I hate configuring things like GPS devices that run super restricted verisons of linux or some other OS. They never seem to deal with error handling very well. For example, here’s the oddball command for fetching ntp.conf and ntp.keys from a ntp server onto a Symmetricom GPS receiver. This is what you want to see, it just works. But in the many failures leading up to this configuration, it was finding problems fetching the files or having the correct access but it was happily coasting right along, overwriting its own configuration with jibberish and rebooting it self only to find the configuration was bollocks.
So, for a single core single processor machine a load average of 1 means that on average there’s a process in the running or runnable state at all times. This means the CPU average is at 100% too. If another process wants to run, it has to wait in the queue before being executed. For a multi-core or multi-processor system, you’re not CPU bound until the load average equals the total number of cores. You can calculate a system’s CPU utilization by dividing the load average by the number of processors you see in /proc/cpuinfo