4 ms·
I’m guessing it’s because they’re using all the computing power the GPU has to offer in CUDA mode, as opposed to sharing the GPU with other functions (when in R
by sheepscreek 1y ago
I’m guessing it’s because they’re using all the computing power the GPU has to offer in CUDA mode, as opposed to sharing the GPU with other functions (when in RTX).
- colechristensen 1y agoYup this is an "assume spherical cow" situation where it's not dishonest, but you can't draw any real world conclusions from the experiment unless you happen to be working in a very restricted space.
- ChocolateGod 1y agoWouldn't you need to in a real world scenario make the CUDA cores aware of the game geometry adding more work on the CPU?
- touisteur 1y agoIdeally you don't make the cuda cores aware but rather the ray-tracing circuitry. RT cores are designed to perform ray-triangle intersections in a BVH. You get the teraflops and memory bandwidth (or more of it) if you fit the RT-core computing model. And in most cases it's ok to spend time on one CPU function (creating and loading the BVH) against the hundred thousands of frames you'll be drawing on GPU.
- colechristensen 1y agoA whole lot of stuff is going on during gaming and graphics rendering with trick upon trick to squeeze out every last bit of performance. Unless you're an expert in a graphics rendering stack or a game engine it's hard to have these conversations in a meaningful way.
- atq2119 1y agoMore likely it's because the scene they're using is completely unrepresentative of what people are interested in: almost no triangles, primarily procedural nodes (for spheres), and in general a fairly simple scene.