steakhal accepted this revision. steakhal added a comment. In D110357#3066207 <https://reviews.llvm.org/D110357#3066207>, @ASDenysPetrov wrote:
> Ok. Let's see what the benefits it brings. According to our measurements, it has some effects but is probably difficult to draw clear conclusions. By using the github/martong/constraint_assignor_rem <https://github.com/martong/llvm-project/commit/eb184186e60f74111345d328b51498e696b41f42>, it seems like the runtime is within measurable errors, and the memory consumption and the total result count remained approximately the same. F19624597: image.png <https://reviews.llvm.org/F19624597> F19624602: image.png <https://reviews.llvm.org/F19624602> F19624615: image.png <https://reviews.llvm.org/F19624615> It seems like we have more results since we more often have more concrete values instead of unsimplified symbols, which makes the analyzer 'smarter' and reports more problems. At least, that's my theory. Sorry for not sharing directly the csa-testbench summary HTML, but that could contain sensitive information. If you require, I could repeat the test and publish the results to the publicly reachable demo server to let you all inspect and verify my theory. Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D110357/new/ https://reviews.llvm.org/D110357 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
