We present an evaluation of 32-bit POSIT arithmetic through its implementation as accelerators on FPGAs and GPUs. POSIT, a floating-point number format, adaptively changes the size of its fractional part. We developed hardware designs for FPGAs and software for GPUs to accelerate linear algebra operations using Posit(32,2) arithmetic. Our FPGA- and GPU-based accelerators in Posit(32,2) arithmetic significantly accelerated the Cholesky and LU decomposition algorithms for dense matrices. In terms of numerical accuracy, Posit(32,2) arithmetic is approximately 0.5 - 1.0 digits more accurate than the standard 32-bit format, especially when the norm of the elements of the input matrix is close to 1. Evaluating power consumption, we observed that the power efficiency of the accelerators ranged between 0.043 - 0.076 Gflops/watts for the LU decomposition in Posit(32,2) arithmetic. The power efficiency of the latest GPUs as accelerators of Posit(32,2) arithmetic is better than that of the evaluated FPGA chip.

Matrix computations are a fundamental tool in scientific and engineering applications. Among many such applications, Convolutional Neural Networks (CNN) that can be effectively computed by matrix-matrix multiplications are being popular and an efficient implementation of CNN is highly important. In this study, we have designed an parallel processor for the matrix computations using torus interconnect topology, and we implemented Cannon's algorithm for matrix-matrix multiply-add. We have evaluated the scalability of the proposed processor on a reconfigurable FPGA platform. More precisely, the designed processor with 8 × 8 functional units with 16 bit floating-point multiply-add unit was evaluated on Cyclone IV FPGA chip, with performance of 27 GFlops. We also implemented CNN calculations on our processor. We compared the matrix based approach and our proposed method. As a result, our method is 25 times faster than the matrix based approach if the processor has 8×8 functional units, image size is 32×32 and filter size is 5 × 5.

Media space is a promising but still immature technology to connect distributed sites. We developed a simple additional function that moved a remote camera forward when a local user approached a display so that the approach was amplified by a remote person's expanding image accompanied by motion parallax. We conducted an experiment in which we observed that a movable camera enhanced social telepresence, which is the feeling of facing a remote person in the same room. Despite the camera's movement, subjects believed that the camera did not move and a zoom-in function expanded the image. Surprisingly, a zoom-in camera that expanded the image as the movable camera did, however, was ineffective probably because of a lack of motion parallax. Although we explained nothing about the camera, most subjects noticed that their walking caused the view's expansion. If a remote person initiated the camera's movement, social telepresence could not be enhanced.

Recently, various robots capable of having a video chat with distant people have become commercially available. This paper shows that movement of these robots enhances distant people's presence that the robot operator feels. We conducted an experiment to compare the degrees of social telepresence produced by fixed, rotatable, movable, and automatically moving cameras. In this experiment we found that forward-backward movement of the camera significantly contributed to social telepresence, while rotation did not. We also found that this effect disappeared when the camera moved automatically. We propose the user-controllable movement of cameras as a fundamental function for video-based communication systems.