-
Sub-task
-
Resolution: Fixed
-
P2
-
23
-
Measurement made on host:
- macOS Monterey 12.6.3
- Mac mini (M1, 2020)
-
- Apple M1 - 8 cores (4 perf & 4 effic)
- 16 GB Memory
With the introduction of JDK-8324186... a number of benchmarks are showing regression in macOS-aarch64:
-9% Renaissance-Mnemonics on macOS-aarch64
-5% Renaissance-ParMnemonics on macOS-aarch64
-4% Renaissance-RxScrabble on macOS-aarch64
-4% Renaissance-Scrabble on macOS-aarch64
-1.2% SPECjbb2005-ZGC on macOS-aarch64
-1.2% Crypto-AESGCMBench.encrypt on macOS-aarch64
Regressions were Isolated to jdk-23+8-514 :JDK-8324186 "Use "dmb.ishst+dmb.ishld" for release barrier"
Isolation graphs are attached for Mnemonics and ParMnemonics, but the isolation graphs for all benchmarks above show the regression introduced in jdk-23+8-514
-9% Renaissance-Mnemonics on macOS-aarch64
-5% Renaissance-ParMnemonics on macOS-aarch64
-4% Renaissance-RxScrabble on macOS-aarch64
-4% Renaissance-Scrabble on macOS-aarch64
-1.2% SPECjbb2005-ZGC on macOS-aarch64
-1.2% Crypto-AESGCMBench.encrypt on macOS-aarch64
Regressions were Isolated to jdk-23+8-514 :
Isolation graphs are attached for Mnemonics and ParMnemonics, but the isolation graphs for all benchmarks above show the regression introduced in jdk-23+8-514
- duplicates
-
JDK-8325269 Quadruple dmbs could be emitted after JDK-8324186
-
- Closed
-
- relates to
-
JDK-8328878 Rework [AArch64] use "dmb.ishst + dmb.ishld" for release barrier
-
- Closed
-
-
-
- Closed
-
