Paper on the Inside Aspects of Intel Ubuntu Processors

namespace. Figure Y shows CDFs of the page load times obtained from this experiment. As shown, Firefox marginally outperforms Chrome on each considered configuration. On a 25 Mbps link with 30 ms minimum RTT, Firefox has a median page load time 4.6% lower than Chrome. Finally, on a 1 Mbps link with 300 ms minimum RTT, Firefox’s median page load time is 6.3% faster than that of Chrome. the 95 percentile page load time is On a 14 Mbps link with 120 ms minimum RTT, the 95th percentile page load time is 6.4% lower for Firefox as compared to Chrome.

Page load time is directly affected by the order in which resources are requested and processed. Changes to the lay- out and javascript parsing engines affect this order as they dictate when objects will be rendered and when processing of javascript occurs which in turn dictates when requests are made. Similarly, network decisions such as maximum con- nections per destination host and maximum total number of connections per process dictate when requests can be made as certain connections may be blocked until existing connec- tions are available for reuse. While Chrome and Firefox share the same maximums for connections per destination host and total number of connections per process, they use different layout and javascript parsing engines. As a result, the order in which they request objects is not always the same. Evaluating this, however, is difficult as page content is everchanging and network conditions are not constant which may affect the order in which requested objects are received and subsequent requests are made.

Using Mahimahi, we are able to overcome these run-to-run disparities by using the same recorded web content while loading pages locally inside ReplayShell. We use Chrome and Firefox to load a recorded copy of http://www.bing. com. We record the order of requested objects using Chrome Developer Tools [? ] and Firefox Developer Tools [? ]. Table T shows the the differences in object request order between...