sc94597 said:
It gets even trickier than that. Depending on the data-paths available in the shading units, "maximum TFLOPS" can be counted in a way that the TFLOPS number becomes inflated (with respect to gaming workloads, not say scientific computing.) Nvidia is counting FP32 and FP32/INT32 cores in the estimate of their TFLOPS since Ampere, for example. This is despite in a gaming load 26% of the calculations are likely INT32 and not FP32 (according to Nvidia's estimate.) And then there is the complication that is ray-tracing and "ray-tracing tflops." Using Tflops as a metric of real-world gaming performance between different architectures, platforms, and especially the brands is just not a good idea. I might have missed it though, where were people discussing performance per TFLOPS? |
Maybe I misread and I don't feel like going back to reading every comment, but I'm pretty sure Soundwave suggested that the Switch 2 could do more than Series S/X and PS5 (which he at least twice referred to as a "RDNA 1.5") per a TFLOPS due to the more advanced architecture and technologies like DLSS. Basically, he thinks Switch can "effectively" do nearly as much with 3 or 3.5 TFLOPS as Series S can with 4TFLOPS.
I know very well that TFLOPS is a misleading metric, but it's what countless people still use to know how powerful a console is on paper. And they determine a TFLOPS efficiency based on how it materializes in games (resolution, fps, settings etc). I know it's not that simple and countless factors go into this.
Last edited by Kyuu - on 11 September 2023
