Non-contact mechanical and chemical analysis of single living cells by microspectroscopic techniques

Innovative label-free microspectroscopy,which can simultaneously collect Brillouin and Raman signals,is used to characterize the viscoelastic properties and chemical composition of living cells with sub-micrometric resolution.The unprecedented statistical accuracy of the data combined with the high-frequency resolution and the high contrast of the recently built experimental setup permits the study of single living cells immersed in their buffer solution by contactless measurements.The Brillouin signal is deconvoluted in the buffer and the cell components,thereby revealing the mechanical heterogeneity inside the cell.In particular,a 20%increase is observed in the elastic modulus passing from the plasmatic membrane to the nucleus as distinguished by comparison with the Raman spectroscopic marker.Brillouin line shape analysis is even more relevant for the comparison of cells under physiological and pathological conditions.Following oncogene expression,cells show an overall reduction in the elastic modulus(15%)and apparent viscosity(50%).In a proof-of-principle experiment,the ability of this spectroscopic technique to characterize subcellular compartments and distinguish cell status was successfully tested.The results strongly support the future application of this technique for fundamental issues in the biomedical field.

Tetris:A Heuristic Static Memory Management Framework for Uniform Memory Multicore Neural Network Accelerators

Uniform memory multicore neural network accelerators(UNNAs)furnish huge computing power to emerging neural network applications.Meanwhile,with neural network architectures going deeper and wider,the limited memory capacity has become a constraint to deploy models on UNNA platforms.Therefore how to efficiently manage memory space and how to reduce workload footprints are urgently significant.In this paper,we propose Tetris:a heuristic static memory management framework for UNNA platforms.Tetris reconstructs execution flows and synchronization relationships among cores to analyze each tensor’s liveness interval.Then the memory management problem is converted to a sequence permutation problem.Tetris uses a genetic algorithm to explore the permutation space to optimize the memory management strategy and reduce memory footprints.We evaluate several typical neural networks and the experimental results demonstrate that Tetris outperforms the state-of-the-art memory allocation methods,and achieves an average memory reduction ratio of 91.9%and 87.9%for a quad-core and a 16-core Cambricon-X platform,respectively.