LabVIEW
How to make this histogram function faster
Yes, overall, you only have 2862 nonzero bins, a very small percentage. Hard to tell without knowing more about the image sensor and how the data is acqured processed and quantized to 12 bits. Maybe the "12bit" specification is a bit "generous" but 2862 is not that different from 4k so it is probably OK. 😉
Couple of things:
1. The parallel loop does seem to speed things up if you've shifted the data back to 12 bits - about 50% improvement on my machine.
2. The distribution of values "missing" is not constant - at the start, there's usually 2 values between gaps, but sometimes 1, but later it changes to 8, and then bigger again:
This is probably not too surprising as the output from a supposedly 12-bit sensor - there are usually values that cannot be obtained. It's just to do with the way the sensor output is quantized, which also leads to some of the other patterns in the data where certain values are accentuated rather than eliminated:
