Without clamping gamma.

Cosine-Weighted Importance Sampling. Min samples: 16. Max samples: 512. Shadow min samples: 1. Shadow max samples: 4. Glossy reflection and refraction samples: 1. GI samples: 1. 4 min 31.322 sec. It is noisier than using 1024 max samples, but from afar, its not that different.

Cosine-Weighted Importance Sampling. Min samples: 16. Max samples: 1024. Shadow min samples: 1. Shadow max samples: 4. Glossy reflection and refraction samples: 1. GI samples: 1. 8 min 47.517 sec.

Uniform Hemisphere Sampling. Min samples: 16. Max samples: 64. Shadow min samples: 1. Shadow max samples: 4. Glossy reflection and refraction samples: 1. GI samples: 1. 26.6087 secs. Noisier than the cosine-weighted sampling.

Got it! Turned out to be an issue with clamping. Cosine-Weighted Importance Sampling. Min samples: 16. Max samples: 64. Shadow min samples: 1. Shadow max samples: 4. Glossy reflection and refraction samples: 1. GI samples: 1. 29.0197 secs.

Got the emission working properly, but the image is too bright as a whole....possibly an issue with the attenuation?

Cool render. Got this when I used the mirror reflection vector R = reflect-V, N as the sampling axis instead of the surface normal.

Got something! Still missing the emissive object in the reflection, and its effect on the white sphere/wall, but very close.