Many of Speak Freely's compression and encryption modes require substantial amounts of computation. In order to transmit and receive audio in real time, without regular pauses, your computer must be fast enough to compress and/or encrypt audio at least as fast as it arrives from the microphone or across the network. The performance of machines running Windows spans an enormous range, and even machines with identical processor chips may deliver dramatically different results due to variations in cache size, memory architecture, and the presence or absence of a floating point unit or coprocessor. To give an idea how well suited various compression and encryption modes are to your own computer, Speak Freely includes a built-in performance measurement (or benchmark) facility, which evaluates how quickly your computer executes the various compression and encryption algorithms compared to the requirement for real time audio.
To run the benchmark, display the Help/Performance Benchmark dialogue. Before the benchmark has been run, all the fields will be question marks. To start the benchmark, press the Run button. Speak Freely runs each of the algorithms, both in the mode used when transmitting and that used for reception, for three seconds and fills in items in the table with a percentage giving the speed at which the algorithm executed compared to the real time audio rate. A result of less than 100% indicates your computer cannot perform that algorithm quickly enough to keep up with audio; any mode showing less than 100% isn't suitable for your machine. A measurement substantially higher than 100% indicates the mode should be usable. Note, however, that for successful two-way communication the modes you select must not only be within the capacity of your own computer but also that of the person you’re talking to; if one of you has a much faster machine than the other, you'll have to negotiate settings suitable for both.
The benchmark takes about a minute to run; progress is indicated by the filling in of fields in the table every three seconds. You can stop the benchmark at any time by pressing the Stop button. Since Windows is, more or less, a multitasking system, other concurrently-running applications may reduce the performance figures reported by the benchmark by "stealing" compute cycles from Speak Freely. To get an idea of how Speak Freely will actually behave, it's best to run the benchmark with the same background applications you'll be running when you actually communicate using Speak Freely. (Since Windows is not noted for its real-time response, for optimum performance it's best to run Speak Freely as the only active application.)
When you combine various compression and encryption modes, the performance required is often not simply the sum of the modes selected. Compression, for example, reduces the number of bytes the encryption algorithms must process, and in some cases adding the overhead of compression may enable you to use an encryption mode your computer isn't fast enough to apply to uncompressed data. The performance benchmark can only give you general guidelines; if you have a machine like the 400 MHz Pentium II on which the sample benchmark at the start of this section was run, any combination of modes will work fine. If you want to use a combination of modes that are above but uncomfortably close to 100%, it's best to experiment with local loopback or an echo server determine if your machine is fast enough to squeak by.