[x64] Introduce FMA3 instructions on scalar data elements.

test/cctest/test-disasm-x64.cc

Index: test/cctest/test-disasm-x64.cc
diff --git a/test/cctest/test-disasm-x64.cc b/test/cctest/test-disasm-x64.cc
index fac9226279d7833fa72e386d05c5ebb7d4fe6d03..2ff1c93a5719d54990d7710f54657bcb8f972542 100644
--- a/test/cctest/test-disasm-x64.cc
+++ b/test/cctest/test-disasm-x64.cc
@@ -51,7 +51,7 @@ TEST(DisasmX64) {
   CcTest::InitializeVM();
   Isolate* isolate = CcTest::i_isolate();
   HandleScope scope(isolate);
-  v8::internal::byte buffer[2048];
+  v8::internal::byte buffer[4096];
   Assembler assm(isolate, buffer, sizeof buffer);
   DummyStaticFunction(NULL);  // just bloody use it (DELETE; debugging)
 
@@ -394,6 +394,14 @@ TEST(DisasmX64) {
     __ xorps(xmm0, Operand(rbx, rcx, times_4, 10000));
 
     // Arithmetic operation
+    __ addss(xmm1, xmm0);
+    __ addss(xmm1, Operand(rbx, rcx, times_4, 10000));
+    __ mulss(xmm1, xmm0);
+    __ mulss(xmm1, Operand(rbx, rcx, times_4, 10000));
+    __ subss(xmm1, xmm0);
+    __ subss(xmm1, Operand(rbx, rcx, times_4, 10000));
+    __ divss(xmm1, xmm0);
+    __ divss(xmm1, Operand(rbx, rcx, times_4, 10000));
     __ addps(xmm1, xmm0);
     __ addps(xmm1, Operand(rbx, rcx, times_4, 10000));
     __ subps(xmm1, xmm0);
@@ -402,6 +410,9 @@ TEST(DisasmX64) {
     __ mulps(xmm1, Operand(rbx, rcx, times_4, 10000));
     __ divps(xmm1, xmm0);
     __ divps(xmm1, Operand(rbx, rcx, times_4, 10000));
+
+    __ ucomiss(xmm0, xmm1);
+    __ ucomiss(xmm0, Operand(rbx, rcx, times_4, 10000));
   }
   // SSE 2 instructions
   {
@@ -464,6 +475,74 @@ TEST(DisasmX64) {
     }
   }
 
+  // FMA3 instruction
+  {
+    if (CpuFeatures::IsSupported(FMA3)) {
+      CpuFeatureScope scope(&assm, FMA3);
+      __ vfmadd132sd(xmm0, xmm1, xmm2);
+      __ vfmadd132sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmadd213sd(xmm0, xmm1, xmm2);
+      __ vfmadd213sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmadd231sd(xmm0, xmm1, xmm2);
+      __ vfmadd231sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+
+      __ vfmadd132sd(xmm9, xmm10, xmm11);
+      __ vfmadd132sd(xmm9, xmm10, Operand(r9, r11, times_4, 10000));
+      __ vfmadd213sd(xmm9, xmm10, xmm11);
+      __ vfmadd213sd(xmm9, xmm10, Operand(r9, r11, times_4, 10000));
+      __ vfmadd231sd(xmm9, xmm10, xmm11);
+      __ vfmadd231sd(xmm9, xmm10, Operand(r9, r11, times_4, 10000));
+
+      __ vfmsub132sd(xmm0, xmm1, xmm2);
+      __ vfmsub132sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmsub213sd(xmm0, xmm1, xmm2);
+      __ vfmsub213sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmsub231sd(xmm0, xmm1, xmm2);
+      __ vfmsub231sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+
+      __ vfnmadd132sd(xmm0, xmm1, xmm2);
+      __ vfnmadd132sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmadd213sd(xmm0, xmm1, xmm2);
+      __ vfnmadd213sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmadd231sd(xmm0, xmm1, xmm2);
+      __ vfnmadd231sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+
+      __ vfnmsub132sd(xmm0, xmm1, xmm2);
+      __ vfnmsub132sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmsub213sd(xmm0, xmm1, xmm2);
+      __ vfnmsub213sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmsub231sd(xmm0, xmm1, xmm2);
+      __ vfnmsub231sd(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+
+      __ vfmadd132ss(xmm0, xmm1, xmm2);
+      __ vfmadd132ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmadd213ss(xmm0, xmm1, xmm2);
+      __ vfmadd213ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmadd231ss(xmm0, xmm1, xmm2);
+      __ vfmadd231ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+
+      __ vfmsub132ss(xmm0, xmm1, xmm2);
+      __ vfmsub132ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmsub213ss(xmm0, xmm1, xmm2);
+      __ vfmsub213ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfmsub231ss(xmm0, xmm1, xmm2);
+      __ vfmsub231ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+
+      __ vfnmadd132ss(xmm0, xmm1, xmm2);
+      __ vfnmadd132ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmadd213ss(xmm0, xmm1, xmm2);
+      __ vfnmadd213ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmadd231ss(xmm0, xmm1, xmm2);
+      __ vfnmadd231ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+
+      __ vfnmsub132ss(xmm0, xmm1, xmm2);
+      __ vfnmsub132ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmsub213ss(xmm0, xmm1, xmm2);
+      __ vfnmsub213ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+      __ vfnmsub231ss(xmm0, xmm1, xmm2);
+      __ vfnmsub231ss(xmm0, xmm1, Operand(rbx, rcx, times_4, 10000));
+    }
+  }
   // xchg.
   {
     __ xchgq(rax, rax);