Measured to be more performant when reading bitcode.
Details
Details
Diff Detail
Diff Detail
Event Timeline
Comment Actions
That's a good point, I've been focused on the Reader only because this is where my current bottle-neck is :)
Comment Actions
What test have you used? I will give this a quick try on "opt -disable-verifier -disable-output" on a LTO of clang.
Comment Actions
LGTM.
The results I got with 10 runs were
master:
11.167095958 seconds time elapsed ( +- 0.34% )
patch:
10.860838399 seconds time elapsed ( +- 0.39% )