Hello ! Is this okay to set frameratelimit to lets say 10 for example?

The thing is i dont want to use time in my game and i need to slow things up on the screen. Setting framerate limit to lower value solve the problem for me but is this safe to do that?

Hello

How would I copy last N Frames in Unity3d ?

Is it a must to use a stack of texture2D ? check if its not empty, pop it ?

or there is a clever way of doing so ?

@0x00004 I mean it's "Safe" .. it would also probably get your game labelled as unplayable

@AhmedSaleh don't know of a better, more proper way to do it. But I would reuse the textures if possible, Allocating such a large object every frame is not the best idea if you can avoid it

    if(s.Count == 4)
    {
        int count = 0;

       filterMaterialAssetLeft.SetTexture(_historyIDs_flow[count++], s.Pop());
    }

that's what I did now

where s is a stack

@Tyyppi_77 I want to save Last N Frames ?

Is that a stack or a fifo

yes

a queue ?

or a circular queue ?

@Tyyppi_77

sorry

Hello!

The government has told us the parameters on how the holiday time should be spent. And of course, it can't please everyone...

also, I've started to build llvm/clang to see what it can do for us..

Building clang from the source, then use it as a static analyzer for our software. (And probably check what clang-format can do.)

You mean as opposed to installing the visual studio plugin?

I don't intend to add any custom features, no.

I have not been able to find "binaries" aside from the VS/VSCode integration.

The plan is to integrate the tools with scripts, and I think, at least here, having the devs download and extract a zip is simpler than having them install and use the plugin.

Oh, look!

Nice! I'll try that.

Could you do a "drop in replacement"?

replacing the old binaries by new ones?

Ah, right, nice!

When configuring linters in sublimetext, you also need to specify the binary path

Now I need to figure out how to use that in a script...

yep

I've managed to do that with cppclean and cppcheck.

Likely, yes, at least for now. I understand there is a python script already made that will do a diff and format only the updated code; maybe down the road that'll be what we'll use; I don't know yet.

That's a clever way to use that tool :P

Yes, well the script would be fired manually, but the idea is to run it before requesting a code review.

Some here don't like much the idea of having a difference between what is reviewed and what is then committed. (Which is understandable.)

Yep!

In fact, the script I have now runs cppcheck, cppclean, our JSON files and XML files so that they are "clean". So this script would be improved by adding clang-tidy and clang-format.

yes, a pre-commit hook would work. (But as I said, some people here don't like the idea of having something that modify their stuff before they commit).

I suppose one could set it up to run only when committing to branches and not to the trunk.

Yeah; I don't know. It sure is not instant.

instantaneous?

Hmm. Well most of us use VS, but my boss uses like ultraedit or his favourite unix editor like vi or something.

This could also be split; what's fast can be made as a hook, and what's slower needs a manual intervention.

Yeah, we don't really have that, though.

Well... we do, but that's something I don't want to touch.

I figured that adding the changes to the "reports" of these tools to the code review would help highlight more potential defects.

Quick question here... Is there a performance advantage in an ecs if you go pure ? So components without any methods inside ? And if theres one, what if the component references a class which contains the method to call ? Like "Rotation{ vector3 rot }... rot.rotateBy()" ?

Do you know anything about cache misses?

Not really... we arent that deep into it yet. But i know that the cache loads the next lines of code and jumping around fills up the cache, which means he loads fewer lines which results in performance loss... thats why a ecs is more performant. Atleasts thats how i understood it

Yes, that's basically that.

There is instruction cache and data cache.

If your data is contiguous in memory, you have a gain. If you don't have to jump a lot instruction-wise, you have a small gain.

I don't think whether the function is in the component or in the system matters much in this case.

The best thing you could do is try and profile both and see if there is a difference.

I suppose instructions take less space in memory and it is why it is less important.

Thanks :) Currently my transform component looks like this : "Transform{ x,y,z }" but i wanted to switch to a library which implements tons of the vector and matrix operations to speed up the development... but then my transform would simply reference one of their vector classes and use their methods instead, thats why im asking

Reference, its java :(

Do you have anything like "inlining" in java?

@Tyyppi_77 Yes. Also, I think there is something like path prediction?

Thats actually a good question... i dont even know what inlining is, i used it in c++ sometime ago with templates... without inline the templates didnt worked properly

Yes; it was just a point that you need to check how the data is laid out and used more than how you structure the code ;)

Yes, I don't know.

IMHO, ECS should be used first to help development speed, and only then with performance.

I also have no clue... atleast it helps the development speed ^ im using "artemis odb", which had a pretty good benchmark compared to other java ecs implementations

When I profile my app, I notice that the getComponent<GfxTransform>() is the heaviest one. It is also the most used, hence the high time. BUt it's still nothing compared to how render() does...

@genaray Keep on working on your thing, profile and fix what seems optimizable ;)

You'll probably not see any major difference between both approaches.

And even if you learn something with this project, it might be different with the next project.

Alright thanks ^^

In response to inlining with Java: normanmaurer.me/blog/2014/05/15/Inline-all-the-Things

Take away points:

- JIT automatically detects "hot" methods and try to inline them for you

- you can influence what is considered "hot"

- you can manually inline the code yourself

- but Java doesn't have an inline keyword like some languages

- and the JIT might opt not to inline some hot code for various performance reasons