http://www.open-mpi.org/software/hwloc/v1.0/

./configure && make && ./utils/lstopo

On a dual socket nehalem with HT/SMT on lstopo gives (phys=X is the linux processor number):

$ ./hwloc-1.0/utils/lstopo
Machine (24GB)
 NUMANode #0 (phys=0 12GB) + Socket #0 + L3 #0 (8192KB)
  L2 #0 (256KB) + L1 #0 (32KB) + Core #0
   PU #0 (phys=0)
   PU #1 (phys=8)
  L2 #1 (256KB) + L1 #1 (32KB) + Core #1
   PU #2 (phys=1)
   PU #3 (phys=9)
  L2 #2 (256KB) + L1 #2 (32KB) + Core #2
   PU #4 (phys=2)
   PU #5 (phys=10)
  L2 #3 (256KB) + L1 #3 (32KB) + Core #3
   PU #6 (phys=3)
   PU #7 (phys=11)
 NUMANode #1 (phys=1 12GB) + Socket #1 + L3 #1 (8192KB)
  L2 #4 (256KB) + L1 #4 (32KB) + Core #4
   PU #8 (phys=4)
   PU #9 (phys=12)
  L2 #5 (256KB) + L1 #5 (32KB) + Core #5
   PU #10 (phys=5)
   PU #11 (phys=13)
  L2 #6 (256KB) + L1 #6 (32KB) + Core #6
   PU #12 (phys=6)
   PU #13 (phys=14)
  L2 #7 (256KB) + L1 #7 (32KB) + Core #7
   PU #14 (phys=7)
   PU #15 (phys=15)