More Categories
8 years ago
Södermanland, Sweden

Good idea with the GOTY tag

Don_Elgorro likes this
Friesland, Netherlands

I fixed the runs not showing thing. Im assuuming Ork's DM run is on the GOTY edition then? I left it set to Yes for now.

California, USA

Hey there,

I just submitted runs for dm and assault. I may run a few other categories. I played ut99 online from 2007ish-2014ish and was a part of a few different online communities. I still hop on an instagib or bt server every now and then.

Everything NotBaDocha said is true. If people run it with the GOTY edition, they are doing different maps in certain categories, and more maps in certain categories. Times will be very different. There also seems to be discrepancy on start and stop times as well.

Load times are affected randomly as i've met people with better pc setups then mine that load the game slower than me and vice versa. In-game times and separate categories would probably be the best bet although the timer is limited to seconds, and it rounds it off if you land in between. Plus, nearly everyone who has ran the game so far has done it in real time. However, if more people were to run it, especially in the shorter categories like assault, you will come to find that the varying load times may have a large impact on the overall real time, which technically, wouldn't be fair. I guess this would be more of an issue if a few people really grinded this game out. To be specific, I'm not talking about the menu inbetween maps. I'm talking about how long it takes a map to end after you complete the final objective (like the final frag in dm) and how long it can take a map to load after you have selected it from the menu. Normally the latter is consistent and the former is inconsistent.

Unless there is some way to tweak the game to make the load times even. I only played the ladder once or twice before I decided to try to run the game, so I haven't really experimented with the ladder much.

Just my two cents. I'll continue to submit runs regardless of whats decided here.

Edited by the author 7 years ago
Friesland, Netherlands

Just to clarify, by "how long it takes a map to end after you complete the final objective" you mean, for example, the time from 4:56 - 4:59 video timestamps (or 3:05 - 3:08 on Livesplit) in this run: http://www.speedrun.com/run/8yvn2p4m ? Or just the tiny fraction where it says "Loading" in blue letters before just jumping back into the menu?

Edited by the author 7 years ago
California, USA

Yes, that is exactly what I'm talking about. It can hang for different pc's for different amounts of time at the end of each map as far as I know. There was a dm or assault vid from another runner I was looking at in particular, but I cannot remember which one. I do not have time tonight to do comparisons, and tomorrow, I am leaving to go out of the country for six days, so I can do the comparisons in about a week. Maybe it just seems that way to me (and the others I used to play with). To be fair, I have never created or seen hard evidence for it. It definitely feels that way though.

I'm sort of mixed on the whole thing. In-Game timer is extremely limited with rounding, and it would not catch someone making a mistake clicking through the menu. However, if there is hard evidence that loading times differ, than Real Time would not be fair imo.

It seems like you all have it figured out though in a way that works for the few people who run it. I do not mean to try to change your system.

Edited by the author 7 years ago
Portugal

Might have something to do with the video driver you're using, IF that does influence the times I'm positive using openGL over any version of Direct3D will make them finish faster.

It also depends on the maps you're playing, best example that comes to mind is the assault map HiSpeed, it takes forever to finish both on attack and defence, probably because you have to wait for "cutscene" at the end.

http://www.speedrun.com/run/emk8j05m (27:54 / 28:42)

California, USA

Yes, I am aware of the dependence on the maps. I don't think there is a way to not notice the long ending at the end of high speed lol. I, too, was wondering if the different renders mattered as well as any graphical settings. It could just be left to that. I have always played with the latest version of opengl available.

On a complete side note: I created a guide for turning on and off the goty maps in the ladder. The guide also includes an explanation of how you can activate certain ladders without playing through the game. It is a not too complicated process of editing the user ini. Thought I would notify this thread since some of the original posts were about turning on and off the GOTY maps. Let me know if there are any issues:

http://www.speedrun.com/ut99/guide/1v96k

Edited by the author 7 years ago
cyborg likes this
Friesland, Netherlands

Could look into making a livesplit component for loadless timers, but that'd solve nothing until we are sure if that waiting part is actually affected by it. Looking forward to those comparisons you'll make when you return.

California, USA

Just posting to let you two know I haven't forgotten about this. Had to work and such when I got back from vacation. In my small bits of free time I have grinded down the assault and dm runs a bit. I am also somewhat working on a novice full-game run. I am hoping to check out different renders and graphic settings to see if there can be some manipulation of the overall load time. Hoping to post here with my results, submit a few runs, and tweak my guide a bit by this Sunday.

Edited by the author 7 years ago
Naywyn likes this
California, USA

Sorry for the delay, I have completed enough testing and comparisons that I feel comfortable making a post.

First, I created a keybind so that whenever I hold down a specific key I would enter ghost mode (it is a cheat), and whenever I let go of the key it would deactivate ghost mode. I used this bind to let me cruise through high speed. I did this 5 times for a series of renders to see if specific renders could activate the load times at the end of maps. I chose high speed because it is the longest load time due to the helicopter animation, so, any possible variance in load times would be easier to see instead of another map with a shorter load time. I timed the load times using live split, and I then averaged out the 5 runs to get a decent representation of the load time for that specific render:

[center]software rendering: 13.89s, ~20 fps d3d: 13.92s, ~60 fps d3d8: 13.96s, ~120 fps d3d9: 13.93s, ~160 fps d3d10: 13.94s, ~160 fps OpenGL: 13.94s, ~160 fps[/center]

The difference was minimal (which kind of surprised me). 1/100's of seconds could possibly even be human error. I played around with various texture settings and game settings, but the average was still right around 13.9Xs. What got interesting was when I pulled up the three current assault videos that are listed on the site. I found some discrepancies between the load times of each player:

http://i.imgur.com/8Q3FhPv.png

So, Column A are the loads for the map. There is a load between Offense and Defense (Frigate O for example), and a load between Defense and the game menu (Frigate D for example). All of the times are in seconds, and the times represent the time to load between O-D and D-Menu. mmsrhino and I have very similar load times between Offense and Defense and between the defense and game menu. NotBaDocha has some slower speeds. The rows highlighted in blue are load times where he is over a second slower than either of us for that particular load time. I calculated the total load time for each of us in Row 15, and you can see that there is about a 6 second difference between ocha's overall load time and the two of us. There are other differences, but they are smaller for the most part.

Six seconds in an assault run can be a lot. It does not look like it right now as there are only three recorded times, but right now I have a pb of 6:58 Real Time (video is pending right now). My best ever possible with my current route would probably be 6:50-6:51. Six seconds could end up being a lot if other people ever decided to run the game.

The issue LOOKS to be that ocha's computer kind of struggles in his assault run. You can see that it is a bit choppier than mine and rhino's. I discussed this with a friend who played UT online even longer than me, and we THINK (can't really know for sure) that the struggling indicates that his FPS fluctuates. We think this fluctuation causes the load times to take even longer. If you ever want to check your fps in UT you can open up the console and type "timedemo 1" without the quotes. His FPS may be fluctuating due to many reasons: trying to run too high of a resolution. trying to run too high of graphics settings, possibly trying to run UT99 and a streaming program at the same time, running too many programs at once etc. There are a lot of reasons as to why it could be problematic, and I don't even know if it is possible to know for sure that the fluctuating fps is the problem.

The final thing regarding load times I'd like to mention is seen on Row 17 of the above graphic. The difference between Real Time and In-game time do not line up with the difference in load times for each player. Rhino's total load time is less than mine; however, my difference between real time and in-game time is lower than his. NotBaDocha has the highest total load time by a lot, but he has a better difference between Real TIme and In-game time than mmsrhino. I believe this is due to the combination of the Total Load Times (which are not counted in the In-game times) and each player's menuing (choosing the next map to play and also not counted in In-game time). If you watch the vid, my menuing seems to be the fastest as I hit esc and quickly click my mouse onto assault and then to the right arrow to select the next map asap. When you add each player's individual menuing times (which I did not calculate) with their Total Load Times, it comes out to the different order you see on Row 17.

Edited by the author 7 years ago
Naywyn and cyborg like this
California, USA

(Another post to break the wall of text). Another thing I found:

I could find absolutely no way to eliminate the lag in the utmenu after you finish a map on the ladder. You know how you can't instantly click the right arrow to move onto the next map? That lag. Yes, you can hit the back arrow and then go forward, or you can hit esc and then reenter the ladder and proceed to the next map. I mean that there is no setting in the Unreal Tournament or User inis that allow you to just simply click the right arrow without doing any of the above and without a 3 or so second lag. It is either hit the back arrow or hit the esc key for now. This is the same for when you unlock a new ladder (if you are doing a full-game run for example). Usually there is a short animation that shows the unlocking of the next ladder. You can only skip that if you hit esc right when you see the animation start. There is no ini setting that instantly skips this animation.

Sorry for the wall(s) of text. If you ask, I try to deliver. I know no one runs it, and there is a chance nobody will run it again, but this info is here if anyone needs it now. My original opinion about going by In-game time isn't correct I think. It SEEMS that if you can have the settings set so your pc can run ut smoothly, then the load times will be pretty similar. Idk for sure though.

Edited by the author 7 years ago
Naywyn and cyborg like this
Portugal

Great insight boss, thanks for going through the effort to check the render times, I'm surprised that the load times aren't different between d3d9/10 and OGL, good to know!

ocha's problem may just be down to low refresh rate. When I used to record clips for frag videos @30fps long ago (I'd lower UT's framerate as well to avoid blurring and frameloss) I noticed load times in general were much slower even tho my fps didn't fluctuate, although this is pretty farfetched with nowadays' computers.

If there's interest by the other runners I can put a copy of my ini files, or even my UT (minus any extra maps/demos/textures etc) up on the resources, my load times all seem consistent with my settings.

PS good job on your latest assault run, sub 7 is sick! :)

bigboss and Naywyn like this
Södermanland, Sweden

Good finding Boss. I though OpenGL would be the fastest but apparently not, and also how fast is software mode at for example 120fps instead of 20 does it make a difference?

Edited by the author 7 years ago
California, USA

Thanks Cyborg!

Yeah there are a ton of settings for each render that could be making his game stutter. I am pretty sure that it is safe to assume that discrepancies between different load times are primarily due to having an improperly calibrated UT install. And, I thought/expected the exact same thing Jonte. I am going to try to up the fps on software rendering and d3d; however, both renders have issues concerning speed and consistency. I want to do that and play with a few more ini settings.

Having examples of proper ini's in the resources is probably a good idea actually!

Also, kind of going back to one of the original questions in this thread regarding versions and such, mmsrhino's and Dein's Novice full-game runs are the GOTY version. If you go into like 10 minutes or so into each run you can see they have the arcane temple map, and that is a GOTY edition exclusive. Them having GOTY switched on also helps to explain the huge time difference as it has a few more maps overall.

I am considering trying out a 4x4x4x4x4 or a 3x3x3x3x3 run. Basically do 3 or 4 maps in each category (4 would= all of the challenge maps). I am pretty sure you could just tweak the user ini save game files to have all of the ladders unlocked (challenge ladder won't unlock unless you beat all of the maps in all of the other ladders normally). Any thoughts on that? It would be like a shorter version of the full-game run. Just a taste of each category.

Edit: 3x3x3x3x4 may be the best of both worlds. Do the three maps in each ladder which would normally unlock the next category, and then do all 4 challenge matches to win the game.

Edited by the author 7 years ago
Naywyn likes this
Friesland, Netherlands

Wouldn't it be better to make full ILs for each map for every game mode (including Last Man Standing), in practice session with default settings (30 kills / 3 flags / 100 dominations), rather than come up with an arbitrary category that pretends it's a full run but really isn't

Portugal

The issue is practice section bots behave differently than campaign/ladder bots, they're less objective orientated and more prone to camping. They can also be tweaked to have worse aim, ignore bot-pathing, etc.

California, USA

Yeah, I was going to say the exact same thing. The practice session has a tab all the way to the right which makes it very easy to manipulate the bots you play against. IL's are not a bad idea though. Could create them for the ladder maps.