Algorithms for hardware allocation in data path synthesis

Abstract

Abstract-The most creative step in synthesizing data paths executing software descriptions is the hardware allocation process. New algorithms for the simultaneous costlresource constrained allocation of registers, arithmetic units, and interconnect in a data path have been developed. The entire allocation process can be formulated as a twodimensional placement problem of microinstructions in space and time. This formulation readily lends itself to the use of a variety of heuristics for solving the allocation problem. We present simulated-annealingbased algorithms which provide excellent solutions to this formulation of the allocation problem. These algorithms operate under a variety of user-specifable constraints on hardware resources and costs. They also incorporate conditional resource sharing and simultaneously address all aspects of the allocation problem, namely register, arithmetic unit and interconnect allocation, while effectively exploring the existing tradeoffs in the design space.

Accelerating SPICE Model-Evaluation using FPGAs

“…number of floating-point operators) and partitioning the nodes based on locality using a high-quality partitioner MLPart [23]. We then provision the number of hardware operators of each type according to instruction frequency and allocate them to partitions using need-proportional distribution [24], [25]. Each partition can process operations of a single operator type.…”

Behavioral synthesis of fault secure controller/datapaths using aliasing probability analysis

“…A related methodology, based on simulated annealing, has also been published [18]. The synthesis methodology is outlined in this paper and the parts that deal with incorporating fault-security are detailed.…”

“…Any equation of form (18) in n variables, with each having a range of fHY IY F F F Y P x À Ig, which is not an identity, will have at most p à P xnÀI solutions. To see this, we can simply fix the values of all but one of the variables and solve for the remaining one variable.…”

Area efficient DSP datapath synthesis

“…[5], have developed an integer programming model which can perform optimisation on the whole datapath synthesis problem. Devadas and Newton [2] use simulated annealing to perform global datapath optimisation. However, neither of these systems are designed to take physical domain effects into account during synthesis.…”