saghir@eecg.toronto.edu (Mazen Saghir) (04/16/91)
I was reading about DSP architectures and how benchmarks are important when comparing different architectures when I got to wondering about the nature of those benchmarks. How good are benchmarks like 1024x1024 FFTs, FIRs, or IIRs when comparing different DSP architectures. Wouldn't an application-specific benchmark be more suitable when comparing DSPs, or would "general" benchmarks suffice?