This test program solves the benchmark problem a 8x8x8 mesh
and is intended for examining the speed of the code, including

	- optimization of the code
	- different compiler flags
	- different compilers
	- different computers

To run the benchmark, type

	time ./ns-benchmark

The test run should end with something like

	      UP residual: 2.9976e-05
        Assemble vector b
        Assemble matrix A
        Assemble vector b
      Non linear residual: 2.9976e-05
  Write data to file
  Solution has been computed to endtime T = 0.108253
  Closing DOLFIN.

Version 0.2.10
--------------

About the same as version 0.2.9


System:   Linux -.-.chalmers.se 2.4.19-pre4-ac4 #1 tor apr 4 14:57:21 DST 2002 i686 unknown
CPU:      Amd Athlon 1400MHz
Compiler: gcc 2.96 20000731 (Red Hat Linux 7.1 2.96-98)
Flags:    -O2 -march=athlon
Time:     13.860u 0.010s 0:14.78 93.8%    0+0k 0+0io 214pf+0w
Date:     2002-04-17 (Thomas Svedberg)

System:   Linux bamse 2.4.18pre2 #2 SMP ons jan 9 09:37:33 MET 2002 i686 unknown
CPU:      Intel Pentium III (Coppermine) 1000MHz
Compiler: gcc 2.95.4
Flags:    -O2
Time:     23.640u 0.070s 0:24.76 95.7%    0+0k 0+0io 215pf+0w
Date:     2002-03-05

Version 0.2.9
-------------

9 times faster!

System:   Linux bamse 2.4.18pre2 #2 SMP ons jan 9 09:37:33 MET 2002 i686 unknown
CPU:      Intel Pentium III (Coppermine) 1000MHz
Compiler: gcc 2.95.4
Flags:    -O2
Time:     22.810u 0.080s 0:27.15 84.3%    0+0k 0+0io 217pf+0w
Date:     2002-03-05

Version 0.2.8
-------------

System:   Linux bamse 2.4.18pre2 #2 SMP ons jan 9 09:37:33 MET 2002 i686 unknown
CPU:      Intel Pentium III (Coppermine) 1000MHz
Compiler: gcc 2.95.4
Flags:    -O2
Time:     215.360u 1.230s 5:18.85 67.9%   0+0k 0+0io 219pf+0w
Date:     2002-03-05
