>I'll dig out !SICK and do some figures for you. As I said earlier, RedSquig is faster on my Athlon 2100 than a real RiscPC, so your "lot" is already here.
A RiscPC, with a StrongARM ? I think not. Besides measure speed with !Sick on an emulated computer may not yield accurate results (as it depends on how Sick times things and how accurate the "clock" is when emulated). If for example if Windows is running background tasks will the "clock" count that in or not ? You'd be better off with a stopwatch timing REAL tasks doing something.
>As for USB api... uh, well how could they a) second-guess that folks are suddenly going to produce their own USB APIs without talking to ROL and b) produce any sane API based on no knowledge of how USB hardware would be implemented on a RiscPC. I think you're being just a touch silly now.
The implementation is independant of the API. Which is why we have essentially similar USB hardware but completely different API's.
ROL should simply have laid down a "standard" and put it up to others to implement USB in whatever way they like but ensure the API followed that as described by ROL. As to information USB is well known with it's own Special Interest Group and commercial implementers freely providing information (e.g., Microsoft and various Linux implementers).
Alternatively ROL could have "invited" interested parties to discuss an API that could be arrived by agreement (much as Microsoft did when setting up CDMRW/Mt Rainier). Either way ROL had to be proactive and they were not.
If all else failed ROL could "invite" proposed standards from companies and then "annoint" one as the approved standard. [They already control the admin of filetypes and the like, this is the same authoritive control but in a different area]