-
Bug
-
Resolution: Fixed
-
P3
-
1.2.0
-
None
-
1.1.6
-
generic
-
windows_nt
-
Not verified
Issue | Fix Version | Assignee | Priority | Status | Resolution | Resolved In Build |
---|---|---|---|---|---|---|
JDK-2020115 | 1.2.0 | Deepa Viswanathan | P3 | Resolved | Fixed | 1.2beta4 |
All the benchmarks report 0.0010 secs on Win32 with the JIT on. This happens
with JDK1.2beta4 and JDK1.1.6
The following seems to be the problem -
At the end of a benchmark run a BenchmarkTime is contructed.
BenchmarkTime time = new BenchmarkTime(System.currentTimeMillis() - startTime);
and here's the constructor --
public BenchmarkTime (long millis){
if (millis <= 0)
millis = 1;
this.millis = millis;
}
it turns out millis value is 0. But, if this value is printed out
using System.out.println before or in the constructor, the right
value is reported.
- backported by
-
JDK-2020115 SpecJava v18 reports bogus timing results with JIT on.
- Resolved