I read the FAQ, but it's still not working. Here's my trivial program:

Code: Select all

#include<Eigen/Core>
using namespace Eigen;
void foo(VectorXd foo, VectorXd bar)
{
   double ll=0;
  EIGEN_ASM_COMMENT("begin");
  ll += foo.transpose() * bar;
  EIGEN_ASM_COMMENT("end");
}


Here's how I compile it:

Code: Select all

g++  -I"/opt/R/x86_64-pc-linux-gnu-library/3.3/Rcpp/include" -I"/opt/R/x86_64-pc-linux-gnu-library/3.3/RcppEigen/include" foo.cpp -O2 -c -S -o myprogram.s


Here's part of the asm output:

Code: Select all

   .cfi_startproc
#APP
# 8 "foo.cpp" 1
   #begin
# 0 "" 2
#NO_APP
   movq   8(%rsi), %rax
   cmpq   %rax, 8(%rdi)
   jne   .L5
#APP
# 10 "foo.cpp" 1
   #end
# 0 "" 2
#NO_APP
   ret
.L5:
   pushq   %rax
   .cfi_def_cfa_offset 16
   movl   $_ZZN5Eigen13CwiseBinaryOpINS_8internal17scalar_product_opIddEEKNS_9TransposeIKNS4_INS_6MatrixIdLin1ELi1ELi0ELin1ELi1EEEEEEEKS6_EC4ERSA_RSB_RKS3_E19__PRETTY_FUNCTION__, %ecx
   movl   $132, %edx
   movl   $.LC0, %esi
   movl   $.LC1, %edi
   call   __assert_fail
   .cfi_endproc


If I use the trivial code from the FAQ then it seems to work:

Code: Select all

#include<Eigen/Core>
using namespace Eigen;
void foo(Vector4f& u, Vector4f& v, Vector4f& w)
{
  EIGEN_ASM_COMMENT("begin");
  u = v + 3*w;
  EIGEN_ASM_COMMENT("end");
}


Compiles to:

Code: Select all

   movaps   .LC0(%rip), %xmm0
   mulps   (%rdx), %xmm0
   addps   (%rsi), %xmm0
   movaps   %xmm0, (%rdi)


This looks good, right? What am I doing wrong with my dot product?

Thanks.

it has been optimized away by the compiler because "ll" is not used. To get something meaningful, add "return ll;"