[correct group address this time] On Tue, Sep 9, 2008 at 4:59 PM, Michael Nordman <[EMAIL PROTECTED]> wrote:
> ======================================================================== > > http://mondrian.corp.google.com/file/7970017///depot/googleclient/gears/opensource/gears/test/manual/workerpool_limit.html?a=4 > File > //depot/googleclient/gears/opensource/gears/test/manual/workerpool_limit.html > (snapshot 4) > ------------------------------------ > Line 75: numResponses++; > google style thing, prefer pre-increment > ------------------------------------ > Line 92: numWorkers++; > ditto > ------------------------------------ > Line 93: document.getElementById("numWorkers").innerHTML = numWorkers; > your timings also include DOM manipulation stuff, not sure how variable > that > part is across browsers but if it varies by more than Gears > createWorker/sendMessage does, it could dominate the time measurements. > > i wonder if you remove incremental UI updates and just publish the results > at > the end, how much the numbers change by? Hmm, good point. The measurement here wasn't really meant as hard data when I added it, but that makes it a bit misleading. I limited the DOM updates to happen every after every 100 workers at the same time the latency average is being calculated. It speeds things up noticeably in FF and IE on my workstation. > ======================================================================== > > -- > To respond, reply to this email or visit > http://mondrian.corp.google.com/7970017 >
