I like your approach and there are some good insights in the other posts.
Let me suggest you back the decision up a step. Business objectives drive business requirements. What are the business requirements your tool (the workstation with whatever OS and applications) set will support? If you can get a well defined set of business requirements, you can use these to drive technical objectives.
The technical objectives are the step before technical requirements, which is where you are currently at. The technical objective is where such issues as local versus server processing, level of security, responsiveness, central/remote/mobile work locations, compatability, interoperability, support, and the like are considered. Also considered at this point are user profiles - a gouping of two to five unique types of users based on function and perhaps a three level skill description within each profile that helps define how user friendly any solution must be.
The technical objectives will drive the technical requirements. Perhaps unix or linux will meet one or more user profiles while MS may be necessary for others. Perhaps all can be supported with a thin client with minimal OS - in effect a terminal station.
As for the hw aspect - do not assume that any manufacturer's platform with a common CPU, memory, OS, and clock speed will perform the same. The physical architecture can directly impact performance. Consider as well the environment. A device built for rugged environments may be preferred even if its performance is less than similar devices from other manufacturers.
Of course, the time to do things right is often missing. In that case, keep this thought from Robert McLeod Gray in mind - "Whatever you do, you'll regret it."
All the best to you.