Another cls based on chunks as cls-rz, but this time it's mt.
Nothing to say, just set up cls.ini, you know how to use a cls.
FP8 is slow, even in store mode because it precompresses data and then compresses it, so in level 0 it just precompresses, you get better result using level 0 fp8+srep+lzma rather than srep+lzma.
You can't set up how much threads to use because cls automatically defines it. Usually it uses all threads except one.
Configuring ini file:
There is already an example inside, but this is how it's done.
cls.ini
[fp8]
Level=5
ChunkSize=64m
Enjoy!
Original fp8 homepage in encode