This paper describes algorithm to parallelize loop with loop-carried dependency between iterations in the specific case. This algorithm can be used for hardware compilation and allows increase performance of synthesiz...
详细信息
ISBN:
(纸本)9781424413478
This paper describes algorithm to parallelize loop with loop-carried dependency between iterations in the specific case. This algorithm can be used for hardware compilation and allows increase performance of synthesized devices. The paper represents case and condition when the suggested algorithm should be applied and introduces criterion for algorithm parameters chosen to keep up optimal balance between performance and hardware overhead The hardware realization of ZP-coder on the base of investigated algorithm and its comparison with analogical devices are also represented.
We empirically study power consumptions of two linear transformations and lossless data compression algorithms on a TI MSP432 microcontroller. We developed a custom circuit using a TI INA219 breakout board and an Ardu...
详细信息
ISBN:
(纸本)9781467392686
We empirically study power consumptions of two linear transformations and lossless data compression algorithms on a TI MSP432 microcontroller. We developed a custom circuit using a TI INA219 breakout board and an Arduino Uno R3 board and programmed it to record power usage automatically. We validated the accuracy of our measurements with that obtained from TI EnergyTrace software. Our experimental results show that arithmetic coder consumed the least power and required the lowest time for a compression gain of 21.89%. However, when data was preprocessed with the move-to-front transformation and compressed with an arithmetic coder, computation time and energy usage increased only by 21%, while compression gain nearly doubled to 38.39%.
In the paper optimization of lossless image coders based on Adaptive Neural Networks (AdNN) is addressed. Firstly, a detailed analysis of the influence of AdNN parameters on coder performance (average bitrate, time co...
详细信息
ISBN:
(纸本)9781457720482
In the paper optimization of lossless image coders based on Adaptive Neural Networks (AdNN) is addressed. Firstly, a detailed analysis of the influence of AdNN parameters on coder performance (average bitrate, time complexity) is done. Secondly, an improved technique denoted AdNN(+) is proposed. Its main features are introduction of contexts, variable training window size, and post-processing by NLMS algorithm. Experiments show that indeed, the new method is better than others based on neural networks, and that it can even compete with the best existing image lossless coding algorithms.
Tato práce se zaobírá novodobým kompresním řetězcem JPEG 2000. První část je rozborem technik používaných v základní části standardu a jejich možností ...
详细信息
Tato práce se zaobírá novodobým kompresním řetězcem JPEG 2000. První část je rozborem technik používaných v základní části standardu a jejich možností v kompresním formátu. Druhá část se věnuje implementaci kompresního řetězce z pohledu předzpracování, až k přivedení výstupu do vlastního formátu. Závěrem práce je provedeno srovnání s aktuálními implementacemi z paměťového a výkonnostního hlediska.
暂无评论