https://bugs.llvm.org/show_bug.cgi?id=33054
Bug ID: 33054
Summary: `upgradeCUVariable` is non linear
Product: libraries
Version: trunk
Hardware: PC
OS: Windows NT
Status: NEW
Severity: enhancement
Priority: P
Component: DebugInfo
Assignee: [email protected]
Reporter: [email protected]
CC: [email protected], [email protected],
[email protected], [email protected],
[email protected]
I have in my hands a modestly sized piece of bitcode [250MB] produced with 3.8
(dumped after the mover merged input modules at LTO time) that takes circa
1000secs to be parsed. The overall time spent in the optimizer is about 1/10th
of that (120secs).
The profiler shows all the time goes in `parseMetadata` while materializing, in
particular in `upgradeCUVariables` on metadata operations.
- 94.82% llvm::MetadataLoader::MetadataLoaderImpl::upgradeCUVariables
- 42.72% llvm::GlobalObject::getMetadata
- 21.45% llvm::MDGlobalAttachmentMap::get
[...]
+ 29.10% llvm::GlobalObject::addMetadata
+ 11.56% llvm::GlobalObject::eraseMetadata
[...]
I'm afraid I'm not in a position to share the bitcode as it's owned by one of
our internal game teams, but I may try to reproduce on something publicly
available.
Adrian, you committed this code, any ideas on how we can make it more
efficient?
--
You are receiving this mail because:
You are on the CC list for the bug._______________________________________________
llvm-bugs mailing list
[email protected]
http://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-bugs
