Details
-
Type:
Bug
-
Status: Resolved
-
Priority:
P3
-
Resolution: Not an Issue
-
Affects Version/s: 11
-
Fix Version/s: None
-
Component/s: performance
-
Labels:
Description
11-b4 is showing a 2-4% performance regression on Solaris/SPARC in the AES-GCM encrypt/decrypt microbenchmarks:
• micro.benchmarks.crypto.small.AESGCMBench.decrypt-algorithm:AES_GCM_NoPadding-dataSize:1024-keyLength:128-provider
• micro.benchmarks.crypto.small.AESGCMBench.encrypt-algorithm:AES_GCM_NoPadding-dataSize:1024-keyLength:128-provider
The regressions reproduce in:
• Triage (ie, re-run on the same machine)
• Re-triage (ie, run on a different machine)
JVMswap results are inconclusive (ie, about half the regression is recovered). Hard to say if the cause is in the JVM , the JDK, or both.
• micro.benchmarks.crypto.small.AESGCMBench.decrypt-algorithm:AES_GCM_NoPadding-dataSize:1024-keyLength:128-provider
• micro.benchmarks.crypto.small.AESGCMBench.encrypt-algorithm:AES_GCM_NoPadding-dataSize:1024-keyLength:128-provider
The regressions reproduce in:
• Triage (ie, re-run on the same machine)
• Re-triage (ie, run on a different machine)
JVMswap results are inconclusive (ie, about half the regression is recovered). Hard to say if the cause is in the JVM , the JDK, or both.