H120 plugin speed -- games play *fast*

[Chronon]

I've noticed that in comparison to my Gigabeat F40 the H120 that I just Rockboxed plays a lot of the games at a very high speed.  Brickmania, for example, is very difficult to play because the gameplay is so fast.  Is there a particular reason for this?  I searched the wiki, but couldn't find any info on this -- and obviously the manual does not contain inter-target comparisons.  It's not a huge issue, since I would opt to use the Gigabeat for games anyway.  It's just a curiosity at the moment.

[LambdaCalculus]

Wow, that's kinda strange. One of my friends has an H120, and he says that the games all play just fine. Has it always been like this for you, or has it started with a recent SVN build, perhaps? I'm rather curious about this as well.

[Llorean]

Well, if you're using Brickmania for an example, it's a unique case. The gameplay speed is dependent on the target's resolution (due to some quirks in how it's programmed, if I understand correctly) so because of this, lower resolution screens tend to be faster I guess.

[Chronon]

Hmm... okay.  Is Bubbles the same?  Because I've noticed that the bubbles fire faster and the "Hurry Up!" prompt comes much faster on it than on my Gigabeat.  

That makes a lot of sense, though.  I haven't noticed this on all plugins/games.

[LambdaCalculus]

Actually Llorean, now that you mention it, I have noticed a difference in speed between my iPod video and my girlfriend's iriver H320 when playing certain games. Brickmania is a bit slow on my player (which is 320x240), compared to the iriver (which is 220x176). My H120-toting friend hasn't informed me about the performance of all of the games, but a couple like Bubbles and Star do feel a wee bit faster on the H120 than on the H320 or my iPod.

[Llorean]

For brickmania at least, it's resolution dependent because the game itself isn't moving faster, it's just the ball and paddle are moving different amounts relative to the screen size. So, if the "Faster" prompt in Bubbles is showing up at different times on different targets, it may be that there's a bug (or at least, something that someone should look at for lining up cross-target). I don't know if "Faster" is based on a set time past, or if it's based on something else.


Edit: It's possible something like this affects many plugins, making them "feel" slower. For example, any plugin that moves something 1 pixel every 1/10 second, or some such would of course move things slower or faster depending on screen size. I imagine in a lot of cases adapting a plugin to a specific screen size has simply been a case of creating new graphics for it, and setting new bounds.

[Chronon]

Well, what I noticed on playing Bubbles on a Nano versus a 5g iPod is that the bubbles would all fly around the screen much faster -- apparently the same effect as with brickmania.  This difference in time of flight could explain my perception of faster hurry-up prompts.  The other games that I notice this behavior with would also seem to behave in the same way as Brickmania as far as this goes: Xobox and Chopper.

My curiosity is satisfied.  

[LambdaCalculus]

I just tested Bubbles on my iPod and on my Archos; both display the "HURRY!" message after 8 seconds, but the Archos does fire the bubbles much faster than on the iPod.

Chopper does seem to suffer from this as well. I tried it on my friend's H120 and found it to be *really* speedy. On mine, it's easy to reach 5000+ feet because of the slower speed (which does seem to be resolution dependent, but I'm only speculating).

I have noticed that even Doom appears to be running a little faster on the H320 than on my iPod. Anyone test Doom out to see if this is true?

[Chronon]

I guess the faster motion of the bubbles tricked me!

[Llorean]

In the case of Doom, it may just be performance issues. Bigger screens are also harder to render.

For *most* of the plugins though, due to their relative simplicity, it's more likely how the code works than performance concerns. But tis is "more likely" not "certainly"

[LambdaCalculus]

As always, thanks for the info, Llorean!  

[pixelma]

Bubbles is somehow dependent on cpu frequency when it can't keep up with drawing the animation slows down. On some targets you can really see a difference whether the cpu is boosted or not (not on Archos of course). There is even a patch in the tracker to circumvent the problem by letting the plugin always boost the cpu - a wrong approach in my opinion. If I remember correctly there is also "max fps" define (which could explain why it seems slower on the gigabeat).

As for chopper, there are cycletime defines in the code that set the speed, I guess they were chosen on each target like they felt right, I'd be surprised if someone compared them before.

That's the bit I know about some plugins, don't know about the others. There seem to be a lot of reasons that plugins run differently on different players - some even seem solveable by coding...

Edit: another thought

[Chronon]

Thanks for the info, pixelma.  Although that would seem to suggest that the motion would be the fastest on the Gigabeat rather than the Nano (as I seem to recall).

Yes, I can imagine that we could normalize the performance of these plugins across targets a bit if we put some effort into it.  Although, hardly a priority at the moment, I think.

[pixelma]

I also think that this isn't a high priority at all. *taking the chance to point to the additional idea I had about bubbles on the gigabeat because it seems it was a bit hidden*

[Chronon]

Thanks.  I missed the "max FPS" constant that you mentioned the first time through.