Vrack 2 Re-released

hmmm, i wonder whether this has anything to do with the recent idea AndrewB mentioned in #doom about 'Wad of the year' and recent criticisims about Vrack2's lack of a reject table ;-)

Nope nope.

this increased fps by pretty much for me, maybe as much as 10

No. He did it to SHUT ME UP!!!!!

Heh =).

Any other changes besides the two listed?

...by reading the textfile. dumb me.

I also added deathmatch starts and some corpses to bring the thing count up to 1337, and I updated the text file with new URL to my site etc.

But that's it. It's the same level. No misalignments fixed. No taking care of the detail-lacking ugliness that inhabits the level. No extra health added for those who are wimps incapable of playing the game.

I loved Vrack1, but for some reason when Vrack2 came out, there was something about its gameplay that left me cold.

Having played it again (1 hr 9 frikken minutes) I now realise the error of my ways and bow down to the majesty that is Vrack2.

Plays a damn site quicker for me now as well. But that could be something to do with a significant hardware upgrade since the first release. :-)

Just a question...

is there a port in existence that won't horribly and happily murder itself if you try to actually play this puppy in any sort of multiplayer mode?

"I also added deathmatch starts and some corpses to bring the thing count up to 1337"

rofl... my guess is that this release is fred trying to keep himself in the news ;)

You realise what this means, of course?

Our FRICKING DEMOS probably won't work any more. Curse you, Fred! CURSE YOU and your CURSED REJECT TABLES!

=)

And Enjay, yeah, you certainly will see a big FPS improvement, that's what a reject table is *for*. It allows Doom to stop worrying about monsters that are out of view, basically, and in a map as big as Vrack 2 this is a significant factor, because in the first version your computer was computing every single awake monster all the time, which puts a bit of strain on the poor thing.

arioch: Legacy :)

I played it coop in PrBoom, not a problem...

I never knew Vrack2 was finished already so im glad of the 're-release'. Thanks for the reject table, my slow PC needs all the help it can get.

Mystic: Vrack2 have been finished for atleast 6 months.

hey Fredrik, gotta update your links. Went to your site but the DL links still point to CDROM.com.

Sweet level!

You'll only see a "big" frame improvement IF you have a -slow- computer. Hashed this out with stphrz on newdoom:)

There is no detectable difference (with/wo a reject) on a PIII800 at 1024x768. Heck I couldn't even tell at 1600x1200.

Wanna argue? Go read the newdoom debate under Editors and see what stphrz and I argued about:)

This level was a nice example of a large level with lot's of "things" to be able to prove or disprove the hype:)

What port did you use?

Hmm, interesting read...

Gah! My mouse slipped.

I asked because when I ran the Vrack2(with reject) on the latest version of ZDoom I got zero framerate increase. Even on my P200. The reject table did nothing at all. With ZDoom v1.22, Boom and PrBoom it worked though.

Gah - lol. ZDOOM 1.23 b45c1.456.f (ok really 28a, I do love ZDOOM version numbers though - keeps it different and interesting:)

I looked at the source of 1.23 dated 8/2/2001 and in there the REJECT is still checked - and I think it still is. He did drop the blockmap stuff - since he posted sometime ago that he determined that it actually was slower:)

I may be remembering wrong? Anyway, that wouldn't surprise me, since code written for a 386/486 could easily be dumped and switched to different code better suited for the newer Pentiums. Remember how Quake was able to switch to FP? Actually faster than all that integer hacking they did.

I think I listed the module name in that thread. Other ports probably have similar code, since BOOM made the most changes in that area (unless they didn't use it?).

[you can play ZDOOM with a -0- length BLOCKMAP - the BLOCKMAP size issue is gone]

6 months? then I probably have it already. I cant be bothered searching my backups. I prefer to have the newer version anyway.

my machine is a p2 667, using mbf and running around vrack2 in -skill 5 with -devparm gave me like 3-6 dots, whereas vrack2b gave me 1-3 (mostly 1, or with the second blinking). and on -skill 4 the reject makes it run almost perfectly (as opposed to a lesser slowdown)

If you want to give "numbers", please also list your memory (the other kind), video card and resolution. Otherwise it's meaningless.

that would be

128 ram
2mb trident video card
320 x 200 res

many people don't have fast computers, and many play doom on slower machines even if they do

Hey Rellik, how did you get it? I really want this one.

Hey Fredrik, nice site, its a lot better than when I last visited. Update those download links, that is so annoying when you really want something and the link is dead.

Ok, thanks.

Get a new(er) video card and you won't believe the difference. There shouldn't be any difference at 320x200 with a PII667:)

Anyway an older TNT cost hardly anything, even a TNT2 is inexpensive and is lightyears faster matching your cpu better.

He did drop the blockmap stuff - since he posted sometime ago that he determined that it actually was slower

No, the blockmap is still there. It can vastly speed up collision checking because you only need to check the things in a small area instead of everything on the map. If the blockmap is 0-length, it just gets generated automatically when the level is loaded.

I assume you are talking about the bit in my log where I talk about using the blockmap for sight calculations. The Doom source released by id traversed the BSP to do LOS. Both Heretic and Hexen use the blockmap to do the same thing (so I assume Doom originally did, too). Way back when ZDoom 1.14 was new, I replaced the BSP-based LOS code in ZDoom with the blockmap-based code from Hexen. It wasn't until just a few months ago that I actually did some comparative timings of the two approaches. I found that using the blockmap to determine LOS in Vrack2 instead of using the BSP was generally between 4-6 times faster. At best, using the BSP was just as fast as using the blockmap, but it was never faster. I was quite surprised.

Ok, I did get confused:) I made a wrong conclusion when I saw a -0- blockmap worked with ZDOOM (and remembered you saying something about the blockmap on your forum). Didn't know you generated the blockmap.

Need to see what you did that since, as I recall, I tested a level with a blockmap > 128kb (if it had generated) and the format won't support that (offset issue). Before that it crashed at > 64kb and a -0- blockmap would make you run through walls.

But I differ with you on the REJECT in the real world. Although technically there's a speed difference in collision checking, that is noticable only for SLOW machines (the whole shebang - cpu/memory/video).

The overhead for the video code far outweighs the time for reject checking (which stays constant regardless of the res). IOW, the ratio of reject cpu time/video display cpu in a 486 vs a PIII800 is radically different.

Sure it helps, but it really does not make a big difference on a PIII800. You were probably on a PII300 (?) if you tested it. So it's all relative.

I can run Vrack at 1600x1200 with no problem without a reject voodoo3 or ge2. As a -human- I can't tell the difference with or without. There probably is a technical measurable difference.

Stphrz and I hashed this out. He ran the test on a 2000 or so linedef level of his with ~500 monsters I think. Go look at the newdoom node thread if you are interested.

Run it with nosound/music to focus on this issue. But only on a PIII500 or above is this true. PIII500 at 800x600, otherwise the res itself just slows one down.

I was talking about the BLOCKMAP, not the REJECT. Specifically, the use of the BLOCKMAP for LOS checking versus the use of the BSP for LOS checking. The gains from using a REJECT may be marginal on fast computers, but that doesn't mean the same thing applies to the BLOCKMAP. Even the Unreal engine uses a 3D version of a blockmap to speed up collision detection.