In the future if you could add Nvidia RTX support so Cycles could use the Tensor cores you might get up to an 8x speedup with Cycles using RTX alone. With Octane they just came out with RTX support and they get about a 2x speedup. Not quite 8x, but still with your E-Cycles doing 2x if you can get 2x more it would be 4x speedup.
I think a lot of people here are a little unbelieving of your results because they are so good and because you don’t have an easier to understand explanation of how you do it. Is it only a good denoiser that speeds it up? An optimization of the raytracing code? A faster memory optimization? A pure streamlining of the Cycles code? I hope you see where I’m going with this. Add to the first post a simple explanation of how you are speeding up Cycles instead of only having, it’s all in the class if you want to spend a few days learning it, would go a long way to gain trust from the start. In the very first lines it should be.
E-Cycles speeds up Cycles with:
- Better memory optimizations
- Faster raytracing algorithm
- Etc. whatever you did to make it faster