* [PATCH] perf/ui/tui: don't force a refresh during progress update
@ 2013-10-23 21:59 Patrick Palka
2013-10-25 10:34 ` Ingo Molnar
2013-10-26 0:25 ` [PATCH rebase] " Patrick Palka
0 siblings, 2 replies; 5+ messages in thread
From: Patrick Palka @ 2013-10-23 21:59 UTC (permalink / raw)
To: linux-kernel; +Cc: Peter Zijlstra, Paul Mackerras, Ingo Molnar
Each call to tui_progress__update() would forcibly refresh the entire
screen. This is somewhat inefficient and causes noticable flickering
during the startup of perf-report, especially on large/slow terminals.
It looks like the force-refresh in tui_progress__update() serves no
purpose other than to clear the screen so that the progress bar of a
previous operation does not subsume with that of a subsequent operation.
But we can do just that in a much more efficient manner by clearing only
the region that a previous progress bar may have occupied before
repainting the new progress bar. Then the force-refresh could be
removed with no change in visuals.
This patch disables the slow force-refresh in tui_progress__update() and
instead calls SLsmg_fill_region() on the entire area that the progress
bar may occupy before repainting it. This change makes the startup of
perf-report much faster and appear much "smoother".
It turns out that this was a big bottleneck in the startup speed of
perf-report -- with this patch, perf-report starts up ~1.6x faster (0.8s
vs 0.5s) on my machines. (These numbers were measured by running "time
perf report" on an 8MB perf.data and pressing 'q' immediately.)
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Ingo Molnar <mingo@redhat.com>
Signed-off-by: Patrick Palka <patrick@parcs.ath.cx>
---
tools/perf/ui/tui/progress.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/tools/perf/ui/tui/progress.c b/tools/perf/ui/tui/progress.c
index 6c2184d..641049a 100644
--- a/tools/perf/ui/tui/progress.c
+++ b/tools/perf/ui/tui/progress.c
@@ -17,13 +17,14 @@ static void tui_progress__update(u64 curr, u64 total, const char *title)
if (total == 0)
return;
- ui__refresh_dimensions(true);
+ ui__refresh_dimensions(false);
pthread_mutex_lock(&ui__lock);
y = SLtt_Screen_Rows / 2 - 2;
SLsmg_set_color(0);
SLsmg_draw_box(y, 0, 3, SLtt_Screen_Cols);
SLsmg_gotorc(y++, 1);
SLsmg_write_string((char *)title);
+ SLsmg_fill_region(y, 1, 1, SLtt_Screen_Cols - 2, ' ');
SLsmg_set_color(HE_COLORSET_SELECTED);
bar = ((SLtt_Screen_Cols - 2) * curr) / total;
SLsmg_fill_region(y, 1, 1, bar, ' ');
--
1.8.4.rc3
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH] perf/ui/tui: don't force a refresh during progress update
2013-10-23 21:59 [PATCH] perf/ui/tui: don't force a refresh during progress update Patrick Palka
@ 2013-10-25 10:34 ` Ingo Molnar
2013-10-26 0:25 ` [PATCH rebase] " Patrick Palka
1 sibling, 0 replies; 5+ messages in thread
From: Ingo Molnar @ 2013-10-25 10:34 UTC (permalink / raw)
To: Patrick Palka; +Cc: linux-kernel, Peter Zijlstra, Paul Mackerras, Ingo Molnar
* Patrick Palka <patrick@parcs.ath.cx> wrote:
> Each call to tui_progress__update() would forcibly refresh the entire
> screen. This is somewhat inefficient and causes noticable flickering
> during the startup of perf-report, especially on large/slow terminals.
>
> It looks like the force-refresh in tui_progress__update() serves no
> purpose other than to clear the screen so that the progress bar of a
> previous operation does not subsume with that of a subsequent operation.
> But we can do just that in a much more efficient manner by clearing only
> the region that a previous progress bar may have occupied before
> repainting the new progress bar. Then the force-refresh could be
> removed with no change in visuals.
>
> This patch disables the slow force-refresh in tui_progress__update() and
> instead calls SLsmg_fill_region() on the entire area that the progress
> bar may occupy before repainting it. This change makes the startup of
> perf-report much faster and appear much "smoother".
>
> It turns out that this was a big bottleneck in the startup speed of
> perf-report -- with this patch, perf-report starts up ~1.6x faster (0.8s
> vs 0.5s) on my machines. (These numbers were measured by running "time
> perf report" on an 8MB perf.data and pressing 'q' immediately.)
>
> Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
> Cc: Paul Mackerras <paulus@samba.org>
> Cc: Ingo Molnar <mingo@redhat.com>
> Signed-off-by: Patrick Palka <patrick@parcs.ath.cx>
> ---
> tools/perf/ui/tui/progress.c | 3 ++-
> 1 file changed, 2 insertions(+), 1 deletion(-)
>
> diff --git a/tools/perf/ui/tui/progress.c b/tools/perf/ui/tui/progress.c
> index 6c2184d..641049a 100644
> --- a/tools/perf/ui/tui/progress.c
> +++ b/tools/perf/ui/tui/progress.c
> @@ -17,13 +17,14 @@ static void tui_progress__update(u64 curr, u64 total, const char *title)
> if (total == 0)
> return;
>
> - ui__refresh_dimensions(true);
> + ui__refresh_dimensions(false);
> pthread_mutex_lock(&ui__lock);
> y = SLtt_Screen_Rows / 2 - 2;
> SLsmg_set_color(0);
> SLsmg_draw_box(y, 0, 3, SLtt_Screen_Cols);
> SLsmg_gotorc(y++, 1);
> SLsmg_write_string((char *)title);
> + SLsmg_fill_region(y, 1, 1, SLtt_Screen_Cols - 2, ' ');
> SLsmg_set_color(HE_COLORSET_SELECTED);
> bar = ((SLtt_Screen_Cols - 2) * curr) / total;
> SLsmg_fill_region(y, 1, 1, bar, ' ');
Nice! This is something I noticed as well, never figured out the
root cause of it.
Acked-by: Ingo Molnar <mingo@kernel.org>
Thanks,
Ingo
^ permalink raw reply [flat|nested] 5+ messages in thread
* [PATCH rebase] perf/ui/tui: don't force a refresh during progress update
2013-10-23 21:59 [PATCH] perf/ui/tui: don't force a refresh during progress update Patrick Palka
2013-10-25 10:34 ` Ingo Molnar
@ 2013-10-26 0:25 ` Patrick Palka
2013-10-28 5:52 ` Namhyung Kim
2013-11-12 21:54 ` [tip:perf/urgent] perf ui tui progress: Don' t " tip-bot for Patrick Palka
1 sibling, 2 replies; 5+ messages in thread
From: Patrick Palka @ 2013-10-26 0:25 UTC (permalink / raw)
To: linux-kernel
Cc: Peter Zijlstra, Paul Mackerras, Arnaldo Carvalho de Melo,
Namhyung Kim, Ingo Molnar
Each call to tui_progress__update() would forcibly refresh the entire
screen. This is somewhat inefficient and causes noticable flickering
during the startup of perf-report, especially on large/slow terminals.
It looks like the force-refresh in tui_progress__update() serves no
purpose other than to clear the screen so that the progress bar of a
previous operation does not subsume that of a subsequent operation.
But we can do just that in a much more efficient manner by clearing only
the region that a previous progress bar may have occupied before
repainting the new progress bar. Then the force-refresh could be
removed with no change in visuals.
This patch disables the slow force-refresh in tui_progress__update() and
instead calls SLsmg_fill_region() on the entire area that the progress
bar may occupy before repainting it. This change makes the startup of
perf-report much faster and appear much "smoother".
It turns out that this was a big bottleneck in the startup speed of
perf-report -- with this patch, perf-report starts up ~2x faster (1.1s
vs 0.55s) on my machines. (These numbers were measured by running
"time perf report" on an 8MB perf.data and pressing 'q' immediately.)
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Ingo Molnar <mingo@kernel.org>
Acked-by: Ingo Molnar <mingo@kernel.org>
Signed-off-by: Patrick Palka <patrick@parcs.ath.cx>
---
Rebased against
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git perf/core
And also CC'd two more relevant persons.
tools/perf/ui/tui/progress.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/tools/perf/ui/tui/progress.c b/tools/perf/ui/tui/progress.c
index 3e2d936..c61d14b 100644
--- a/tools/perf/ui/tui/progress.c
+++ b/tools/perf/ui/tui/progress.c
@@ -18,13 +18,14 @@ static void tui_progress__update(struct ui_progress *p)
if (p->total == 0)
return;
- ui__refresh_dimensions(true);
+ ui__refresh_dimensions(false);
pthread_mutex_lock(&ui__lock);
y = SLtt_Screen_Rows / 2 - 2;
SLsmg_set_color(0);
SLsmg_draw_box(y, 0, 3, SLtt_Screen_Cols);
SLsmg_gotorc(y++, 1);
SLsmg_write_string((char *)p->title);
+ SLsmg_fill_region(y, 1, 1, SLtt_Screen_Cols - 2, ' ');
SLsmg_set_color(HE_COLORSET_SELECTED);
bar = ((SLtt_Screen_Cols - 2) * p->curr) / p->total;
SLsmg_fill_region(y, 1, 1, bar, ' ');
--
1.8.4.rc3
^ permalink raw reply related [flat|nested] 5+ messages in thread
* Re: [PATCH rebase] perf/ui/tui: don't force a refresh during progress update
2013-10-26 0:25 ` [PATCH rebase] " Patrick Palka
@ 2013-10-28 5:52 ` Namhyung Kim
2013-11-12 21:54 ` [tip:perf/urgent] perf ui tui progress: Don' t " tip-bot for Patrick Palka
1 sibling, 0 replies; 5+ messages in thread
From: Namhyung Kim @ 2013-10-28 5:52 UTC (permalink / raw)
To: Patrick Palka
Cc: linux-kernel, Peter Zijlstra, Paul Mackerras,
Arnaldo Carvalho de Melo, Ingo Molnar
Hi Patrick,
On Fri, 25 Oct 2013 20:25:49 -0400, Patrick Palka wrote:
> Each call to tui_progress__update() would forcibly refresh the entire
> screen. This is somewhat inefficient and causes noticable flickering
> during the startup of perf-report, especially on large/slow terminals.
>
> It looks like the force-refresh in tui_progress__update() serves no
> purpose other than to clear the screen so that the progress bar of a
> previous operation does not subsume that of a subsequent operation.
> But we can do just that in a much more efficient manner by clearing only
> the region that a previous progress bar may have occupied before
> repainting the new progress bar. Then the force-refresh could be
> removed with no change in visuals.
>
> This patch disables the slow force-refresh in tui_progress__update() and
> instead calls SLsmg_fill_region() on the entire area that the progress
> bar may occupy before repainting it. This change makes the startup of
> perf-report much faster and appear much "smoother".
>
> It turns out that this was a big bottleneck in the startup speed of
> perf-report -- with this patch, perf-report starts up ~2x faster (1.1s
> vs 0.55s) on my machines. (These numbers were measured by running
> "time perf report" on an 8MB perf.data and pressing 'q' immediately.)
Acked-by: Namhyung Kim <namhyung@kernel.org>
Thanks,
Namhyung
^ permalink raw reply [flat|nested] 5+ messages in thread
* [tip:perf/urgent] perf ui tui progress: Don' t force a refresh during progress update
2013-10-26 0:25 ` [PATCH rebase] " Patrick Palka
2013-10-28 5:52 ` Namhyung Kim
@ 2013-11-12 21:54 ` tip-bot for Patrick Palka
1 sibling, 0 replies; 5+ messages in thread
From: tip-bot for Patrick Palka @ 2013-11-12 21:54 UTC (permalink / raw)
To: linux-tip-commits
Cc: acme, linux-kernel, paulus, hpa, mingo, a.p.zijlstra, patrick,
namhyung, tglx
Commit-ID: d53e57d039c323fe3a43630e9f729df48134e2c9
Gitweb: http://git.kernel.org/tip/d53e57d039c323fe3a43630e9f729df48134e2c9
Author: Patrick Palka <patrick@parcs.ath.cx>
AuthorDate: Fri, 25 Oct 2013 20:25:49 -0400
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Mon, 11 Nov 2013 15:56:39 -0300
perf ui tui progress: Don't force a refresh during progress update
Each call to tui_progress__update() would forcibly refresh the entire
screen. This is somewhat inefficient and causes noticable flickering
during the startup of perf-report, especially on large/slow terminals.
It looks like the force-refresh in tui_progress__update() serves no
purpose other than to clear the screen so that the progress bar of a
previous operation does not subsume that of a subsequent operation. But
we can do just that in a much more efficient manner by clearing only the
region that a previous progress bar may have occupied before repainting
the new progress bar. Then the force-refresh could be removed with no
change in visuals.
This patch disables the slow force-refresh in tui_progress__update() and
instead calls SLsmg_fill_region() on the entire area that the progress
bar may occupy before repainting it. This change makes the startup of
perf-report much faster and appear much "smoother".
It turns out that this was a big bottleneck in the startup speed of
perf-report -- with this patch, perf-report starts up ~2x faster (1.1s
vs 0.55s) on my machines. (These numbers were measured by running "time
perf report" on an 8MB perf.data and pressing 'q' immediately.)
Signed-off-by: Patrick Palka <patrick@parcs.ath.cx>
Acked-by: Ingo Molnar <mingo@kernel.org>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1382747149-9716-1-git-send-email-patrick@parcs.ath.cx
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/ui/tui/progress.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/tools/perf/ui/tui/progress.c b/tools/perf/ui/tui/progress.c
index 3e2d936..c61d14b 100644
--- a/tools/perf/ui/tui/progress.c
+++ b/tools/perf/ui/tui/progress.c
@@ -18,13 +18,14 @@ static void tui_progress__update(struct ui_progress *p)
if (p->total == 0)
return;
- ui__refresh_dimensions(true);
+ ui__refresh_dimensions(false);
pthread_mutex_lock(&ui__lock);
y = SLtt_Screen_Rows / 2 - 2;
SLsmg_set_color(0);
SLsmg_draw_box(y, 0, 3, SLtt_Screen_Cols);
SLsmg_gotorc(y++, 1);
SLsmg_write_string((char *)p->title);
+ SLsmg_fill_region(y, 1, 1, SLtt_Screen_Cols - 2, ' ');
SLsmg_set_color(HE_COLORSET_SELECTED);
bar = ((SLtt_Screen_Cols - 2) * p->curr) / p->total;
SLsmg_fill_region(y, 1, 1, bar, ' ');
^ permalink raw reply related [flat|nested] 5+ messages in thread
end of thread, other threads:[~2013-11-12 21:55 UTC | newest]
Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-10-23 21:59 [PATCH] perf/ui/tui: don't force a refresh during progress update Patrick Palka
2013-10-25 10:34 ` Ingo Molnar
2013-10-26 0:25 ` [PATCH rebase] " Patrick Palka
2013-10-28 5:52 ` Namhyung Kim
2013-11-12 21:54 ` [tip:perf/urgent] perf ui tui progress: Don' t " tip-bot for Patrick Palka
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).