Tue, 19 Jan 2016 12:04:18 +0100 |
Doug Simon |
use JVMCI_VERSION constant to avoid too much eagerness when loading mx_graal_8
|
Tue, 19 Jan 2016 12:20:24 +0100 |
Manuel Rigger |
remove redundant modifiers.
|
Tue, 19 Jan 2016 10:55:27 +0100 |
Manuel Rigger |
rename JVMCIJDKConfig to reflect the JVMCI renaming changes.
|
Tue, 19 Jan 2016 10:28:33 +0100 |
Josef Eisl |
TraceRA: avoid streams in TraceRegisterAllocationPhase.
|
Tue, 19 Jan 2016 10:24:05 +0100 |
Josef Eisl |
TraceRA: do not use streams in TraceBuilder asserts.
|
Tue, 19 Jan 2016 10:05:58 +0100 |
Josef Eisl |
TraceRA: don't use streams for TraceBuilderResult.incoming*Edges().
|
Mon, 18 Jan 2016 10:58:42 -0800 |
Tom Rodriguez |
Report number of inlined bytecodes when compilation succeeds
|
Mon, 18 Jan 2016 18:37:09 +0100 |
Manuel Rigger |
Branch probability fix for usage in ConditionalNode.
|
Mon, 18 Jan 2016 10:29:19 +0100 |
Roland Schatz |
Update import: Move CompilationResult and DataSection from JVMCI to Graal.
|
Fri, 15 Jan 2016 16:30:38 +0100 |
Roland Schatz |
Update jvmci import.
|
Fri, 15 Jan 2016 18:27:58 +0100 |
Thomas Wuerthinger |
Simplification of unreachable node tracking in SchedulePhase: Unreachable node bitset is invert of visited bitset.
|
Fri, 15 Jan 2016 18:20:21 +0100 |
Thomas Wuerthinger |
Clean up in the scheduling phase.
|
Fri, 15 Jan 2016 18:02:41 +0100 |
Thomas Wuerthinger |
Remove loop explosion and merging during graph building and FastPE. GraphPE is the successor with equivalent capabilities.
|
Thu, 14 Jan 2016 15:18:33 -0800 |
Christian Wimmer |
GraphPE: refactoring of loop detection
|