Enable micro-optimization of fast-unary-ftruncate and
authorRaymond Toy <toy.raymond@gmail.com>
Tue, 22 Oct 2013 01:10:13 +0000 (18:10 -0700)
committerRaymond Toy <toy.raymond@gmail.com>
Tue, 22 Oct 2013 01:10:13 +0000 (18:10 -0700)
commit07e2d61f67dbd0e099c256052ba70358125cc008
treeeb5af59e897637e1357cfd6cf8d8ce005f785b35
parent622b5df431a87ae3c8a816b7c569f5c5ef85a6d7
Enable micro-optimization of fast-unary-ftruncate and
double-float-bits for x86/sse2. This gives x86 the same
micro-optimizations that were available for sparc and ppc.

 o code/kernel.lisp:
   o Enable fast double-float-bits using the vop instead of calling
     double-float-high-bits/double-float-low-bits.

 o compiler/float-tran.lisp:
   o Make fast-unary-ftruncate known to compiler and enable optimizer
     for it.
   o Make double-float-bits known to compiler

 o compiler/x86/float-sse2.lisp:
   o Implement fast-unary-ftruncate for singles and doubles.
   o Implement double-float-bits.
src/code/kernel.lisp
src/compiler/float-tran.lisp
src/compiler/x86/float-sse2.lisp