ART: Add entrypoint and intrinsic for Math.pow().
MathBenchmarks.java#timePow results on taimen's little cores
fixed at frequency 1401600 with forced JIT compilation:
- before:
- X32: 356.33 (@FastNative), 315.39 (@CriticalNative)
- X64: 357.31 (@FastNative), 315.37 (@CriticalNative)
- after (LICM defeats the benchmark):
- X32: 2.88
- X64: 2.87
- after but with kAllSideEffects to prevent LICM:
- X32: 275.42
- X64: 275.67
Test: Rely on TreeHugger.
Bug: 70727450
Change-Id: Iaa31f70acabbd57c163cfeafe02eed67c1348861
diff --git a/runtime/entrypoints_order_test.cc b/runtime/entrypoints_order_test.cc
index 7c912d0..1fdf439 100644
--- a/runtime/entrypoints_order_test.cc
+++ b/runtime/entrypoints_order_test.cc
@@ -238,7 +238,8 @@
EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pAcos, pAsin, sizeof(void*));
EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pAsin, pAtan, sizeof(void*));
EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pAtan, pAtan2, sizeof(void*));
- EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pAtan2, pCbrt, sizeof(void*));
+ EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pAtan2, pPow, sizeof(void*));
+ EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pPow, pCbrt, sizeof(void*));
EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pCbrt, pCosh, sizeof(void*));
EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pCosh, pExp, sizeof(void*));
EXPECT_OFFSET_DIFFNP(QuickEntryPoints, pExp, pExpm1, sizeof(void*));