[llvm-commits] CVS: llvm/lib/Target/TargetMachine.cpp
Nate Begeman
natebegeman at mac.com
Wed Jul 6 11:59:15 PDT 2005
Changes in directory llvm/lib/Target:
TargetMachine.cpp updated: 1.39 -> 1.40
---
Log message:
First round of support for doing scalar FP using the SSE2 ISA extension and
XMM registers. There are many known deficiencies and fixmes, which will be
addressed ASAP. The major benefit of this work is that it will allow the
LLVM register allocator to allocate FP registers across basic blocks.
The x86 backend will still default to x87 style FP. To enable this work,
you must pass -enable-sse-scalar-fp and either -sse2 or -sse3 to llc.
An example before and after would be for:
double foo(double *P) { double Sum = 0; int i; for (i = 0; i < 1000; ++i)
Sum += P[i]; return Sum; }
The inner loop looks like the following:
x87:
.LBB_foo_1: # no_exit
fldl (%esp)
faddl (%eax,%ecx,8)
fstpl (%esp)
incl %ecx
cmpl $1000, %ecx
#FP_REG_KILL
jne .LBB_foo_1 # no_exit
SSE2:
addsd (%eax,%ecx,8), %xmm0
incl %ecx
cmpl $1000, %ecx
#FP_REG_KILL
jne .LBB_foo_1 # no_exit
---
Diffs of the changes: (+6 -0)
TargetMachine.cpp | 6 ++++++
1 files changed, 6 insertions(+)
Index: llvm/lib/Target/TargetMachine.cpp
diff -u llvm/lib/Target/TargetMachine.cpp:1.39 llvm/lib/Target/TargetMachine.cpp:1.40
--- llvm/lib/Target/TargetMachine.cpp:1.39 Fri May 13 14:48:34 2005
+++ llvm/lib/Target/TargetMachine.cpp Wed Jul 6 13:59:03 2005
@@ -27,6 +27,7 @@
bool NoExcessFPPrecision;
int PatternISelTriState;
bool UnsafeFPMath;
+ bool PICEnabled;
};
namespace {
cl::opt<bool, true> PrintCode("print-machineinstrs",
@@ -52,6 +53,11 @@
cl::desc("Enable optimizations that may decrease FP precision"),
cl::location(UnsafeFPMath),
cl::init(false));
+ cl::opt<bool, true>
+ EnablePIC("enable-pic",
+ cl::desc("Enable generation of position independant code"),
+ cl::location(PICEnabled),
+ cl::init(false));
};
//---------------------------------------------------------------------------
More information about the llvm-commits
mailing list