Hi,
Norman Kircher Just made me aware of this thread.
I created some code (based on G2CPU 1.4.2 community edition) where I'm hitting 800us to 900us on a RTX 4080 Laptop. (equal to a RTX 4060 desktop)
I purposely didn't do indexing to then pull it back to RAM as I don't know what you want to do after.
You can then use this data for boolean comparison, statistics etc before pulling it out of the GPU.
This code works on CUDA and OpenCL and is also LabVIEW RT compatible.
Let me know if it works for you.
Contact me at natan.biesmans@g2cpu.com if you need a more tailor fit implementation.
Br,
Natan Biesmans
CEO G2CPU the GPU and CPU HPC Toolkit for LabVIEW
LabVIEW Champion, CLA, CPI
Deserialize Data with Nan Removal.vi