Re: [Pixman] [PATCH 06/15] pixman-filter: reduced number of samples in Simpson's integration
On Fri, Dec 18, 2015 at 12:17 AM, Bill Spitzakwrote: > Ok to squash them together. Do you want me to do that? > Yes. There will definitely v7 of the patchset (which most likely will be the version getting merged into the tree), so do it in that version. Oded > It actually does not increase the runtime, because the two loops are only > adding every *other* sample. Thus the same number of samples are computed. > > On Thu, Dec 17, 2015 at 1:23 PM, Oded Gabbay wrote: >> >> On Sat, Dec 12, 2015 at 8:06 PM, wrote: >> > From: Bill Spitzak >> > >> > With the cubic fix this is plenty accurate enough, far in excess of the >> > pixman >> > fixed-point error limit. Likely even 16 samples is too many. >> > --- >> > pixman/pixman-filter.c | 2 +- >> > 1 file changed, 1 insertion(+), 1 deletion(-) >> > >> > diff --git a/pixman/pixman-filter.c b/pixman/pixman-filter.c >> > index 7c1da0d..4aafa51 100644 >> > --- a/pixman/pixman-filter.c >> > +++ b/pixman/pixman-filter.c >> > @@ -190,7 +190,7 @@ integral (pixman_kernel_t reconstruct, double x1, >> > else >> > { >> > /* Integration via Simpson's rule */ >> > -#define N_SEGMENTS 128 >> > +#define N_SEGMENTS 16 >> > #define SAMPLE(a1, a2) >> > \ >> > (filters[reconstruct].func ((a1)) * filters[sample].func ((a2) / >> > scale)) >> > >> > -- >> > 1.9.1 >> > >> > ___ >> > Pixman mailing list >> > Pixman@lists.freedesktop.org >> > http://lists.freedesktop.org/mailman/listinfo/pixman >> >> I think it is better to just squash this patch into the previous one, >> as it closely related and actually makes more sense to put them >> together so we can see the run time hasn't increased but actually >> decreased. >> >> Oded > > ___ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman
Re: [Pixman] [PATCH 06/15] pixman-filter: reduced number of samples in Simpson's integration
On Sat, Dec 12, 2015 at 8:06 PM,wrote: > From: Bill Spitzak > > With the cubic fix this is plenty accurate enough, far in excess of the pixman > fixed-point error limit. Likely even 16 samples is too many. > --- > pixman/pixman-filter.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/pixman/pixman-filter.c b/pixman/pixman-filter.c > index 7c1da0d..4aafa51 100644 > --- a/pixman/pixman-filter.c > +++ b/pixman/pixman-filter.c > @@ -190,7 +190,7 @@ integral (pixman_kernel_t reconstruct, double x1, > else > { > /* Integration via Simpson's rule */ > -#define N_SEGMENTS 128 > +#define N_SEGMENTS 16 > #define SAMPLE(a1, a2) \ > (filters[reconstruct].func ((a1)) * filters[sample].func ((a2) / > scale)) > > -- > 1.9.1 > > ___ > Pixman mailing list > Pixman@lists.freedesktop.org > http://lists.freedesktop.org/mailman/listinfo/pixman I think it is better to just squash this patch into the previous one, as it closely related and actually makes more sense to put them together so we can see the run time hasn't increased but actually decreased. Oded ___ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman
Re: [Pixman] [PATCH 06/15] pixman-filter: reduced number of samples in Simpson's integration
Ok to squash them together. Do you want me to do that? It actually does not increase the runtime, because the two loops are only adding every *other* sample. Thus the same number of samples are computed. On Thu, Dec 17, 2015 at 1:23 PM, Oded Gabbaywrote: > On Sat, Dec 12, 2015 at 8:06 PM, wrote: > > From: Bill Spitzak > > > > With the cubic fix this is plenty accurate enough, far in excess of the > pixman > > fixed-point error limit. Likely even 16 samples is too many. > > --- > > pixman/pixman-filter.c | 2 +- > > 1 file changed, 1 insertion(+), 1 deletion(-) > > > > diff --git a/pixman/pixman-filter.c b/pixman/pixman-filter.c > > index 7c1da0d..4aafa51 100644 > > --- a/pixman/pixman-filter.c > > +++ b/pixman/pixman-filter.c > > @@ -190,7 +190,7 @@ integral (pixman_kernel_t reconstruct, double x1, > > else > > { > > /* Integration via Simpson's rule */ > > -#define N_SEGMENTS 128 > > +#define N_SEGMENTS 16 > > #define SAMPLE(a1, a2) \ > > (filters[reconstruct].func ((a1)) * filters[sample].func ((a2) / > scale)) > > > > -- > > 1.9.1 > > > > ___ > > Pixman mailing list > > Pixman@lists.freedesktop.org > > http://lists.freedesktop.org/mailman/listinfo/pixman > > I think it is better to just squash this patch into the previous one, > as it closely related and actually makes more sense to put them > together so we can see the run time hasn't increased but actually > decreased. > > Oded > ___ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman
[Pixman] [PATCH 06/15] pixman-filter: reduced number of samples in Simpson's integration
From: Bill SpitzakWith the cubic fix this is plenty accurate enough, far in excess of the pixman fixed-point error limit. Likely even 16 samples is too many. --- pixman/pixman-filter.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/pixman/pixman-filter.c b/pixman/pixman-filter.c index 7c1da0d..4aafa51 100644 --- a/pixman/pixman-filter.c +++ b/pixman/pixman-filter.c @@ -190,7 +190,7 @@ integral (pixman_kernel_t reconstruct, double x1, else { /* Integration via Simpson's rule */ -#define N_SEGMENTS 128 +#define N_SEGMENTS 16 #define SAMPLE(a1, a2) \ (filters[reconstruct].func ((a1)) * filters[sample].func ((a2) / scale)) -- 1.9.1 ___ Pixman mailing list Pixman@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/pixman