Search

The CUDA core of NVIDIA GPUs had been one of the most efficient computation units for parallel computing. However, recent rapid developments in deep neural networks demand an even higher level of computational performance. To meet this requirement, NVIDIA has introduced the Tensor core in recent generations. However, their impressive enhancements in computational performance have newly brought high pressure on the memory hierarchy. In this paper, first we identify the required memory bandwidth in the memory hierarchy as the computational performance increases in actual GPU hardware. Through a comparison of the CUDA core and the Tensor core in V100, we find that the tremendous performance increase of the Tensor core requires much higher memory bandwidth than that in the CUDA core. Moreover, we thoroughly investigate memory bandwidth requirement over Tensor core generations of V100, RTX TITAN, and A100. Lastly, we analyze a hypothetical next-generation Tensor core introduced by NVIDIA through a GPU simulation, through which we propose an inter-warp multicasting microarchitecture that reduces redundant shared memory (SMEM) traffic during the GEMM process. Our evaluation shows that inter-warp multicasting reduces the SMEM bandwidth pressure by 33% and improves the performance by 19% on average in all layers of ResNet-152 and BERT-Large. [ABSTRACT FROM AUTHOR]

Background: Muscle flap is a valuable option in soft tissue reconstruction. Denervated skeletal muscle is known to undergo degeneration. However, information regarding histological and genetic changes in muscle free flap without reinnervation over long-term follow-up remains unclear., Methods: We collected flap muscles obtained during secondary exploration surgery after more than 15 years of previous muscle free flap without reinnervation. Compared with normal muscle and fat, histomorphometric and gene expression analysis of flap muscle were performed., Results: During the study period, we collected 5 samples of previous muscle free flap. The mean ± SD postoperative duration after free flap was 18.6 ± 4.0 years. All flap muscles were replaced with adipose tissue based on gross and histological findings. In flap muscle, the expression of gene related to muscle-specific MYH2 gene was downregulated, whereas the expression of genes related to adipose, fibroadipogenic progenitor, and blood vessel was upregulated compared with that of normal muscle. Vascular density and pattern were also similar to those in normal fat., Conclusions: We demonstrated that muscle free flap without reinnervation eventually converts into adipose tissue regardless of spontaneous reinnervation during muscle regeneration. The long-term findings of the present study will be valuable for muscle flap selection and prognosis.