format PE console
include '%fasminc%\win32a.inc'

  ; preparation
  mov    eax, v1
  mov    ebx, v2
  xor    ecx, ecx
  xorps  xmm0, xmm0

  ; did not work (if PAE is OFF - works fine)
  ; comment next 2 line to check working example
  movups xmm0, [eax+ecx*4]
  mulps  xmm0, [ebx+ecx*4]

  ; work fine
  movups xmm0, [eax+ecx*4]
  movups xmm1, [ebx+ecx*4]
  mulps  xmm0, xmm1

  invoke  ExitProcess

v1  dd  0,0,0,0
v2  dd  0,0,0,0

data import
 library   kernel32,'KERNEL32.DLL'
 import    kernel32,\
           ExitProcess,'ExitProcess'
end data
    

align 16

v1  dd  0,0,0,0 
v2  dd  0,0,0,0