summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorTobias Grosser <tobias@grosser.es>2017-02-12 10:53:02 +0000
committerTobias Grosser <tobias@grosser.es>2017-02-12 10:53:02 +0000
commit288c450cf697c90fbfc42258e37ead6ec8460465 (patch)
tree07b898f28cdb3d48896d2f16a70093d71ec66580
parent9fe37df27ccd4ebe26b4dee76ed3e43f1616bec7 (diff)
downloadbcm5719-llvm-288c450cf697c90fbfc42258e37ead6ec8460465.tar.gz
bcm5719-llvm-288c450cf697c90fbfc42258e37ead6ec8460465.zip
[ScopDetectDiagnostics] Do not format unnamed array names
Formatting unnamed array names is expensive in LLVM as the this requires deriving the numbered virtual instruction name (e.g., %12) for an llvm::Value, which is currently not implemented efficiently. As instruction numberes anyhow do not really carry a lot of information for the user, we just print 'unknown' instead. This change reduces the scop detection time from 24 to 19 seconds, for one of our large-scale inputs. This is a reduction by 21%. llvm-svn: 294894
-rw-r--r--polly/lib/Analysis/ScopDetectionDiagnostic.cpp2
1 files changed, 1 insertions, 1 deletions
diff --git a/polly/lib/Analysis/ScopDetectionDiagnostic.cpp b/polly/lib/Analysis/ScopDetectionDiagnostic.cpp
index 8f8401af56b..b2b566bdffd 100644
--- a/polly/lib/Analysis/ScopDetectionDiagnostic.cpp
+++ b/polly/lib/Analysis/ScopDetectionDiagnostic.cpp
@@ -443,7 +443,7 @@ std::string ReportAlias::formatInvalidAlias(std::string Prefix,
assert(V && "Diagnostic info does not match found LLVM-IR anymore.");
if (V->getName().size() == 0)
- OS << "\"" << *V << "\"";
+ OS << "\" <unknown> \"";
else
OS << "\"" << V->getName() << "\"";
OpenPOWER on IntegriCloud