Hardware Accelerated Text Display
Abstract
Web browsers and e-book are some of the most dominant applications on mobile devices today. They spend a significant amount of time handling text in these documents. Based on the experimental results from different commercial web browsers, the majority of the time spent to display text is dedicated to layout design and painting the bitmaps of the character glyphs on the screen; the time needed to rasterize the bitmaps of these glyphs is negligible. Many efforts have been made in software to improve the performance of text layout and display and very few are trying to come up with parallel processing schemes for System-On-Chip (SoC) designs to better handle this graphic processing. This work introduces a new novel hardware-software hybrid algorithm which performs the layout design of text and displays it faster by using a small piece of hardware which can easily be added to the SoCs of today's mobile devices. This work also introduces a novel method for applying kerning to layout design process. The performance of the algorithms are compared to WebKit, the most widely used web rendering framework, and has resulted in a 29X and 192X performance increases in layout design when kerning is both used and not used respectively.