aengelke wrote:
Another data point: I can build LLVM (incl. tools) in 6-7 minutes on my laptop
now (M2 Macbook Air, 8 cores); building unit-tests take another 2.5mins.
Interestingly, with PCH optimizations are comparably expensive here:
```
ExecuteCompiler 3272
Frontend 1533
Source 476
PerformPendingInstantiations 570
CodeGen Function 300
Backend 1699
Optimizer 1056
InlinerPass 264 (90.27us per run; 2934128 runs)
InstCombinePass 163 (10.69us per run; 15292003 runs)
SimplifyCFGPass 67 (3.77us per run; 18032531 runs)
EarlyCSEPass 58 (9.87us per run; 5880387 runs)
SROAPass 51 (5.69us per run; 9020246 runs)
BlockFrequencyAnalysis 47 (9.41us per run; 5048403 runs)
BranchProbabilityAnalysis 33 (6.72us per run; 5048403 runs)
AAManager 30 (2.57us per run; 11970407 runs)
MemorySSAAnalysis 29 (7.63us per run; 3921385 runs)
IPSCCPPass 28 (12019.39us per run; 2410 runs)
PostOrderFunctionAttrsPass 23 (3.98us per run; 5868256 runs)
CodeGenPasses 639
wall-time (8c/8t) ~400s
```
Most expensive CUs >15s:
```
15.769s unittests/SandboxIR/SandboxIRTest.cpp
16.174s lib/CodeGen/SelectionDAG/DAGCombiner.cpp
16.238s lib/Transforms/IPO/MemProfContextDisambiguation.cpp
16.457s unittests/Frontend/OpenMPIRBuilderTest.cpp
16.702s unittests/ADT/DenseMapTest.cpp
16.779s lib/Transforms/IPO/AttributorAttributes.cpp
16.920s lib/Target/AArch64/AArch64ISelLowering.cpp
28.694s tools/llvm-readobj/dir/ELFDumper.cpp
28.865s lib/Target/X86/X86ISelLowering.cpp
30.295s lib/Passes/PassBuilder.cpp
33.410s lib/Transforms/Vectorize/SLPVectorizer.cpp
54.741s unittests/Frontend/OpenMPDecompositionTest.cpp
```
https://github.com/llvm/llvm-project/pull/173868
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
