It can help, under some circumstances, but it's also liable to make some sites not render properly or suffer other issues, and it may not always be easy to spot the differences.
Pipelining in this context means launching multiple HTTP requests in parallel rather than waiting for each request to complete before the next one starts (serialised requesting). From memory, the HTTP 1.0 standard provides support for up to 4 concurrent requests, while HTTP 1.1 drops that down to 2. Irrespective, there are plenty of servers which limit themselves to servicing only 1 single request at a time per client. All additional concurrent requests will be ignored, and the results will vary depending on the page design and the browser you're using. Whatever it is, it won't be pretty.
There's no harm in trying it, as long as you remember to put it back if your favourite sites are suddenly not working properly.
Tweaking these settings does nothing whatsoever to speed up the "single large file" transfer scenario.
