While BrowsingBench and DPIBench are embedded computer system-level benchmarks, they are implicitly testing the underlying multicore hardware (as well as software stacks). Take BrowsingBench as an example. The primary function of this embedded computer is to perform html page loads on a client device (i.e., smartphone, tablet). The page-load operation involves a series of steps that includes serial functions (i.e., enter or click URL, fetch initial HTML, parse the HTML, and determine the workload) and parallel functions that could take advantage of a multicore device (i.e., parsing, Javascript, image decoding, page rendering after all elements have been assembled, and animation). Inherently, if the client device’s browser and operating system are designed appropriately, we will see significant performance improvements on these parallel functions.
How are these technical problems best solved, by industry and the EEMBC?
refer to: http://embedded-computing.com/articles/moving-qa-markus-levy-founder-president-eembc/