pcc wrote:

The performance impact of this change on firefox-x64 on an M2 Ultra was
```
    N           Min           Max        Median           Avg        Stddev
x 2801     1.2972602     1.4316316     1.3621663     1.3620049   0.019557658
+ 2801     1.2877738     1.4305345     1.3634474     1.3630622   0.019953389
Difference at 95.0% confidence
        0.00105723 +/- 0.00103472
        0.077623% +/- 0.0759707%
        (Student's t, pooled s = 0.0197565)
```
The delta is fairly small so I don't think that performance is a strong reason 
not to make this change. That being said, after having given this more thought 
I do feel that the compatibility reason given in 
https://github.com/llvm/llvm-project/pull/133531#issuecomment-2795136942 has 
some merit so I intend to rework this with a new relocation type for directly 
producing IRELATIVE in the output.

https://github.com/llvm/llvm-project/pull/133531
_______________________________________________
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits

Reply via email to