Merge pull request #3806 from elasota/fmt-clarifications

Correct FSE probability bit consumption in specification
This commit is contained in:
Yann Collet
2023-11-01 10:59:54 -07:00
committed by GitHub
+2 -1
View File
@@ -1083,7 +1083,8 @@ It depends on :
Presuming an `Accuracy_Log` of 8,
and presuming 100 probabilities points have already been distributed,
the decoder may read any value from `0` to `256 - 100 + 1 == 157` (inclusive).
Therefore, it must read `log2sup(157) == 8` bits.
Therefore, it may read up to `log2sup(157) == 8` bits, where `log2sup(N)`
is the smallest integer `T` that satisfies `(1 << T) > N`.
- Value decoded : small values use 1 less bit :
__example__ :