projects / oota-llvm.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: a8cb100) | patch
For 16 and 32-bit multiplies, use the IMUL instruction instead of the MUL instruction.
authorChris Lattner <sabre@nondot.org>
Sat, 21 Jun 2003 17:16:58 +0000 (17:16 +0000)
committerChris Lattner <sabre@nondot.org>
Sat, 21 Jun 2003 17:16:58 +0000 (17:16 +0000)
commit0f1c461423ddc3fcea0fd8813b934838cc2b95e5
tree140bada7ea29224a76ddc1349ddb00c2be997219tree | snapshot
parenta8cb10070688cdfbbbb77491edb987f977ca798ecommit | diff
For 16 and 32-bit multiplies, use the IMUL instruction instead of the MUL instruction.
This allows us to not force the use of the EAX/AX registers!

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@6830 91177308-0d34-0410-b5e6-96231b3b80d8
lib/Target/X86/InstSelectSimple.cpp diff | blob | history
lib/Target/X86/X86ISelSimple.cpp diff | blob | history
C/C++ LLVM-based compilers that forbids OOTA behaviors