Re: Vector-alignment patch performance regressions

2012-08-08 Thread Mans Rullgard
On 8 August 2012 16:48, Matthew Gretton-Dann wrote: > On 8 August 2012 16:24, Ulrich Weigand wrote: >> >> Hello, >> >> I've had a look at the mp3player performance regressions (just with *some* >> data sets) with the vector-alignment patch. Interestingly it turns out >> that the patch basically

Re: Vector-alignment patch performance regressions

2012-08-08 Thread Mans Rullgard
On 8 August 2012 16:24, Ulrich Weigand wrote: > > Hello, > > I've had a look at the mp3player performance regressions (just with *some* > data sets) with the vector-alignment patch. Interestingly it turns out > that the patch basically does not change the generated code for the hot > spot (inv_md

Re: Vector-alignment patch performance regressions

2012-08-08 Thread Matthew Gretton-Dann
On 8 August 2012 16:24, Ulrich Weigand wrote: > > Hello, > > I've had a look at the mp3player performance regressions (just with *some* > data sets) with the vector-alignment patch. Interestingly it turns out > that the patch basically does not change the generated code for the hot > spot (inv_md

Vector-alignment patch performance regressions

2012-08-08 Thread Ulrich Weigand
Hello, I've had a look at the mp3player performance regressions (just with *some* data sets) with the vector-alignment patch. Interestingly it turns out that the patch basically does not change the generated code for the hot spot (inv_mdct routine) at all. (The *only* change is which bits of th