Occasionally a heavy Mauve addict asks what sort of compute hardware would be ideal for computing Mauve alignments.  The answer depends in part on the length and number of sequences being aligned.  When aligning a large number of sequences, Mauve is both compute intensive and memory intensive.  When aligning very large genomes such as mammalian genomes, Mauve requires multiple independent hard drives to perform external sort operations.

PC hardware 

 I will use Dell hardware as an example, although I am in no way endorsing purchase of Dell hardware and other manufacturers likely offer better compute hardware.   Dell offers the Dimension series of workstations that can be configured with large memory and multiple CPUs.  One setup that would make CPU and memory hungry programs like Mauve happy might be a Dell p490, configured with two quad-core Xeon 2.4Ghz CPUs and 4GB or more of RAM.  Such a setup offers 8 effective CPUs for less than $3000 as of November 2007.  However, 8 cpu cores does not mean that Mauve will run 8 times faster.  Some parts of the algorithm are not parallel, and parts that are parallel tend to be memory I/O intensive and will probably have to fight for the shared memory bus resource.

 Mac hardware

 Since its transition to Intel CPUs, Apple hardware is very similar to Dell hardware.  Apple offers Mac Pro machines which are comparable to the eight-core machine described in the previous section.

Other hardware 

 Companies such as IBM, SGI, Sun, and others manufacture large memory model, multi-CPU machines.  Such hardware is typically much more expensive than a PC, but usually has the advantage of high-throughput memory I/O from each CPU.  The parallel portions of the aligner stand a much better chance of giving a linear speedup with added CPUs on this type of hardware.  Although pre-compiled binaries are not available for these architectures, the aligner source code can be compiled using gcc.  Other compilers have not been tested.