I have a test page that does 1000 images with 2-4 PLEENQs on each image, and performance is good with an 8-gig of ram mac mini and an iPhone 5 (my test platform for the mobile version).
I do about 10 million impressions a day, and the request to the server is a pretty consistent ~200ms, along with an async load that won't affect overall page load.
I do about 10 million impressions a day, and the request to the server is a pretty consistent ~200ms, along with an async load that won't affect overall page load.