GPGPU Performance Estimation with Core and Memory Frequency Scaling

Graphics processing units (GPUs) support dynamic voltage and frequency scaling to balance computational performance and energy consumption. However, simple and accurate performance estimation for a given GPU kernel under different frequency settings is still lacking for real hardware, which is important to decide the best frequency configuration for energy saving. We reveal a fine-grained analytical model to estimate the execution time of GPU kernels with both core and memory frequency scaling. Over a 2 x range of both core and memory frequencies among 20 GPU kernels, our model achieves accurate results (4.83 % error on average) with real hardware. Compared to the cycle-level simulators, our model only needs simple micro-benchmarks to extract a set of hardware parameters and kernel performance counters to produce such high accuracy without kernel source analysis.