Jump to content
Search In
  • More options...
Find results that contain...
Find results in...

deathz0r

Members
  • Content count

    3737
  • Joined

  • Last visited

About deathz0r

  • Rank
    Forum Staple

Recent Profile Visitors

3240 profile views
  1. deathz0r

    Doom Streams

    going through more NanoWadMo WADs, starting with The Lost Magic by Vortale https://twitch.tv/deatsaw
  2. The amount of RAM is irrelevant for SDRAM, since the smallest SDRAM sticks I'm aware of are 16MB - the more important reason for why SDRAM (late-socket 7 motherboards until early-Pentium 4/Athlon XP) is preferred over EDO (late-486 until mid-socket 7) or FPM (mid-286 until mid-socket 7 - both EDO and FPM can be found as 72-pin sticks that are mechanically and electrically compatible) RAM is simply because the thoroughput of SDRAM is significantly higher than both EDO and FPM. Additionally, the L1 cache of socket 7 CPUs has anywhere between twice to four times the thoroughput of 486 CPUs and since this was still an era where L2 cache was on the motherboard, there were improvements with the mobo chipsets combined with lower latency and increased thoroughput of newer cache chips which also significantly improve the performance. Different mobo chipsets can also impact performance, but I don't have enough mobos sitting around to get data that granular. (also I killed the socket 7 mobo that I did the playtesting from Z1M3 onwards, and prices for good socket 7 mobos are stupid on eBay) So why is that relevant? For shits and giggles I started doing my playtesting playthrough on a 486 DX4-100 setup (with a good ol' VLB video card and FPM RAM) and I got through Z1M1 and Z1M2, noticing that there was increasing amount of stutter the further I progressed. When I was testing Z1M3, I was noticing atrocious performance with vanilla in even when using a 1.4Ghz Pentium 3 setup (quick sidenote: L2 cache is now baked in the CPU and running at full speed at this point, so there's no longer a performance penalty between L2 cache on a mobo and a CPU) where it was getting choked hard and continuously stuttering non-stop. Something didn't seem right about that and I told essel about it. A bit of optimising with animated textures happened and suddenly Z1M3 became a lot more playable. I decided to go back and check the start of Z1M2 (as there's noticable delay between being teleported into the main map) and figured out from that what was the minimum CPU and type of RAM required - Z1M2 being a good indicator because it has a ton of midtex work at the start of the main map and 486 CPUs choke on heavy midtex usage, 386 CPUs choke even harder relative to their performance. My estimate for 32MB is based on the testing before animated texture optimisation - I suspect 16MB would have a small amount of stutter in certain areas. I have picked up a "cheap" socket 7 mobo a couple days ago which has both SDRAM and EDO/FPM slots, so I can do a bit more testing for whether there's a difference in the stuttering for texture/flat heavy areas like Z1M3. I'll get to that soon(tm)
  3. This definitely happens with PrBoom+/DSDA-Doom when uncapped framerate is enabled - that's a side effect with how they handle the interpolation between gametics which is very noticeable with any instant-speed movement.
  4. Having tested this on vanilla with a REAL 90s PC, I can say that you need a Pentium MMX @ 166Mhz and SDRAM to run this somewhat smoothly but there's only so much you can do to get around the 8MB heap size limit of vanilla, so there's still unavoidable stutter especially when there's lots of textures and flats. Using doom32 and having sufficient RAM (most likely 32MB) will most likely remove the main performance bottleneck, but I was interested in making sure that KDiZDiZD [WORKS IN INTENDED SOURCE PORT] and I'm glad that I got to see cool shit going on and helped point out performance bottlenecks. It makes me want to see FastDoom's dehacked support get improved massively, because I'm curious to see it working at good speed on a 486DX4-100.
  5. deathz0r

    Doom Streams

    a comfy and casual UV continuous playthrough of Hell Revealed, part two (requested by @SaladStream ) https://twitch.tv/deatsaw
  6. deathz0r

    Doom Streams

    a comfy and casual UV continuous playthrough of Hell Revealed, part 1 (requested by @SaladStream https://twitch.tv/deatsaw
  7. deathz0r

    Does the sidedef limit matter? (Boom Format)

    As boris said, if DB2-derived editors still hit over 65,535 sidedefs after compression then an error message will pop up saying that it's unable to save. I took this screenshot while working on Poogers MAP29: It's also possible to hit the vertices limit (65,535) before hitting the linedef limit if you have an obscene amount of free-floating lines, such as very complex voodoo scripts for example.
  8. deathz0r

    Doom Streams

    a comfy and casual UV saveless pistol-start playthrough of Valiant, MAP26-?? https://twitch.tv/deatsaw
  9. deathz0r

    Doom Streams

    a comfy and casual UV saveless pistol-start playthrough of Valiant, MAP20-?? (requested by @Daerik back in September '21 or something) https://twitch.tv/deatsaw
  10. deathz0r

    ENDOOM Appreciation Thread

    I deliberately used Hocus Pocus' exit splash as a template for the ENDOOM for POOGERS, which was fun recreating from scratch: Looks like the vertical rainbow was removed sometime between Hocus Pocus and ROTT. I also like HR2's ENDOOM as well:
  11. deathz0r

    POOGERS.wad [Final release]

    Hey look it's on /idgames now, the final final final release! MediaFire mirror Changelog:
  12. deathz0r

    POOGERS.wad [Final release]

    Remember when I said final release in that last post? Yeah. POOGERS v1.1 (link removed as of 11/06/2022) There was a few issues brought up and since there was a delay in a response from the /idgames upload, this is effectively a hotfix release which would only cause a desync if anyone had made a max demo of MAP29 that involved player movement input on two scrollers in secret/easter egg areas, and I don't see any demos of MAP29 existing :) I've checked all existing demos and confirmed there is no desync for those. This version has also been uploaded to /idgames since there is currently no existing download link. Changelog:
  13. deathz0r

    POOGERS.wad [Final release]

    POOGERS final release! (link removed as of 03/06/2022) Now with a credits map (MAP36, after MAP30 on UMAPINFO ports), an ENDOOM lump and a ton of other changes under the hood! This version has also been uploaded to /idgames, so get hyped and check out the good shit brother today!
  14. deathz0r

    Doom Streams

    working on the credits map for POOGERS (MAP36) and making cool shit ™️ with voodoo dolls again, let's finish this today and get POOGERS uploaded to /idgames! https://twitch.tv/deatsaw
  15. Is there a technical reason for this being removed from 0.25? EDIT: helps to put in a link to the source
×