??? 02/24/12 12:55 Read: times |
#186178 - Sales talk Responding to: ???'s previous message |
That white paper contained a bit too much colored sales talk.
An interesting paragraph: "To combat non-SEU failures, a few manufacturers include a software watchdog in their digital cameras. In theory, the watchdog will restart the camera if there is a problem. However, a camera with a software watchdog can’t reset itself once the camera has locked up because the watchdog is part of the digital electronics itself, which is frozen." But few embedded developers uses just a software watchdog. Software watchdogs are mechanisms to verify that multiple threads are in working order, i.e. to supervise the software in the processor - not the processor itself. Another interesting thing is that the white paper did not mention anything about why that external watchdog device would be any better than watchdogs in the cameras. What watchdog supervises the watchdog? Owners of Axis cameras can quickly establish that with 24 cameras, you do not see one hung camera on average every two weeks. If it's because of built-in watchdogs or because they just don't suffer from even close to that number of hangs, I can't say. But no need for external watchdog magic. And thinking about equipment I work with - with tenths of thousands of networked and constantly monitored units out in the field, I do not see many lockups/day. I hardly see any reboots that are not the result of software update processes etc - the devices reports back to the server why they did a reboot. If reporting a power-on, the vehicle the unit is installed in have normally been to service and had the battery disconnected. On one hand, technology have either improved or that white paper is seriously overrating the probability of a SEU. On another hand, released products are normally way better to take care of themselves than that white paper indicates - at least if we ignore toys. |