Forester, not a geopspatial or IT professional, currently using QGIS and SAGA for generally county level geoprocessing of multipspectral imagery and DEM's (cluster analysis, zonal statistics, stream channel modeling). Currently exceeding practical processing limits for laptop/windows configuration using HP EliteBook w/4GB Ram, Intel Core i5, 32 bit Win7 OS. Need advice on logical next step in hardware/OS/software. Future ambitions include deployment of PostGIS and derivation of canopy height models from stereo matched imagery. Thinking low end workstation and Ubuntu OS. Guidance/advice/considerations and recommendations much appreciated. I'm a one man shop trying to operate a bit beyond my capabilities. I do not speak fluent IT or GIS, need a balance of horsepower, idiot proof, moderate learning curve and intermediate support. Cost is important but functionality, flexibility and capability are primary. Thanks.
On the hardware side, I'd recommend staying with something in the i5 line, as that gets you the best bang for your buck, and any extra money might be better spent on improving other hardware.
Your best bet to get the most out of your hardware is to look at what components are limiting you now. When you're running an process and look at task manager, what is slowing you down? Are you running out of ram? Are you 100% on all cores, or only just one or two? Are you limited on disk i/o?
Where the bottleneck is should determine where you should focus your efforts. I would recommend upgrading to an hybrid SSD/HDD drive in your new machine, because for a lot of GIS work they are very effective, due to the way data is used (the drive caches data to the SSD that is used frequently, so if you're working on the same project data all week long, it'll cache the data early on, and then you'll have some of the speed of a SSD without the cost of a full SSD drive). On the other hand, if you've got the money, just grab a large SSD drive and run everything off of there.
Hardware GPU acceleration can be few and far between, but if you have aspirations of running CUDA enabled models it could be a good investment. I would recommend some kind of discrete graphics card to take that load off of the CPU and RAM, but don't throw money at it thinking that your rendering times are going to be amazing.