Speculative computation for coprocessor synthesis

Time critical parts of a hardware-software system suitable for coprocessor implementation often contain nested loops. When loop pipelining is employed for high performance, control dependencies (conditional branches) in any part of the loop can become a dominant limitation to pipeline utilization. Speculative computation based on multiple branch prediction is a systematic approach that overcomes the problem of control dependencies and enables wide parallelism. We present the concept and some practical examples showing a speedup of up to three, with little hardware overhead