Maximilien Breughe

TL;DR: This paper presents the benchmarking method for evaluating ML inference systems, MLPerf Inference, and prescribes a set of rules and best practices to ensure comparability across systems with wildly differing architectures.

TL;DR: MLPerf Inference as mentioned in this paper is a benchmarking method for evaluating ML inference systems with different architectures and architectures. And it is based on the first call for submissions garnered more than 600 reproducible inference-performance measurements from 14 organizations, representing over 30 systems that showcase a wide range of capabilities.

TL;DR: A mechanistic analytical performance model is proposed that is built from understanding the internal mechanisms of the processor, and is shown to be accurate with an average performance prediction error of 3.2% compared to detailed cycle-accurate simulation using gem5.

TL;DR: The proposed mechanistic performance model for superscalar in-order processors models the impact of non-unit instruction execution latencies, inter-instruction dependencies, cache/TLB misses and branch mispredictions, and achieves an average performance prediction error of 2.5% compared to detailed cycle-accurate simulation.

TL;DR: A number of methods for selecting representative inputs are proposed and evaluated and it is shown that the optimum design point can be found with as few as three inputs.