they're looking for a compressor that can do more than 200MB/s on a 10mW machine (that's including radio, so it has to run on a CPU clocked like original 8086) and yield 200x size improvement. speaking from the perspective of a data compression person, this is completely unrealistic. the best statistical models that i have on hand yield ~7x compression ratio after some tweaking, but they won't run under these constraints.
I thought 200x is too extreme as well. In compression literature, is there a way to estimate the upper limit on lossless compressibility of a given data set?
There is not, because there could always be some underlying tricky generator that you just haven't discovered, and discovering that pattern is basically equivalent to solving the halting problem. (See https://en.wikipedia.org/wiki/Kolmogorov_complexity#Uncomput...)
As a trivial example, if your dataset is one trillion binary digits of pi, it is essentially incompressible by any regular compressor, but you can fit a generator well under 1 kB.
The same, since lossy compression can never be worse than lossless compression. (Also, it is more complex since you have to define your loss somehow. These Neuralink samples seemingly come as .wav files, but you probably wouldn't want to encode them with MP3!)
