To all,
After about eight years, I finally had to upgrade my machine. It was testing my patience due to its slowness. The old PC was a dual Xeon 2.66 GHz 32-bit processor. Each processor is single core, but it has HT for a total of 2 threads (total for the system is 4 threads). The new machine is a dual Xeon 5650 2.66 GHz processor with 6 cores each for a total of 24 threads. The old system was Windows XP 32-bit; whereas, the new system is Windows 7 64-bit. Finally, the old system had 4 GB of ram, and the new system has 24 GB. All the benchmark tests were run with the console open.
Old dual processor Xeon 2.66 GHz:
Benchmark_M74: 78.09 seconds
Benchmark_M74_parallel: 35.54 seconds
New dual processor Xeon x5650 2.66 GHz:
Benchmark_M74: 83.42
Benchmark_M74_parallel: 12.13
I'm a little surprised at how slow the processors are when executing with non-parallel code. I suppose the "poor" performance of the new chip (i.e. slower than the older chip) is due to the 64-bit overhead and larger memory pool??? It definitely goes to show that cpu speed is everything.
I did run a comparison on StarAlignment and Integration. It took the old system 8203 seconds to align thirty-three 4096x4096 images (Kodak 16803 chip). It took the new system 1184 seconds. Integrating the images with noise evaluation and Winsorized sigma clipping. The old system took 2132 seconds while the new system took 1041 seconds.
I'm very pleased with the improvement I'm seeing, but as stated earlier, cpu speed is everything if you want the fastest computations. In the future, I may experiment with overclocking.
Wade