In real-time systems, energy consumption is one of the most critical challenges. Dynamic voltage and frequency scaling (DVFS) algorithms have been widely applied to balance the trade-off between performance and power consumption. We first review some classic DVFS methods, e.g., LA-EDF and CCEDF, and then a hybrid DVFS algorithm (Soft-LA2) that we proposed recently. Soft-LA2 needs domain knowledge to set the trade-off parameter manually to achieve energy saving compared with classic methods. Motivated by recent deep learning technologies applied in DVFS, in this paper, we propose a new framework for Soft-LA2. Our method automatically optimizes parameters for Soft-LA2, which leads to more power saving compared with random parameter setting in Soft-LA2. Simulation results show that under certain neural network architecture setting, our method can find the best parameters automatically to achieve power saving performance. Furthermore, with test data, our proposed method saves 4.9% and 9.6% power consumption compared with relevant learning method DQL-EES and random setting Soft-LA2, respectively.
8 December 2025 in Concurrency and Computation: Practice and Experience
Endpoint structure–activity relationships (SARs) for irreversible inhibitors can integrate reversible binding, covalent chemistry, conformational behavior, and cellular factors, making it difficult to assign a potency change to a single molecular event. We reanalyzed a previously reported series of irreversible inhibitors of wild-type (WT) HER2 and HER2 A775_G776insYVMA (HER2YVMA) as six matched analogue pairs. Three complementary analyses were kept physically distinct. Pre-reaction molecular dynamics (MD) trajectories were used to quantify the fraction of frames satisfying a predefined near-attack-conformation (NAC)-like distance–angle criterion; continuous structural descriptors calculated from OpenComplex2 (OC2)-generated covalently linked HER2–inhibitor structures were used to characterize structural differences after covalent linkage; and covalent-adduct MD trajectories were used to characterize post-bond-formation dynamics. The Michael addition step itself was neither simulated nor measured. Cyclization of compound 3 to compound 9c increased the NAC-like frame fraction in HER2WT from 0.0217% to 0.358% but decreased it in HER2YVMA from 0.745% to 0.333%, with the overall direction preserved across all six sensitivity definitions. In HER2YVMA, the modification distinguishing compounds 4a and 5a increased the NAC-like fraction from 2.620% to 6.244% despite a loss of cellular potency. The analogous tail modification increased this fraction in the 8c/10e series, whereas the HER2YVMA 9c/11f response was starting conformation sensitive. The distribution of OC2-generated covalently linked structures across the predefined structural coordinate and the final-adduct MD readouts likewise varied with HER2 context and scaffold, although some differential effects were sensitive to the coordinate reference. These analyses complement experimental SARs but do not provide a measure of reaction kinetics or thermodynamic populations.
Endpoint structure–activity relationships (SAR) for irreversible inhibitors can integrate reversible binding, covalent chemistry, conformational behavior, and cellular factors, making it difficult to assign a potency change to a single molecular event. We reanalyzed a previously reported series of irreversible inhibitors of wild-type (WT) HER2 and HER2 A775_G776insYVMA (HER2YVMA) as six matched analogue pairs. Three complementary analyses were kept physically distinct. Pre-reaction molecular dynamics (MD) trajectories were used to quantify the fraction of frames satisfying a predefined near-attack-conformation (NAC)-like distance–angle criterion; continuous structural descriptors calculated from OC2-generated covalently linked HER2–inhibitor structures were used to characterize structural differences after covalent linkage; and covalent-adduct MD trajectories were used to characterize post-bond-formation dynamics. The Michael-addition step itself was neither simulated nor measured. Cyclization of compound 3 to compound 9c increased the NAC-like frame fraction in HER2WT from 0.0217% to 0.358% but decreased it in HER2YVMA from 0.745% to 0.333%, with the direction preserved across all six sensitivity definitions. In HER2YVMA, the modification distinguishing compounds 4a and 5a increased the NAC-like fraction from 2.620% to 6.244% despite a loss of cellular potency. The analogous tail modification increased this fraction in the 8c/10e series but reduced it in the 9c/11f series. The distribution of OC2-generated covalently linked structures across the predefined structural coordinate and the final-adduct MD readouts likewise varied with HER2 context and scaffold, although some differential effects were sensitive to the coordinate reference. No single structural readout tracked endpoint potency across all six pairs. These analyses complement experimental SAR but do not provide a measure of reaction kinetics or thermodynamic populations.