Sub-1Ghz Testers Needed

Community Forums/Showcase/Sub-1Ghz Testers Needed

Gabriel(Posted 2004) [#1]
NOTE : NEW VERSION!! This has profiling code compiled in it. What I need you to do is play the game a bit, bowl some shots, push it hard, make it slow down. Then zip up the "profile" folder that the game creates and email it to me :

sybixsus@...


This should help me figure out exactly where the slowdown is. Many thanks.


If you already have the Beta, you just need the betaprof patched exe.

http://www.blackwolfgames.com/bowling/betaone.exe - MAIN D/L ( UNCHANGED )

http://members.aol.com/sybixsus/betaprof.zip - PATCHED EXE - PROFILING INCLUDED

Much appreciated.


Rob(Posted 2004) [#2]
I suspect you're not using tokamak properly AT ALL. I have tokamak doing 100's times more than simple bowling. You need to ensure tokamak's timing is set up very well.

I have tested 10 (!) cars at full frame rate with steering on my 800mhz. A ball and a few skittles doesn't come close.

Perhaps there's something really odd going on with tokamak timing - now THAT can cause problems...


Rob(Posted 2004) [#3]
Right I've tested it... when I hit them at full force dead in the middle there is a drop in framerate. It'll drop to an estimated 10fps for the duration.

It shouldn't though, because there's another bowling game done in blitz tokamak with absolutely no slowdown. Why not visit the playerfactory forum and resolve the issue?


Gabriel(Posted 2004) [#4]
Okie dokie, thanks for the test and the suggestion. I was actually thinking about some of the things I've seen in Tokamak and wondering what I was doing that made it so slow. I'm only retrieving pin and ball positions and rotations, so I guess you're right. maybe it is to do with setting it up right in the first place. I'll post over on the Tokamak forums as you suggest.


Gabriel(Posted 2004) [#5]
Of course, it's worth pointing out that the other bowling game made with Tokamak may not be using a terrain mesh at all, whereas I'm using a fairly complex one, and many of the other demos may be using either simple or no terrain meshes.

I don't know how much of a difference that might make, but it's a consideration.


Gabriel(Posted 2004) [#6]
Ok, with a little help from Sweenie, some things are becoming clearer and yet the overall situation is more confusing. Can anyone else with a sub 1Ghz CPU test this to see if the results are consistent or if it's patchy?

Rob : You definitely turned shaders off? The demo has them on by default and I'm wondering if it's actually something funky that the cube-mapped pins do when they're being tossed around that's causing it.

You see, I've modified the source to display the number of milliseconds the physics is taking and it's taking around 0.05 millisecs most of the time, peaking at around 0.1. Even on a PC which is ten (!!) times slower than mine, one millisecond for the physics is pretty darn low. And most important of all, the physics time doesn't spike at ALL when the ball hits the pins.

[edit]Ah Hell.. I'll sort this out one way or another. I've uploaded a patch that displays the milliseconds the physics are taking in the top left corner of the screen. People who don't have the beta yet need the beta AND this patched exe. Rob and anyone who tried the previous beta version, if you wouldn't mind testing it again, you'll only need this patched exe. You can leave the old one around or overwrite with this one. Makes no odds.

http://members.aol.com/sybixsus/beta101.zip[/edit]

I can't think of anything else that changes when the ball hits the pins. I mean I get the pin positions and rotations every frame. I could ( and probably will ) optimize it so that I don't get the positions and rotations of the pins until the ball is nearly to the end of the alley, but at the moment, I'm doing it all the time.


Red Ocktober(Posted 2004) [#7]
Tested it on a 733mHz machine with GeForce2...

it slowed down as the ball approached the pins at 1024x768

it slowed down considerably when pins were hit at 1024x768

a little slowdown when pins were hit at 640x480

in practice 2 mode the slowdown seemd more.


nice babe animation while bowling...
needs background noise, people bowling...
needs change of music, current one sounds like alternate sound track from MYST...

(don't ya just luv it when people come around at the 25 hour and decide that they know how to make your game so much better :) )

overall... nice looking app that may find a following.

Good Luck



--Mike


Gabriel(Posted 2004) [#8]
Thanks Mike.

(don't ya just luv it when people come around at the 25 hour and decide that they know how to make your game so much better :) )



Indeed I do, which is why I posted this so early. This version is around 2 weeks of work, so that I can get these issues sorted before I get too far into development.

BTW MIke, did you download the patched EXE? Could you let me have some numbers ( top left corner ) to see how many milliseconds the physics engine is taking when the pins get busy and the game slows down?


I'm thinking it could be something entirely different than physics now. Maybe the camera gets too close to the pins and fillrate becomes an issue? Or cube mapping? Did you have shaders turned on or not?


Red Ocktober(Posted 2004) [#9]
no shaders...

the numbers are

in 640x480 59420416
in 1024x768 44195328

--Mike


Gabriel(Posted 2004) [#10]
Erm, no that's the amount of videomemory not being used during the menus. I'm talking about the numbers during the actual bowling. As the ball rolls down the alley ( if you have the 101 patched exe ) there's a number in the top left corner. It's only onscreen during the chase camera bit, and I need a rough idea of the average figure and the maximum. Also whether or not the number gets much higher when the pins are hit.


Red Ocktober(Posted 2004) [#11]
k coming... sorry...

--Mike


Red Ocktober(Posted 2004) [#12]
the numbers range from 0.0119636 to a high of 0.0127921 in 640x480...

they seem to spike higher in 1024x768 around 0.024xxxx to 0.0258xxxx

i hope that helps a bit.

-- Mike


Gabriel(Posted 2004) [#13]
Many thanks, Mike. Those are the same kind of numbers I'm getting on my girlfriend's Intel P4 HT 3Ghz+ monster. Essentially the physics are taking no more than 1/40th of a millisecond even on the highest resolution. So Tokamak is insanely fast, and physics do not appear to be the problem. Very, very odd.

But it's great news too, because it means I don't have to scrap the game. I just have to look for a problem somewhere else.


Agamer(Posted 2004) [#14]
Good to here but bad to see you have still got a problem!


LineOf7s(Posted 2004) [#15]
No, it's not your physics - even on 1024x768 the 'physics numbers' you wanted us to quote to you didn't get above 0.029 at its worst. Everything else in the game is smooth and pretty; as reported, it's only just before the ball hits the pins and as the pins recover from that hit that the frame rate falls through the floor.

On a somewhat unrelated note, can I suggest you pare down your control system a lil? For someone who wants to just get from one frame to the next with a minimum of fuss, there's just too many mouse clicks required.

Might I suggest the following for your consideration:

- Remove the need to 'aim and shoot'. Have the pins displayed all the time in (perhaps) a lil window up in the corner - nobody wants to have to click to see what's remaining, it should just be there. If you remove the opportunity/need to click on the pins to see them, then that means you can just go straight into the 'aim and shoot' phase... which of course is just what people wanna do.

- Once you get past that bit, you have click to: aim the line down the alley; choose the angle down the alley; choose the strength of the shot; and then choose the spin of the ball. I appreciate that allows for a lot of control, but (again) if people are there to bowl, they just wanna get bowling - after all, that's the fun part. I can't help feeling there's a better way overall to get most of that functionality into your controls, but at the very least I would suggest you get rid of the 'choose angle' segment - that aspect of the bowl is covered by the spin you add (as far as an arcade bowling game is concerned, anyway). Again, I'm sure there's a better way to implement your control method and as far as your core gameplay goes (ie not counting tournaments or multiplayer or whatever - just the bowling itself) it's the weak link in the experience (imho, of course).

(Of course, this relates mainly to the control method in Practice, as opposed to Practice Two, but Practice Two is only marginally better as far as accessibility)

Apart from that, I find the game excellently presented and produced. Really top notch. See if you can make it a bit more heedlessly fun, and you'll be on a winner.

PS: Having played it a bit more now (surely that's got to be some sort of encouragement?) it's probably not a good thing that I ended up with one hand on the mouse and the other resting on the ESC key so I could skip through the cut-scenes and *get on with it*. An option to display them once at the start and then have them turned off, perhaps? I choose to speak for all humankind when I say that the only part of a bowling game that anyone truly cares about is the bit where the ball rolls down the alley and hits the pins. Everything else just needs to be tolerated until that alley bit can happen again.

PPS: And given how much we need to see them (usually sitting there impudently observing the pitiful efforts of the bowling ball to knock them down), and the high quality of everything else, you really should put a lot more work into making the (close up view of the) pins look exceptionally excellent (or, at least, a lil better than they are. Clearly you - or your graphic artist - are capable).


Gabriel(Posted 2004) [#16]
Thanks for the feedback, Line.

It's certainly good to hear that the physics are not presenting a problem for any rigs, but it's very confusing trying to work out what else is happening when the ball and pins collide that isn't happening at any other time.

A few people have mentioned that control method #1 involves too many clicks. It's probably not the control method I would choose, and I can see why people don't like it. The reason it's there and the reason it's so staccato is because I want it to be accessible to the typical indie gamers who don't play a lot of games and certainly don't play games requiring hand-eye coordination. Control method #2 is the control method for experienced gamers who do have good hand-eye coordination, though I quite accept that method 2 is not right yet.

When you say get rid of the "choose angle segment", do you mean I should just assume it's in a straight line down the alley from where they line up? That's certainly doable, but it does mean that certain shots are impossible and you won't always be able to make up for it with spin. ( Usually, but not always. )

I'll have another look at the controls. Maybe I need three control options. The current Staccato controls, a simpler arcade controls set and the TruBowl controls ( Practice 2 ) which gives you direct control over the shot.

Is three different control sets too much?


Gabriel(Posted 2004) [#17]
New Patched EXE available.

http://members.aol.com/sybixsus/betaprof.zip

This includes full profiling code, thanks to Michael Reitzenstein's preprocessor. Just download the patch and use that exe instead of the original one. Play the game, bowl some shots, push it hard, make it slow down and then zip up the entire "Profile" folder that the game creates. Email it to me here :

sybixsus@...


Many thanks guys.


Red Ocktober(Posted 2004) [#18]
i don't think it's so much the controls being obtuse... as the fact that most people may not realize that they can speed things up a bit by hitting the esc key at ceertain times to bypass some of the animation and camera movement, and place you instantly at the next 'point'...

at first i thought it was a lil cumbersome, but as you play the game, and get used to the esc key shortcut, it seems a lot smoother...

--Mike


Gabriel(Posted 2004) [#19]
Good point, Mike. That should be clearer. I could add some kind of visual indication of that. BTW the right mouse button works instead of escape. Sorry, should have mentioned it earlier I suppose :(


LineOf7s(Posted 2004) [#20]
Personally I thought the control issues and the 'speeding things up by skipping through the pretty stuff by hitting escape key' issues were two separate things - but knowing now that you can use the right mouse button instead of escape has dealt nicely with what I raised before about sitting with one hand on the mouse and the other on ESCape.

As far as the controls go, you're right in that the omission of the 'shot angle' control will make certain relatively 'normal' shots very very difficult indeed, if not impossible. Perhaps the issue I have is less with the amount of clicking that needs doing as much as the amount of clicking that needs doing *before anything happens*. I'm sure I'm sailing perilously close to the "needs a major rewrite" wind, but perhaps if she could start bowling once you set up the 'aiming', and you set the strength and spin whilst she's moving, perhaps that would make things 'flow' a bit better. I'm loathe to compare games to others (the Virtual Tennis/ATB debate is still ringing in my ears), but I keep thinking how other bowling games have managed it, and I keep coming back to 10th Frame on the C64, which had a golf-game-like 3-click system. Now, I'm in no way suggesting that specifically, but in that, you set where your bowler stood and the angle he bowled at, and then strength of shot and the amount of spin were set using the 3-click system whilst the bowler animated his way up to the line.


Gabriel(Posted 2004) [#21]
Well I admit i prefer to plough my own furrow rather than copy something else, but there is a fairly well-known bowling game which uses a very similar control method to my method #1. It's true that I've played a couple of games with the "golf controls" but I really didn't like them much.

What I currently plan to do with control method #1 is combine lining up with setting power ( 2 clicks become 1 ) and combine aiming with setting spin ( 2 clicks become 1 ) which will effectively halve the number of clicks required from four to two.

With method #2, I could make it so that she starts bowling as soon as you move the mouse to set the power and aim, then let you set the spin while she begins bowling. Course then I'm back to her ass getting in the way of the shot, but maybe I can cut to another camera without messing up the spin setting.


LineOf7s(Posted 2004) [#22]
PS: I can't for the life of me get your "updated patch including profiling stuff" link to work, or else I'd send you what data I could...

PPS: On paper at least, "4 clicks becomes 2" sounds very good indeed. I really haven't yet come to terms yet with your TruBowl system as it stands (I realise you're still tweaking), but I feel I'm more of a clicky person usually anyway.


Gabriel(Posted 2004) [#23]
I may have accidentally broken the link to the profiling exe when I was playing with some possible fixes earlier on. The profiles aren't really finding a slowdown anywhere. Dicussions with Specis and Mordy led me to a possible solution, but I've decided to just get on and finish the game now. I'll worry about making it work on low spec PC's more when it's finished. It'll be easier to motivate myself if I have a publisher interested or the prospect of being able to sell it myself when I solve the problem.

If you or anyone else want to have a play with my "fixes" to see what ( if anything ) they achieve, they're here :

http://members.aol.com/sybixsus/3fixes.exe < Self extracting rar with three new exes. One disables Tokamak ( might have to restart the game after each shot as the ball will probably fly off in all directions on this one. ) One which disables updateworld and one which doesn't move the camera at all. The ones which disable updateworld and camera movement should run just as before, no need to restart all the time.


http://members.aol.com/delay.zip < a single zipped exe. This one simply adds some delay commands in the main loop to give Blitz more time to "breathe".


There's nothing to moniror or write down or email me in any of these. If you get any improvement from any of the four exes, just let me know which one and how much improvement. They actually may all offer a slight improvement.

Thanks.


Michael Reitzenstein(Posted 2004) [#24]
Sybixsus - Send the profiles to me and I will take a look at them for you. I might spot something you missed. huntersd@...