Algorithmic Efficiency - Data Presentation

Data Presentation

Output data can be presented to the end user in many ways—such as via punched tape or card, digital displays, local display monitors, remote computer monitors or printed. Each of these has its own inherent initial cost and, in some cases, an ongoing cost (e.g. refreshing an image from memory). As an example, Google's logo is a PNG image of 28K (or 48K), yet the raw text string for "Google" occupies only 6 octets or 48 bits (4,778 or 8192 times less). This graphically illustrates how the presentation of data can significantly affect the overall efficiency of transmission (and also the complete algorithm—since both GIF and PNG images require processing).

It is estimated by "Internet World Stats" that there were 1,733,993,741 internet users in 2009 and, to transmit this new image to each one of them, would require around 136,000 billion (109)octets of data to be transmitted—at least once—into their personal web cache. In "Computational Energy Cost of TCP", co-authors Bokyung Wang and Suresh Singh examine the energy costs for TCP and calculated, for their chosen example, a cost of 0.022 Joules per packet (of approx 1489 octets). On this basis, a total of around 2,000,000,000 joules (2 GJ) of energy might be expended by TCP elements alone to display the new logo for all users for the first time. To maintain or re-display this image requires still more processing and consequential energy cost (in contrast to printed output for instance).