I attend quite a few conferences, taking any opportunity to speak about embedded software matters. Although such events take place throughout the year, there are two notable periods: Summer [July\/August], when everything goes very quiet, and early Fall [late September and October], when everything goes mad. We are just beginning the “quiet” period.<\/p>\n
But there were a couple of events scheduled before Summer took its grip …<\/p>\n
On Tuesday\/Wednesday last week I was in Munich. There was a series of smallish conferences over two days in a single location. I was invited to present at two of them: the ARM Developers’ Conference and the Embedded Systems Symposium. They were both primarily German language events, so I was pleased to be given the opportunity to participate [in English, of course].<\/p>\n
The first session that I delivered was titled “OS Selection for ARM-based Systems”. I discussed in detail the decision to acquire a commercial operating systems, as opposed to developing one in-house, then looked at the parameters for making a selection. It occurred to me that, not long ago, this session would not have been relevant. For many years, ARM were quite obscure, so there were only a few operating systems available. And, for some time after they became better known, the ARM CPUs covered a limited application space, so OS selection was not very challenging.<\/p>\n
The second session was titled “How to Measure RTOS Performance” and covered a topic that I have addressed at a number of conferences. The key message of this session is advice to prospective OS customers who get totally hung up on data sheet numbers, which are extremely difficult to interpret usefully to make an “apples with apples” comparison of different OSes. Someone asked me about the issues with multicore. Now, there is a big can of worms!<\/p>\n