Pradeep Garigipati

@polarathene The FNV-1a algorithm is classic for loop algorithm where each iteration's input is dependent on the previous iteration's output unless there is a parallelized version that I am not...

That is true, I was referring to parallelization of the algorithm w.r.t single input stream of bytes. Sure, hashing on a batch of inputs is trivially parallel, not doubt about...

Thank you for digging up the old code. As I said earlier, you are doing individual accesses here in the below section of code ```rust // Iterates through all inputs(columns)...

> Oh? I had thought that it was taking the whole row and computing against that, instead of each being a separate access. I could always rework the data to...

> Is there a example / documentation on approaching that? I would have to think through it. Top of my head, it would most likely start similar to your OpenCL...

Also let us more move this discussion to slack or perhaps another issue as we moved away from the issues original intent some time back :) which is "code examples"....

> Probably would make more sense to rename this issue and recreate the "Add code examples" one as a new issue at this point :sweat_smile: Perhaps :) , I meant...

> I'd appreciate keeping the history of the discussion here, so if wanting to hide the lengthy discussion, please consider marking as outdated/off-topic rather than deleting :) :+1: I am...

I am still contemplating if this would any benefit to the wrapper in general given that serialization and de-serialization are already supported. Hence, the `timeless` milestone.

@umar456 thanks umar, will take a look at it soon.