This is why benchmarks are often meaningless when measuring 'responsiveness'.
Originally Posted by Rallos Zek
Don't take this the wrong way, but the fact you would even bring up Deadline shows you don't have any real-world experience with it on a Desktop system doing parallel tasks.
Swap to Deadline, start copying a large file off your HDD to a device with high write speed, and then attempt to open, say, LibreOffice Writer.
Then try the same thing with BFQ.
Exactly what is done by one of the benchmarks reported in the BFQ result page (http://algo.ing.unimo.it/people/paol...ed/results.php). The only difference is that the benchmark repeats this action a lot of times to compute more reliable statistics, and clears the cache before each repetition for a fair comparison (among BFQ, CFQ, DEADLINE and NOOP).
Originally Posted by korrode