From b4c05977c28c38d2f81b48d0cb15559dc3d05564 Mon Sep 17 00:00:00 2001 From: Elliott Hughes Date: Wed, 24 Feb 2010 16:36:18 -0800 Subject: Optimize more easy multiplications by constants. Rather than make these changes in the libraries (*10 being a common case), let's do them once and for all in the JIT. The 2^n-1 case could be better if we generated RSB instructions, but the current "fake" RSB is still better than a full multiply. Thumb doesn't support reg/reg/reg/shift instructions, so we can't optimize the "population count <= 2" cases (such as *10) there. Tested on sholes, passion, and passion-running-sapphire (and visually inspected to check we weren't trying to generate Thumb2 instructions there). Also tested with the self-verifier. --- vm/compiler/codegen/arm/ArmLIR.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'vm/compiler/codegen/arm/ArmLIR.h') diff --git a/vm/compiler/codegen/arm/ArmLIR.h b/vm/compiler/codegen/arm/ArmLIR.h index b329aedb7..7d262215e 100644 --- a/vm/compiler/codegen/arm/ArmLIR.h +++ b/vm/compiler/codegen/arm/ArmLIR.h @@ -706,7 +706,7 @@ extern ArmEncodingMap EncodingMap[kArmLast]; /* * Each instance of this struct holds a pseudo or real LIR instruction: - * - pesudo ones (eg labels and marks) and will be discarded by the assembler. + * - pseudo ones (eg labels and marks) and will be discarded by the assembler. * - real ones will be assembled into Thumb instructions. * * Machine resources are encoded into a 64-bit vector, where the encodings are -- cgit v1.2.3