Changeset 62093 in vbox for trunk

Timestamp:

Jul 7, 2016 12:43:07 AM (9 years ago)

Author:

vboxsync

Message:

IEM: Saved 1K (except 32-bit) by doing full+unaligned imm16, imme32 and imm64 loads instead of byte by byte.

File:

• trunk/src/VBox/VMM/VMMAll/IEMAll.cpp (modified) (2 diffs)

trunk/src/VBox/VMM/VMMAll/IEMAll.cpp

@@ -1714 +1714 @@
         uint8_t offOpcode = pVCpu->iem.s.offOpcode;
 # ifdef IEM_USE_UNALIGNED_DATA_ACCESS
-        *pu16 = (uint16_t const *)&pVCpu->iem.s.abOpcode[offOpcode];
+        *pu16 = *(uint16_t const *)&pVCpu->iem.s.abOpcode[offOpcode];
 # else
         *pu16 = RT_MAKE_U16(pVCpu->iem.s.abOpcode[offOpcode], pVCpu->iem.s.abOpcode[offOpcode + 1]);
@@ -1740 +1740 @@
         pVCpu->iem.s.offOpcode = (uint8_t)offOpcode + 2;
 # ifdef IEM_USE_UNALIGNED_DATA_ACCESS
-        *pu16 = (uint16_t const *)&pVCpu->iem.s.abOpcode[offOpcode];
+        *pu16 = *(uint16_t const *)&pVCpu->iem.s.abOpcode[offOpcode];
 # else
         *pu16 = RT_MAKE_U16(pVCpu->iem.s.abOpcode[offOpcode], pVCpu->iem.s.abOpcode[offOpcode + 1]);