ubc-aamodt-group / ray-intersection-predictor Goto Github PK

Hello, I am a new employee of Huawei 2012 laboratory. Recently, I am studying the relevant papers of this research, and I respect your achievements.
I have a few doubts about the code related to the paper. I hope I can get your answer.

1. In the paper, I saw that you used the grid spherical hash function to quantize the origin and direction of the light into a hash value. In Section 4.1 of the paper, I described the relevant information of its construction. I saw the code of the hash function construction in the RayTraceManager.cpp file under the path .\rtao-benchmark\src, but after I read all the codes under the path, in addition to using the function when sorting the light, It is not seen that this function is used to quantize rays when constructing rays
2. Similarly, I didn't see the creation of prediction table, hash value insertion and search in the code under this path
3. In the paper, it is mentioned that when the number of bits n used for the hash larger than a table indexed with m bits, we fold the n-bit hash intom-bits by splitting the hash into ⌈n/m⌉ components and combining them with bitwise-xor, but I have not seen the relevant operation process in the code
   Is the relevant operation built into a file, or do I miss any important information when reading? I hope to get your answer, thank you

Amazing work but I have a few questions for this project.

How do you estimate the latency of the intersection unit in the ray predictor in src/gpgpu-sim/ray-predictor.cc?
That means how did you benchmark the real intersection unit in NVIDIA RT core?
Did you use some micro benchmark to achieve this estimation?

Thanks,
looking forwards to your reply.