• Journal of Broadcast Engineering
• /
• v.28 no.1
• /
• pp.21-30
• /
• 2023

Computer-generated holography requires much more computation costs and memory space rather than image processing. We implemented the diffraction calculation with low-precision and mixed-precision floating point numbers and compared the processing time and quality of the hologram with various precision. We compared diffraction quality with double, single and bfloat16 precision. bfloat16 shows 5.94x and 1.52x times faster performance than double precision and single precision. Also, bfloat16 shows lower PSNR and SSIM and higher MSE than other precision. However, there is no significant effect on reconstructed images. These results show low precision, like bfloat16, can be utilized for computer-generated holography.

• ETRI Journal
• /
• v.43 no.4
• /
• pp.684-693
• /
• 2021

This paper presents a low-cost two-stage approximate multiplier for bfloat16 (brain floating-point) data processing. For cost-efficient approximate multiplication, the first stage implements Mitchell's algorithm that performs the approximate multiplication using only two adders. The second stage adopts the exact multiplication to compensate for the error from the first stage by multiplying error terms and adding its truncated result to the final output. In our design, the low-cost multiplications in both stages can reduce hardware costs significantly and provide low relative errors by compensating for the error from the first stage. We apply our approximate multiplier to the convolutional neural network (CNN) inferences, which shows small accuracy drops with well-known pre-trained models for the ImageNet database. Therefore, our design allows low-cost CNN inference systems with high test accuracy.