😥

Reply 1264 of 1280, by schlang

User metadata
Rank Oldbie
Rank
Oldbie

maybe performance related?

PC#1: K6-III+ 400 | 512MB | Geforce4 | Voodoo1 | SB Live | AWE64 | GUS PNP Pro
PC#2: 486DX2-66 | 64MB | Riva128 | AWE64 | GUS PNP | PAS16
PC#3: 386DX-40 | 32MB | CL-GD5434 | SB Pro | GUS MAX | PAS16

Think you know your games music? Show us: viewtopic.php?f=5&t=37532

Reply 1265 of 1280, by appiah4

User metadata
Rank l33t++
Rank
l33t++
schlang wrote on 2026-09-14, 16:37:

maybe performance related?

I believe the doom netcode basically relies on the 25fps cap, I can see how that would cause issues with fastdoom..

Reply 1266 of 1280, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++
appiah4 wrote on 2026-09-15, 06:11:
schlang wrote on 2026-09-14, 16:37:

maybe performance related?

I believe the doom netcode basically relies on the 25fps cap, I can see how that would cause issues with fastdoom..

35 fps cap?
afaik other doom ports interpolate for higher framerates

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 1267 of 1280, by leileilol

User metadata
Rank l33t++
Rank
l33t++
AncapDude wrote on 2026-09-07, 02:39:

"Removed network gaming support" 😥

good tbh, would be a cheater's engine if it still had it given the texture/flat/lighting removal options available.

FUCK "AI". It is a tool of fascism. We do not need it. We do not use it.

Reply 1268 of 1280, by appiah4

User metadata
Rank l33t++
Rank
l33t++
rasz_pl wrote on 2026-09-15, 22:26:
appiah4 wrote on 2026-09-15, 06:11:
schlang wrote on 2026-09-14, 16:37:

maybe performance related?

I believe the doom netcode basically relies on the 25fps cap, I can see how that would cause issues with fastdoom..

35 fps cap?
afaik other doom ports interpolate for higher framerates

35, yes, sorry.

Reply 1269 of 1280, by matze79

User metadata
Rank l33t
Rank
l33t

What about FastHeretic ? 😀

Reply 1270 of 1280, by y2k se

User metadata
Rank Member
Rank
Member

Last night, while playing with the CPU renderer options, I noticed an issue in the upper-left when I select Pentium MMX. The first several pixels are black, then there is a very bright pixel, before showing normal. There is also a noticeable change in floor textures on the right-half of the screen. System is a Pentium 233 MMX and a Sierra Screamin' 3D video card and I am using the FDOOMVBR exectuable. I didn't notice this behavior on my P3 w/ Ti 4200, though that system hard-locks when I exit any of the executables.

file.php?mode=view&id=250597
file.php?mode=view&id=250596

Tualatin Pentium III-S 1.4, ASUS TUSL2-C, 512MB RAM, GeForce 4 Ti 4200, Voodoo2, SB Live!, 3Com 3C905C, 80GB IDE HDD, Dell 2001FP
P233MMX, Intel LT430TX, 64MB RAM, Sierra Screamin' 3D, Voodoo2, AWE64 Gold, 3Com 3C905B, 40GB IDE HDD, Viewsonic A75f

Reply 1271 of 1280, by 7F20

User metadata
Rank Member
Rank
Member
y2k se wrote on 2026-10-10, 13:51:

System is a Pentium 233 MMX and a Sierra Screamin' 3D video card and I am using the FDOOMVBR exectuable.

IIRC, the Rendition cards are notoriously buggy with just about everything, so I could see how that would just be a problem on its own. Seem to remember that Doom would just straight-up refuse to run Doom correctly on them, but maybe that's been patched idk.

Reply 1272 of 1280, by y2k se

User metadata
Rank Member
Rank
Member
7F20 wrote on 2026-10-14, 04:55:

IIRC, the Rendition cards are notoriously buggy with just about everything, so I could see how that would just be a problem on its own. Seem to remember that Doom would just straight-up refuse to run Doom correctly on them, but maybe that's been patched idk.

Original DOOM ran, just with poor performance compared to other video cards.

Tualatin Pentium III-S 1.4, ASUS TUSL2-C, 512MB RAM, GeForce 4 Ti 4200, Voodoo2, SB Live!, 3Com 3C905C, 80GB IDE HDD, Dell 2001FP
P233MMX, Intel LT430TX, 64MB RAM, Sierra Screamin' 3D, Voodoo2, AWE64 Gold, 3Com 3C905B, 40GB IDE HDD, Viewsonic A75f

Reply 1273 of 1280, by ViTi95

User metadata
Rank Oldbie
Rank
Oldbie
7F20 wrote on 2026-10-14, 04:55:
y2k se wrote on 2026-10-10, 13:51:

System is a Pentium 233 MMX and a Sierra Screamin' 3D video card and I am using the FDOOMVBR exectuable.

IIRC, the Rendition cards are notoriously buggy with just about everything, so I could see how that would just be a problem on its own. Seem to remember that Doom would just straight-up refuse to run Doom correctly on them, but maybe that's been patched idk.

I've never seen this bug before, it's really weird. Since FDOOMVBR renders the scene to a backbuffer in RAM and then copies it to VRAM, the issue must be happening during the backbuffer to VRAM copy. I haven't run into this problem on my Rendition Verite V2200, though to be fair I haven't tested the MMX codepath on that card yet. Can you try FDOOMVBD and FDOOM13H to see if this happens with either of those modes?

https://www.youtube.com/@viti95

Reply 1274 of 1280, by y2k se

User metadata
Rank Member
Rank
Member
ViTi95 wrote on 2026-10-14, 12:17:
7F20 wrote on 2026-10-14, 04:55:
y2k se wrote on 2026-10-10, 13:51:

System is a Pentium 233 MMX and a Sierra Screamin' 3D video card and I am using the FDOOMVBR exectuable.

IIRC, the Rendition cards are notoriously buggy with just about everything, so I could see how that would just be a problem on its own. Seem to remember that Doom would just straight-up refuse to run Doom correctly on them, but maybe that's been patched idk.

I've never seen this bug before, it's really weird. Since FDOOMVBR renders the scene to a backbuffer in RAM and then copies it to VRAM, the issue must be happening during the backbuffer to VRAM copy. I haven't run into this problem on my Rendition Verite V2200, though to be fair I haven't tested the MMX codepath on that card yet. Can you try FDOOMVBD and FDOOM13H to see if this happens with either of those modes?

Now I can't reproduce it. Maybe one confounding factor is I am using renutil to force a higher refresh rate for my CRT. I'm using renutil 4 1 to get 640x480 @ 85 Hz. But I would disregard since I apparently don't have reproducible steps.

Tualatin Pentium III-S 1.4, ASUS TUSL2-C, 512MB RAM, GeForce 4 Ti 4200, Voodoo2, SB Live!, 3Com 3C905C, 80GB IDE HDD, Dell 2001FP
P233MMX, Intel LT430TX, 64MB RAM, Sierra Screamin' 3D, Voodoo2, AWE64 Gold, 3Com 3C905B, 40GB IDE HDD, Viewsonic A75f

Reply 1275 of 1280, by BinaryDemon

User metadata
Rank Oldbie
Rank
Oldbie

I was betting ViTi95 had left in some debug indicators that most people probably wouldn’t notice. Guess it’s not that.

Reply 1276 of 1280, by Darmok

User metadata
Rank Newbie
Rank
Newbie

The other day, I pulled my 386 out of storage with the goal of playing DOOM using the latest version of FastDoom (v1.3.0). To start, I decided to run some benchmarks in both 16-bit and 8-bit ISA bus modes. The 8-bit mode is particularly important to me due to a well-known limitation in the AT architecture, which forces an ISA VGA graphics card to operate in 8-bit mode when used alongside an 8-bit Hercules card—a requirement for enabling the `-hercmap` mode. The test results for all video modes supported by the OTI087 are summarized in a table.

The contents of `bench2.bat` are as follows: `fdoom(xx).exe -iwad doom.wad -benchmark file demo1 bench\arch.bnc`. Sound effects (SFX) are enabled, as I am interested in real-world in-game performance.

file.php?mode=view&id=252008

For my configuration, the best results were achieved using fdoom13h.exe, in both 16-bit and 8-bit modes. However, these results were obtained with `screenblocks` set to 10 and detail level set to "high"—which is too demanding for such a low-end computer. Is there any way to instruct the benchmark program to use a lower `screenblocks` value? I know I could run the tests manually, but the automated sweep through different architectures is much more convenient. I could re-run the tests and post the results here for comparison.

What is the refresh rate of the Hercules screen in 'hercmap' mode? Would it be possible to add a parameter to reduce this refresh rate—perhaps to improve performance in that mode?

Test computer configuration: 386DX@45, 32 MB RAM, 128 kB cache, opti495slc chipset, ISA bus frequency 15 MHz. SVGA OTI087 1 MB, Hercules clone.

For reference – OTI087 performance in the vedspeed 3.0 benchmark:

Spoiler

16-bit mode
16444 Bytes per millisecond Normal Ram
6736 Bytes per millisecond 27.56KHz 70.11Hz 40x25 (text)
6736 Bytes per millisecond 28.01KHz 70.08Hz 80x25 (text)
6736 Bytes per millisecond 28.01KHz 70.08Hz 80x50 (text)
6729 Bytes per millisecond 28.02KHz 70.13Hz 320x200x4 (CGA)
6729 Bytes per millisecond 28.02KHz 70.13Hz 640x200x2 (CGA)
6725 Bytes per millisecond 28.02KHz 70.13Hz 640x200x16 (EGA)
6724 Bytes per millisecond 24.51KHz 70.13Hz 640x350x16 (EGA)
6736 Bytes per millisecond 28.02KHz 70.11Hz 320x200x16 (EGA)
6712 Bytes per millisecond 28.02KHz 70.13Hz 320x200x256 (VGA/MCGA)
6723 Bytes per millisecond 28.76KHz 59.98Hz 640x480x16 (VGA)
6723 Bytes per millisecond 28.76KHz 59.97Hz 640x480x2 (VGA/MCGA)
6725 Bytes per millisecond 43.38KHz 72.35Hz 800x600x2 (S-VGA)
6725 Bytes per millisecond 43.38KHz 72.35Hz 800x600x16 (S-VGA)
6722 Bytes per millisecond 27.75KHz 69.42Hz 640x400x256 (S-VGA)
6722 Bytes per millisecond 34.90KHz 72.81Hz 640x480x256 (S-VGA)
6690 Bytes per millisecond 43.38KHz 72.35Hz 800x600x256 (S-VGA)
6634 Bytes per millisecond 53.71KHz 70.03Hz 1024x768x2 (S-VGA)
6634 Bytes per millisecond 53.71KHz 70.03Hz 1024x768x16 (S-VGA)
6722 Bytes per millisecond 28.76KHz 59.98Hz 360x480x256 (VGA)
5121 Bytes per millisecond 53.71KHz 70.03Hz 1024x768x256 (S-VGA)

8-bit mode
16444 Bytes per millisecond Normal Ram
2892 Bytes per millisecond 26.84KHz 70.10Hz 40x25 (text)
2884 Bytes per millisecond 28.01KHz 70.10Hz 80x25 (text)
835 Bytes per millisecond 18.42KHz 49.83Hz 80x25 (B&W text)
2884 Bytes per millisecond 28.01KHz 70.10Hz 80x50 (text)
2882 Bytes per millisecond 28.02KHz 70.13Hz 320x200x4 (CGA)
2882 Bytes per millisecond 28.02KHz 70.13Hz 640x200x2 (CGA)
2882 Bytes per millisecond 28.02KHz 70.13Hz 640x200x16 (EGA)
2878 Bytes per millisecond 24.52KHz 70.13Hz 640x350x16 (EGA)
948 Bytes per millisecond 18.80KHz 50.86Hz 720x348x2 (Hercules)
2884 Bytes per millisecond 28.02KHz 70.13Hz 320x200x16 (EGA)
2881 Bytes per millisecond 28.02KHz 70.13Hz 320x200x256 (VGA/MCGA)
2881 Bytes per millisecond 28.76KHz 59.97Hz 640x480x16 (VGA)
2881 Bytes per millisecond 28.76KHz 59.98Hz 640x480x2 (VGA/MCGA)
2881 Bytes per millisecond 43.38KHz 72.34Hz 800x600x2 (S-VGA)
2880 Bytes per millisecond 43.38KHz 72.35Hz 800x600x16 (S-VGA)
2881 Bytes per millisecond 27.75KHz 69.45Hz 640x400x256 (S-VGA)
2881 Bytes per millisecond 34.90KHz 72.81Hz 640x480x256 (S-VGA)
2881 Bytes per millisecond 43.38KHz 72.35Hz 800x600x256 (S-VGA)
2881 Bytes per millisecond 53.71KHz 70.03Hz 1024x768x2 (S-VGA)
2881 Bytes per millisecond 53.71KHz 70.03Hz 1024x768x16 (S-VGA)
2881 Bytes per millisecond 28.76KHz 59.98Hz 360x480x256 (VGA)
2403 Bytes per millisecond 53.71KHz 70.02Hz 1024x768x256 (S-VGA)

Last edited by Darmok on 2026-11-02, 07:32. Edited 1 time in total.

Reply 1277 of 1280, by rasz_pl

User metadata
Rank l33t++
Rank
l33t++
Darmok wrote on 2026-11-01, 16:09:

The 8-bit mode is particularly important to me due to a well-known limitation in the AT architecture, which forces an ISA VGA graphics card to operate in 8-bit mode when used alongside an 8-bit Hercules card—a requirement for enabling the `-hercmap` mode.

Indeed MEMCS16# has 128KB granularity, cant selectively enable just for 0xc000 segment.
Wow does that work in practice? Do you have to jumper VGA into 8bit mode to have Herc work? plug vga into 8bit port?

https://github.com/raszpl/sigrok-disk FM/MFM/RLL decoder
https://github.com/raszpl/FIC-486-GAC-2-Cache-Module (AT&T Globalyst)
https://github.com/raszpl/386RC-16 ram board
https://github.com/raszpl/Zenith_ZBIOS Zenith Z-386 MFM-300 ZBIOS disassembly

Reply 1278 of 1280, by Darmok

User metadata
Rank Newbie
Rank
Newbie

Many SVGA cards feature an automatic detection mechanism that determines the ISA bus bit-width upon computer startup. By default, a VGA card initializes in 8-bit mode. During the VGA BIOS initialization process, the BIOS code checks for the presence of an MDA or Hercules adapter; if one is detected—both of which are 8-bit devices—the VGA card remains in 8-bit mode. If no such monochrome adapter is found, the VGA card switches to 16-bit mode. The 8-bit or 16-bit mode is controlled by the contents of a specific register within the SVGA chipset. For my OTI-087-based card, I wrote a couple of simple utilities that allow for switching between 8-bit and 16-bit modes at any time. When I don’t need to use a Hercules adapter, I select 16-bit mode to avoid performance penalties during video memory write operations; if I need to use monochrome modes, I can simply switch back to 8-bit mode.

If a VGA card is set to 16-bit mode, a Hercules adapter will still display characters in text mode, though without color attributes, while in graphics mode, only every second line is rendered—at least, this is the case with my particular Hercules clone. This occurs because the VGA card governs the bit-width of the ISA bus cycle. Since chipsets typically introduce additional wait states during 8-bit bus cycles, it is possible that some monochrome adapters may not function at all when the bus is set to 16-bit mode.

P.S. I might be mistaken, and perhaps the Hercules card only displays every second pixel when in 16-bit graphics mode. It’s been many years since I last tried using a graphics mode in a 16-bit configuration—back when I was experimenting with switching a VGA adapter to a 16-bit bus. I simply don’t recall; my monochrome monitor is broken beyond repair, and I’m currently waiting for a TTL-to-VGA converter to arrive.

Reply 1279 of 1280, by Darmok

User metadata
Rank Newbie
Rank
Newbie

I’ve figured out how to set the `screenblocks` parameter for the benchmark program. In the `\bench\arch.bnc` file, the second value for the desired `size` (`screenblocks`) must be specified as 3 units lower than the actual target value.
I ran a few more tests at a screen refresh rate that I find acceptable. While the `realtics` values ​​vary slightly across multiple test runs, this does not affect the resulting FPS. The results are presented in the table.

file.php?mode=view&id=252231

For the "detail high", "size"8 mode , the most optimal configurations are: `fdoom13h` , `cy386` at 16-bit bus, and `fdoom` , `cy486` at 8-bit bus.
For the "detail low", "size"11 mode , the best options are: `fdoom` , `cy386` at either 16-bit or 8-bit bus.
As we can see, to achieve maximum performance, it is necessary to run tests using the same settings that will be used during actual gameplay.