Specifications for GIS desktop hardware

My main suggestions would be to:

1. Save a little on the CPU by getting a slightly less powerful but much less expensive model and use the savings to get a solid state drive (SSD) to put your operating system and applications on. This will make boot and app startup times very short, as well as effectively eliminate those common little stutters caused by random disk accesses during normal app usage.

   You can also use it to speed up certain types of IO-intensive GIS operations like raster conversion, although space will be at a premium compared to a conventional hard disk drive (HDD). See Looking for quantitative evidence supporting SSD adoption for more details.

   Suggestions:

   • CPU: Intel Core i5-2400 Sandy Bridge 3.1GHz (3.4GHz Turbo Boost) LGA 1155 95W Quad-Core - $189.99 USD

   • SSD: Crucial M4 CT128M4SSD2CCA 2.5" 128GB SATA III MLC Internal Solid State Drive (SSD) with Transfer Kit - $147.99 USD

2. Instead of purchasing a case with power supply unit (PSU) included, which are usually bottom of the barrel and inefficient/unreliable, purchase a case without PSU and a separate name brand PSU with at least an 80 PLUS Bronze rating, which indicates its energy efficiency and quality standards.

   Computer hardware enthusiasts will tell you the PSU is the most important part of your system as a large portion of reliability issues such as random reboots, black screens, BSODs, etc. are often remedied by replacing the PSU. So don't skimp!

   Suggestion: Antec EarthWatts Green EA-430D Green 430W - $54.99 USD

   Computer case choices are largely subjective and its likely most any < $50 case will do fine so I'll leave that up to you. Just remember you don't need one with a PSU included if you get a separate one.

3. Lastly will you be doing complex realtime 3D visualizations, such as with ArcScene, ArcGlobe, or Google Earth?

   While the on-die Intel HD Graphics 2000/3000 chipset on the generation of CPU you specified does have a basic level of 3D graphics compatibility (it supports OpenGL 3.0, which ArcGIS 10 requires at least version 2.0), for more complex scenes it will likely perform inadequately and may lack some advanced features.

   In the most demanding cases you would want a dedicated workstation-level graphics card such as AMD's FirePro or NVIDIA's Quadro series. A consumer-level graphics card may also be sufficient if you don't need the advanced features of a workstation card.

   Suggestion: Stick with your CPU's on-die Intel HD Graphics unless you run into performance or compatibility problems. If you do plan on getting a powerful video card though, you'll want to check its power supply requirements and get a higher wattage PSU to match.

   See also: Optimal video cards for GIS programs

I have always based anything I have built on power and performance. I would never recommend an onboard Graphics capability if you are doing serious GIS work. Remember, also, if this is your development machine, it's likely to be running something like Oracle/SQL Server* and Java, or VS2010 and these are very, very hungry beasts. ArcPy itself likes to chew memory up too.

Essentially, GIS is a numbers game, with most of it based on maths, so raw computational power is a must for me. Then get as much memory in there as you can afford, and do go SSD.

There's no getting away from the fact a very good GIS desktop will cost a lot of money if you want to do it right. I have a stupidly powerful desktop and laptop, the best specc'd money could buy at the time, and if I am developing an Oracle based, java AGS 10 solution, with JS front end, with arcpy geoprceissing scripts, you can't do much else.

*other types of RDBMS are available