pouët.net

nVidia RTX innovation or marketing bullshit?

category: code [glöplog]
?

Are you doing this deliberately?
added on the 2020-12-09 18:14:18 by Gargaj Gargaj
and it actually looks a lot nicer than vulkan!
Deliberately? I'm reading through the dx12 tracing tutorial after looking through the metal equivalents, and the dx12 version has a ton of boilerplate code code and things like that which are badly named to the point where I'd have to look at the headers or docs just to figure out what it is. It's not vulkan bad, but it's a long way from good.

I like my coding to be enjoyable. I like to write stuff that's clear and concise and makes things happen, not tons of boilerplate that makes me think middleware might be a good option. To me that's just having some respect for my own time.
added on the 2020-12-09 18:34:02 by alia alia
middleware? try coding in ue4 and eat those words.. :)
added on the 2020-12-09 20:14:22 by smash smash
Ue4 or vulkan tho? 😬
added on the 2020-12-09 20:41:26 by alia alia
Ue4 is easy, you just have to connect boxes!

Seriously though, do you guys think the ray-tracing stuff will ever come to OpenGL?
added on the 2020-12-11 19:01:59 by BarZoule BarZoule
I'm doubtful... since apple ditched GL it's no longer a viable cross-platform option (too many iOS devices). If you're going high end dx12/metal/vulkan are better options... so there's kind of just the lower end games and hobbyists left with openGL. And I can't see Nvidia + AMD caring enough :/
added on the 2020-12-11 19:10:38 by alia alia
Quote:
Seriously though, do you guys think the ray-tracing stuff will ever come to OpenGL?

on opensource AMD driver it possible

on Nvidia never, funny that RTX in Vulkan was an extension, and extensions should be super easy to do "backporting"... so only reason is Nvidia drop OpenGL support already years ago

I saw how people do "bridges" between DX12/Vulkan/OGL include RTX, it easy to send any data(like data needed for RTX, and read the output in OpenGL) between APIs ( I mean in real-time, not copying anything, all data in GPU mem)
added on the 2020-12-12 01:26:45 by Danilw Danilw
I had no idea Vk/GL Interop was a thing, good to know!
But I still hope they'll have put ray tracing in GL by the time I get to it...
I don't know why, but I prefer trusting drivers to custom low-level code. It's not just about platforms, but also future-proofing.
added on the 2020-12-12 06:21:19 by BarZoule BarZoule
Having had the pleasure of using an RTX card for Blender rendering with Optix.

I can say that it definitely is worth it to get an RTX card.
added on the 2020-12-18 15:50:41 by SCA SCA
Does anybody already know something about the competing tech from AMD (DXR)?
Is it comparable?
added on the 2021-01-08 12:37:31 by benJam benJam
Quote:
Does anybody already know something about the competing tech from AMD (DXR)?
Is it comparable?

it is
but it more complicated to do it "same or faster than Nvidia RTX

lots of work Nvidia do by their driver/software tricks

when on AMD same RTX code will be slower like by 10-20% because AMD do not do "hidden optimizations"

more info about RDNA 2 raytracing and optimizations you can check on this link (I saw this link in twitter, found interesting)
https://interplayoflight.wordpress.com/2020/12/27/rdna-2-hardware-raytracing/

that RTX Quake2 on AMD work ~20% up to 40% slower than on Nvidia
added on the 2021-01-08 22:27:06 by Danilw Danilw

login