So You Think Performance is Cracked Using Standard Floating Point DSPs?
There is no doubt that the performance targets of digital audio systems are becoming easier and more economic to achieve as technology advances. This is particularly true in the areas of DSP devices and ASICs (Application Specific Integrated Circuits). However, it is not enough simply to have large amounts of processing power and dynamic range available, for system performance is also still dependent upon the correct choice of algorithms and system architectures. This is particularly true in the design of large Digital Audio Mixing Consoles where the overall system performance is not determined solely by the performance of a single EQ or even a signal path, but by the total number of processes performed.: This paper looks at the performance of four filter structures as implemented using IEEE floating point arithmetic, and also gives examples of processes where the ASIC implementation of dedicated hardware can be very effective.
This paper costs $33 for non-members, $5 for AES members and is free for E-Library subscribers.