On Tue, Jul 7, 2015 at 8:35 PM, Uros Bizjak wrote:
> BT has *slightly* higher latency than TEST (0.33 vs. 0.25 cycles on a
> modern processor), so I have limited the conversion to -Os in case the
> bit-test is in the low 32 bits.
A small update, in case of -Os, unpatched compiler generates andb
Hello!
After recent x86 EXTZ/EXTZV improvements, we can extend BT splitters
to generate BT instruction with immediate operands. The improvement
can be seen with attached testcases.
The benefit is obvious for BT with immediates 32 <= n <= 63:
0: 48 b8 00 00 00 00 00movabs $0x1000