commit e69d7a9bc042423a89a6d7b6b1c2236911f67b59
author Aart Bik <ajcbik@google.com> Fri Feb 17 11:48:23 2017 -0800
committer Aart Bik <ajcbik@google.com> Fri Feb 17 11:48:23 2017 -0800
tree 06b8836ff5f24fc625be4c40b996b6cefca0258b
parent 6a669aac63ffc17b5e903aa4a1f285fe338eadcf

Added a few more integral SIMD extensions for x86/x86_64 (SSE).

Rationale:
ART vectorizer needs a couple of extra SIMD operations before
sending out the larger general CL.

Test: assembler_x86[_64]_test
Bug: 34083438
Change-Id: Id156283424ad311e6109b360efcd409c671cd5b7

compiler/utils/x86/assembler_x86.h

diff --git a/compiler/utils/x86/assembler_x86.h b/compiler/utils/x86/assembler_x86.h
index 4343e2e..2999599 100644
--- a/compiler/utils/x86/assembler_x86.h
+++ b/compiler/utils/x86/assembler_x86.h
@@ -408,14 +408,9 @@
   void movsd(const Address& dst, XmmRegister src);
   void movsd(XmmRegister dst, XmmRegister src);
 
-  void psrlq(XmmRegister reg, const Immediate& shift_count);
-  void punpckldq(XmmRegister dst, XmmRegister src);
-
   void movhpd(XmmRegister dst, const Address& src);
   void movhpd(const Address& dst, XmmRegister src);
 
-  void psrldq(XmmRegister reg, const Immediate& shift_count);
-
   void addsd(XmmRegister dst, XmmRegister src);
   void addsd(XmmRegister dst, const Address& src);
   void subsd(XmmRegister dst, XmmRegister src);
@@ -436,10 +431,20 @@
   void movdqa(const Address& dst, XmmRegister src);  // store aligned
   void movdqu(const Address& dst, XmmRegister src);  // store unaligned
 
-  void paddd(XmmRegister dst, XmmRegister src);  // no addr variant (for now)
+  void paddb(XmmRegister dst, XmmRegister src);  // no addr variant (for now)
+  void psubb(XmmRegister dst, XmmRegister src);
+
+  void paddw(XmmRegister dst, XmmRegister src);
+  void psubw(XmmRegister dst, XmmRegister src);
+  void pmullw(XmmRegister dst, XmmRegister src);
+
+  void paddd(XmmRegister dst, XmmRegister src);
   void psubd(XmmRegister dst, XmmRegister src);
   void pmulld(XmmRegister dst, XmmRegister src);
 
+  void paddq(XmmRegister dst, XmmRegister src);
+  void psubq(XmmRegister dst, XmmRegister src);
+
   void cvtsi2ss(XmmRegister dst, Register src);
   void cvtsi2sd(XmmRegister dst, Register src);
 
@@ -489,6 +494,24 @@
   void shufps(XmmRegister dst, XmmRegister src, const Immediate& imm);
   void pshufd(XmmRegister dst, XmmRegister src, const Immediate& imm);
 
+  void punpcklbw(XmmRegister dst, XmmRegister src);
+  void punpcklwd(XmmRegister dst, XmmRegister src);
+  void punpckldq(XmmRegister dst, XmmRegister src);
+  void punpcklqdq(XmmRegister dst, XmmRegister src);
+
+  void psllw(XmmRegister reg, const Immediate& shift_count);
+  void pslld(XmmRegister reg, const Immediate& shift_count);
+  void psllq(XmmRegister reg, const Immediate& shift_count);
+
+  void psraw(XmmRegister reg, const Immediate& shift_count);
+  void psrad(XmmRegister reg, const Immediate& shift_count);
+  // no psraq
+
+  void psrlw(XmmRegister reg, const Immediate& shift_count);
+  void psrld(XmmRegister reg, const Immediate& shift_count);
+  void psrlq(XmmRegister reg, const Immediate& shift_count);
+  void psrldq(XmmRegister reg, const Immediate& shift_count);
+
   void flds(const Address& src);
   void fstps(const Address& dst);
   void fsts(const Address& dst);