diff options
| author | Stanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com> | 2021-02-22 12:25:30 -0800 |
|---|---|---|
| committer | Stanislav Mekhanoshin <Stanislav.Mekhanoshin@amd.com> | 2021-02-23 10:22:31 -0800 |
| commit | d1b92c91afd0be8939bddbf04f55ec53cf29227a (patch) | |
| tree | 668581a7596c93b46841c1ec2ee67b7ccbd90c3b /llvm/lib/IR/Module.cpp | |
| parent | 5e54bef4d291d4353bb26e0782539b6f79625f68 (diff) | |
| download | llvm-d1b92c91afd0be8939bddbf04f55ec53cf29227a.zip llvm-d1b92c91afd0be8939bddbf04f55ec53cf29227a.tar.gz llvm-d1b92c91afd0be8939bddbf04f55ec53cf29227a.tar.bz2 | |
[AMDGPU] Set threshold for regbanks reassign pass
This is to limit compile time. I did experiments with some
inputs and found that compile time keeps reasonable for this
pass if we have less than 100000 virtual registers and then
starts to explode somewhere between 100000 and 150000.
Differential Revision: https://reviews.llvm.org/D97218
Diffstat (limited to 'llvm/lib/IR/Module.cpp')
0 files changed, 0 insertions, 0 deletions