To improve the power consumption of parallel applications at the runtime, modern processors provide frequency scaling and power limiting capabilities. In this work, a runtime strategy is proposed to distribute a given...To improve the power consumption of parallel applications at the runtime, modern processors provide frequency scaling and power limiting capabilities. In this work, a runtime strategy is proposed to distribute a given power allocation among the cluster nodes assigned to the application while balancing their performance change. The strategy operates in a timeslice-based manner to estimate the current application performance and power usage per node followed by power redistribution across the nodes. Experiments, performed on four nodes (112 cores) of a modern computing platform interconnected with Infiniband showed that even a significant power budget reduction of 20% may result in a performance degradation of as low as 1% under the proposed strategy compared with the execution in the unlimited power case.展开更多
Recent results in the development of diode-driven high energy, high repetition rate, picosecond lasers, including the demonstration of a cryogenic Yb:YAG active mirror amplifier that produces 1.5 J pulses at 500 Hz re...Recent results in the development of diode-driven high energy, high repetition rate, picosecond lasers, including the demonstration of a cryogenic Yb:YAG active mirror amplifier that produces 1.5 J pulses at 500 Hz repetition rate(0.75 kW average power) are reviewed. These pulses are compressed resulting in the generation of ~5 ps duration,1 J pulses with 0.5 kW average power. A full characterization of this high power cryogenic amplifier, including atwavelength interferometry of the active region under >1 kW average power pump conditions, is presented. An initial demonstration of operation at 1 kW average power(1 J, 1 k Hz) is reported.展开更多
文摘To improve the power consumption of parallel applications at the runtime, modern processors provide frequency scaling and power limiting capabilities. In this work, a runtime strategy is proposed to distribute a given power allocation among the cluster nodes assigned to the application while balancing their performance change. The strategy operates in a timeslice-based manner to estimate the current application performance and power usage per node followed by power redistribution across the nodes. Experiments, performed on four nodes (112 cores) of a modern computing platform interconnected with Infiniband showed that even a significant power budget reduction of 20% may result in a performance degradation of as low as 1% under the proposed strategy compared with the execution in the unlimited power case.
基金supported by the U.S. Department of Energy Accelerator Stewardship programme, Office of High Energy Physics, Office of Science under award DE-SC0016136support by the U.S. Department of Energy, Office of Science SBIR programme under award DE-SC0011375
文摘Recent results in the development of diode-driven high energy, high repetition rate, picosecond lasers, including the demonstration of a cryogenic Yb:YAG active mirror amplifier that produces 1.5 J pulses at 500 Hz repetition rate(0.75 kW average power) are reviewed. These pulses are compressed resulting in the generation of ~5 ps duration,1 J pulses with 0.5 kW average power. A full characterization of this high power cryogenic amplifier, including atwavelength interferometry of the active region under >1 kW average power pump conditions, is presented. An initial demonstration of operation at 1 kW average power(1 J, 1 k Hz) is reported.