I applied this to some pieces of Daala's entropy coder: https://review.xiph.org/763/#msg2 https://review.xiph.org/763/#msg2 The output needs a bit of manual re

I applied this to some pieces of Daala's entropy coder: https://review.xiph.org/763/#msg2 https://review.xiph.org/763/#msg2 The output needs a bit of manual review (see the mixing of 32-bit and 16-bit accesses to the same register in the second example), but it was definitely able to find something non-obvious in a sequence far too long for exhaustive search.