-Re: possible regression of full table scan using coprocessor in 0.94
Andrew Purtell 2012-12-01, 09:24
It would be good if this can be substantiated with a test case that
reproduces the effect noticed, or at least enough detail to write that
test. As is there's not much here to go on.
On Saturday, December 1, 2012, Ted Yu wrote:
> In Q&A session of my talk at Hadoop In China 2012, two people (one using
> 0.94.1, one using 0.94.2) mentioned that full table scan using coprocessor
> in 0.94 exhibited less desirable performance compared to 0.92.x
> I still need to collect more concrete information on this claim.
> Maybe the people who asked me this question can provide more details.
Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)