next »

[jack_dracon]

I don't know if that is a good resource for decisions about tools, but this can help in some ways for a research on the  tools list.
(PS.: I don't know if the result in some aspect is real, but is a pretty high the results for Construct 2).
http://www.scirra.com/blog/59/construct-2-vs-gamemaker-vs-gamesalad-vs-stencyl
Cheers. ;}D

[Theophilus]

It's high results because the makers of Construct published that. I didn't click but I assume there is some amount of bias, especially if you say Construct is high.

[TeeGee]

Hi, I am not a bot?! =D

Wait. You aren't sure if you are a bot or not? Man, that's so Philip K. Dick.

[s0]

yeah lol the official construct site is probably not the credible source for these things

[mcc]

I think that Construct 2 making a case for why they consider their product better than comparable products is useful information even if we must consider that information is not from a purely objective source

Personally though I'm not sure speed is the most important criteria I'd judge one of those programs by. If I were doing something speed-critical, it is unlikely I would create it with a gamemaker-like tool.

[DavidCaruso]

guys they posted their entire design methodology with links to the tests they used to measure each application, why not try to see if there are flaws in that instead of dismissing it based on "bias" or "credibility" or w/e

I'm not too versed in performance testing stuff myself but is creating as many sprites onscreen as possible before it starts lagging really the best way to go about this? It's probably a good test for just pure rendering but in the end for most applications each sprite will have to be executing some game logic, the majority of them won't be sitting there just displaying.

[ScirraTom]

guys they posted their entire design methodology with links to the tests they used to measure each application, why not try to see if there are flaws in that instead of dismissing it based on "bias" or "credibility" or w/e

I'm not too versed in performance testing stuff myself but is creating as many sprites onscreen as possible before it starts lagging really the best way to go about this? It's probably a good test for just pure rendering but in the end for most applications each sprite will have to be executing some game logic, the majority of them won't be sitting there just displaying.

Tom from Scirra here, this is all we're trying to say!  There are currently no benchmarks out there comparing products so we thought we would do one, and we invite other benchmarks from independent sources as well!  We'd love to see other benchmarks in this area.

Our full methodology is explained along with project files to download as mentioned above.  Of course we knew that when we did it we would receive criticism for bias.  But as there is no independent source to refer to we had to do it ourselves.

We've put a lot of effort into optimising Construct 2 for performance and it's perhaps one of its less known features which is why we did the blog post.  We think performance in games can be important (but not all the time!) as it can be a limitation on creativity.  Ashley sums it up best in the other performance blog post:

You might be wondering: why is performance so important? There are a number of reasons. Firstly, games running at a solid 60 fps look better and make for a more enjoyable playing experience. Secondly, slow engines put a limit on creativity. Do you want a moment in your game when 100 enemies rush you and the screen is full of explosions? How about a crazy bullet hell game? If your game engine is slow, the game may be unplayably slow, and then you have to revise your game to something less exciting - how disappointing! Finally, faster engines are more efficient, which means they use less battery on laptops, phones and tablets. So you really want your engine to be as fast as possible!

http://www.scirra.com/blog/58/html5-2d-gaming-performance-analysis

[Superb Joe]

i've sniffed a few boxes in my time. game boxes, i mean. and i definitely would not want to eat a game salad.

[moi]

The best way to optimize HTLM5 for performance is to use something else than HTML5

[iceflame]

It seems bias in my opinion, in fact most of the blog so far just seems like a big marketing plan.

what seems funny is how someone posted a non scirra made stencyl version that performed better shortly after, plus it is not using the new flash stage thing which would probably be much better than WebGL anyway. While it's true HTML5 is getting better you still need to use the GPU to take advantage of that which not all players would have but flash is fast enough without it.

Construct 1 was quite good but always full of bugs and having looked at the forums recently for V2 they clearly no longer have any interest in Exe format unless it's in wrapper (which would be pointless). They clearly seem anti-flash also so i guess flash probably won't be a possible format either. Construct2 seems good if you want HTML5 though but it's still no where near as good as flash really.

[Mikademus]

"It seems biased", damn it! Compare "Your reasoning is biased" and "there is bias in your reasoning". This grammatical error drives me bonkers.
</grammar_nazi>

[ScirraTom]

It seems bias in my opinion, in fact most of the blog so far just seems like a big marketing plan.

what seems funny is how someone posted a non scirra made stencyl version that performed better shortly after, plus it is not using the new flash stage thing which would probably be much better than WebGL anyway. While it's true HTML5 is getting better you still need to use the GPU to take advantage of that which not all players would have but flash is fast enough without it.

Construct 1 was quite good but always full of bugs and having looked at the forums recently for V2 they clearly no longer have any interest in Exe format unless it's in wrapper (which would be pointless). They clearly seem anti-flash also so i guess flash probably won't be a possible format either. Construct2 seems good if you want HTML5 though but it's still no where near as good as flash really.

Admittedly anything positive we post about Construct 2 is generally intended to help promote Construct 2.  If we have an aspect of Construct 2 that is quantitatively better than competitors we do want to tell people about it!

Also if you read all the comments you will see that the alternatives posted in the comments section showing faster benchmarks were specially crafted to perform well in the benchmark which misses the point of the benchmark completely as it's not measuring overhead of the game engine itself.

We really would love to see other peoples tests/methodologies/benchmarks comparing products.  We're confident Construct 2 will come out on top each time if done fairly.

[jotapeh]

Well, here's a couple things that bug me about this:

1) The tests simply don't show the performance differences their tests seem to on my machine (Mac 2.8GHz Quad-Core Intel Xeon, 4GB RAM, latest Chrome 15.0.874.121).

Stencyl outperforms the Canvas2D unaccelerated test, and the WebGL accelerated test doesn't seem to render properly...


2) There's no source code for the Stencyl version. "I couldn't figure out how to Save As.." Doesn't cut it. How do you compare Construct 2 (something you are amazingly well versed in) with your relatively amateur "I can't figure out how to save as" skill level in Stencyl? Not a fair comparison, at least show us your code/project, let us see if you butchered it intentionally. Honesty in reporting or please don't post it at all!


Actually a quick Google search turns up this thread: http://community.stencyl.com/index.php?topic=4972.0

Which shows some of the 'other side of the argument', good read for anyone who is seriously considering Construct 2 or Stencyl..

This feels like preying upon people who don't know very much about programming (their target market, in fact.) It misleads them with performance tests that don't show source code and even if they did, their target market would have little success understanding it.

All in all, feels bad man. This doesn't convince me to try Construct 2 or recommend it to friends, but I suppose I'm not the target market anyway.

[ScirraTom]

Well, here's a couple things that bug me about this:

1) The tests simply don't show the performance differences their tests seem to on my machine (Mac 2.8GHz Quad-Core Intel Xeon, 4GB RAM, latest Chrome 15.0.874.121).

Stencyl outperforms the Canvas2D unaccelerated test, and the WebGL accelerated test doesn't seem to render properly...

If WebGL doesn't render properly, it might be a driver issue which wouldn't be our fault but the graphics vendors fault (are you using the latest drivers?)

2) There's no source code for the Stencyl version. "I couldn't figure out how to Save As.." Doesn't cut it. How do you compare Construct 2 (something you are amazingly well versed in) with your relatively amateur "I can't figure out how to save as" skill level in Stencyl? Not a fair comparison, at least show us your code/project, let us see if you butchered it intentionally. Honesty in reporting or please don't post it at all!

The Stencyl test is available on StencylForge under the name 'Performance Test'.

Actually a quick Google search turns up this thread: http://community.stencyl.com/index.php?topic=4972.0

Unless Stencyl can point out what was wrong with the test we made then making different tests doesn't seem fair.

This feels like preying upon people who don't know very much about programming (their target market, in fact.) It misleads them with performance tests that don't show source code and even if they did, their target market would have little success understanding it.

Not at all, if someone can point out what is wrong with our methodology we would like to hear.

[Desert Dog]

GM user here.

@scirratom: Tip; Export projects as a compressed .gmz. I can't actually load the project file for some reason.

I could see from the files that you were creating a new instance of an object for every 'sprite' in your test. Objects are.. well, objects are 'heavier'. you don't just use them to draw a sprite.. there are particles, tiles, or one object that draws multiple sprites.

May I ask how your Construct2 test worked? E.g. did you do you have an 'object' equivalent, and where you creating a new one for EVERY new sprite?

Good test, though, posting it over at the GMC. One of the Yoyogames programmers promised to do a benchmark test a while ago, and I still haven't seen it. 

[ScirraTom]

Hi DesertDog,

Thanks for the feedback!  Will pass the gmz tip on to Ash to make sure he knows and make sure the GM project is accessible.  I'll also wait until tomorrow for a proper answer about how Construct 2 works as Ashley is the expert there.

As you mention this benchmark really is only one small test in an ocean of possible tests!  However with good testing you always aim to isolate things.  Here we've isolated something that's easily replicable and shows what we believe is a good indicator for performance.  A complex, difficult to reproduce test would get a lot more criticism than we've already received.

It's also going to get the ball rolling for more benchmarks hopefully.  Quantifying performance like this is a good way to keep everyone competitive, and at the end of the day benefit the users in the Indie community.  We're very confident Construct 2 will come out well in whatever benchmark you throw at it though.  Would obviously be very interested in seeing a YoYoGames benchmark let me know if one ever become available!

Tom

[rivon]

Well, the test isn't really telling much. Games isn't just sprites. It's also sounds, music playing in the background, physics, pathfinding and AI, input handling etc. All those need to be fast, not only the graphics.

[AshleysBrain]

Hey, Ashley from Scirra here, Tom asked me to answer this point:

I could see from the files that you were creating a new instance of an object for every 'sprite' in your test. Objects are.. well, objects are 'heavier'. you don't just use them to draw a sprite.. there are particles, tiles, or one object that draws multiple sprites.

May I ask how your Construct2 test worked? E.g. did you do you have an 'object' equivalent, and where you creating a new one for EVERY new sprite?

Construct 2 doesn't separate objects and sprites like GM does.  A Sprite in Construct 2 is an object which contains its own sprite.  We wanted to make the test fair so we took the same approach with all the tools.  This was also intentional since the point was to measure the overhead of the game engine in the renderer - they aren't meant to be images which a bit of code just paints to the screen, but full blown objects in the engine.  The test isn't simulating some particle effect, it's trying to simulate a really big game.  GM came relatively close to Construct 2 so I thought it was a fair comparison since I thought the engines had the same job to do.

Does that help clear up what we were trying to do?  As Tom said we're happy to see other benchmarks, we did this test out of a genuine belief Construct 2 has the fastest renderer out there.  And of course there are many other aspects of an engine to measure, so we're happy to see results for other things too.

[Desert Dog]

Hi Ash,

Yes, that clarified a lot! A sprite for me is an image. :D

It's a bit of an apples&oranges case, tbh, however, your point that C2's 'base object' is light-weight compared to other programs seems to be fair enough, so congrats on that. May arguably make it more useful when devving for phones, and other weak devices. However, a base object, and a game are two different things, and I've had fine results with GM so far.

I've posted your blog post over at the GMC:
http://gmc.yoyogames.com/index.php?showtopic=525101&st=0&gopid=3866532&#entry3866532

Where it would seem most of us misinterpret what 'sprite' meant. However, I must warn you that the GMC has a very strict 'no discuss similar tools' rule, and if you do wish to join in&clarify something, etc, but then somewhere along the lines a mod closes the topic, well, just don't be too surprised. It'd just be one of many topics about other products which have been closed in the past, nothing personal at all.

(I've linked to this topic enough times, so anyone seriously interested can come read here anyway)

[jack_dracon]

Hi, I am not a bot?! =D

Wait. You aren't sure if you are a bot or not? Man, that's so Philip K. Dick.

Hahaha. BladeRunner is awesome and a classic too!