Re: [PATCH 1/6] perf, tools, stat: Abstract stat metrics printing
On Tue, Dec 22, 2015 at 02:16:30AM +0100, Andi Kleen wrote:
> > > - fprintf(out, " ");
> > > + print_metric(ctxp, NULL, NULL, "insn per cycle", 0);
> > > }
> > > total =
> > > avg_stats(_stalled_cycles_front_stats[ctx][cpu]);
> > > total = max(total,
> > > avg_stats(_stalled_cycles_back_stats[ctx][cpu]));
> > >
> > > + out->new_line(ctxp);
> > > if (total && avg) {
> > > ratio = total / avg;
> > > - fprintf(out, "\n");
> >
> > you haven't address my first comment in here
> > http://marc.info/?l=linux-kernel=144662610723134=2
>
> The new_line is always needed because stalled cycles is always printed.
>
> The reason it is always printed is that metric-only needs to see all the
> metrics for its column headers. That's why there are else cases
> everywhere.
please notice extra line below - between instructions and cycles lines
[jolsa@krava perf]$ ./perf stat -e instructions,cycles kill
kill: not enough arguments
Performance counter stats for 'kill':
769,784 instructions #0.78 insn per cycle
990,795 cycles
jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH 1/6] perf, tools, stat: Abstract stat metrics printing
> > - fprintf(out, " ");
> > + print_metric(ctxp, NULL, NULL, "insn per cycle", 0);
> > }
> > total =
> > avg_stats(_stalled_cycles_front_stats[ctx][cpu]);
> > total = max(total,
> > avg_stats(_stalled_cycles_back_stats[ctx][cpu]));
> >
> > + out->new_line(ctxp);
> > if (total && avg) {
> > ratio = total / avg;
> > - fprintf(out, "\n");
>
> you haven't address my first comment in here
> http://marc.info/?l=linux-kernel=144662610723134=2
The new_line is always needed because stalled cycles is always printed.
The reason it is always printed is that metric-only needs to see all the
metrics for its column headers. That's why there are else cases
everywhere.
-Andi
--
a...@linux.intel.com -- Speaking for myself only.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH 1/6] perf, tools, stat: Abstract stat metrics printing
On Mon, Dec 14, 2015 at 06:04:14PM -0800, Andi Kleen wrote:
SNIP
> -double avg, int cpu, enum aggr_mode aggr)
> +void perf_stat__print_shadow_stats(struct perf_evsel *evsel,
> +double avg, int cpu,
> +struct perf_stat_output_ctx *out)
> {
> + void *ctxp = out->ctx;
> + print_metric_t print_metric = out->print_metric;
> double total, ratio = 0.0, total2;
> int ctx = evsel_context(evsel);
>
> @@ -307,119 +302,145 @@ void perf_stat__print_shadow_stats(FILE *out, struct
> perf_evsel *evsel,
> total = avg_stats(_cycles_stats[ctx][cpu]);
> if (total) {
> ratio = avg / total;
> - fprintf(out, " # %5.2f insns per cycle",
> ratio);
> + print_metric(ctxp, NULL, "%7.2f ",
> + "insn per cycle", ratio);
> } else {
> - fprintf(out, " ");
> + print_metric(ctxp, NULL, NULL, "insn per cycle", 0);
> }
> total =
> avg_stats(_stalled_cycles_front_stats[ctx][cpu]);
> total = max(total,
> avg_stats(_stalled_cycles_back_stats[ctx][cpu]));
>
> + out->new_line(ctxp);
> if (total && avg) {
> ratio = total / avg;
> - fprintf(out, "\n");
you haven't address my first comment in here
http://marc.info/?l=linux-kernel=144662610723134=2
also please rebase to latest acme/perf/core
FYI there's been some stat changes, but your branch cleanly rebased for me
thanks,
jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH 1/6] perf, tools, stat: Abstract stat metrics printing
> > - fprintf(out, " ");
> > + print_metric(ctxp, NULL, NULL, "insn per cycle", 0);
> > }
> > total =
> > avg_stats(_stalled_cycles_front_stats[ctx][cpu]);
> > total = max(total,
> > avg_stats(_stalled_cycles_back_stats[ctx][cpu]));
> >
> > + out->new_line(ctxp);
> > if (total && avg) {
> > ratio = total / avg;
> > - fprintf(out, "\n");
>
> you haven't address my first comment in here
> http://marc.info/?l=linux-kernel=144662610723134=2
The new_line is always needed because stalled cycles is always printed.
The reason it is always printed is that metric-only needs to see all the
metrics for its column headers. That's why there are else cases
everywhere.
-Andi
--
a...@linux.intel.com -- Speaking for myself only.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH 1/6] perf, tools, stat: Abstract stat metrics printing
On Tue, Dec 22, 2015 at 02:16:30AM +0100, Andi Kleen wrote:
> > > - fprintf(out, " ");
> > > + print_metric(ctxp, NULL, NULL, "insn per cycle", 0);
> > > }
> > > total =
> > > avg_stats(_stalled_cycles_front_stats[ctx][cpu]);
> > > total = max(total,
> > > avg_stats(_stalled_cycles_back_stats[ctx][cpu]));
> > >
> > > + out->new_line(ctxp);
> > > if (total && avg) {
> > > ratio = total / avg;
> > > - fprintf(out, "\n");
> >
> > you haven't address my first comment in here
> > http://marc.info/?l=linux-kernel=144662610723134=2
>
> The new_line is always needed because stalled cycles is always printed.
>
> The reason it is always printed is that metric-only needs to see all the
> metrics for its column headers. That's why there are else cases
> everywhere.
please notice extra line below - between instructions and cycles lines
[jolsa@krava perf]$ ./perf stat -e instructions,cycles kill
kill: not enough arguments
Performance counter stats for 'kill':
769,784 instructions #0.78 insn per cycle
990,795 cycles
jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Re: [PATCH 1/6] perf, tools, stat: Abstract stat metrics printing
On Mon, Dec 14, 2015 at 06:04:14PM -0800, Andi Kleen wrote:
SNIP
> -double avg, int cpu, enum aggr_mode aggr)
> +void perf_stat__print_shadow_stats(struct perf_evsel *evsel,
> +double avg, int cpu,
> +struct perf_stat_output_ctx *out)
> {
> + void *ctxp = out->ctx;
> + print_metric_t print_metric = out->print_metric;
> double total, ratio = 0.0, total2;
> int ctx = evsel_context(evsel);
>
> @@ -307,119 +302,145 @@ void perf_stat__print_shadow_stats(FILE *out, struct
> perf_evsel *evsel,
> total = avg_stats(_cycles_stats[ctx][cpu]);
> if (total) {
> ratio = avg / total;
> - fprintf(out, " # %5.2f insns per cycle",
> ratio);
> + print_metric(ctxp, NULL, "%7.2f ",
> + "insn per cycle", ratio);
> } else {
> - fprintf(out, " ");
> + print_metric(ctxp, NULL, NULL, "insn per cycle", 0);
> }
> total =
> avg_stats(_stalled_cycles_front_stats[ctx][cpu]);
> total = max(total,
> avg_stats(_stalled_cycles_back_stats[ctx][cpu]));
>
> + out->new_line(ctxp);
> if (total && avg) {
> ratio = total / avg;
> - fprintf(out, "\n");
you haven't address my first comment in here
http://marc.info/?l=linux-kernel=144662610723134=2
also please rebase to latest acme/perf/core
FYI there's been some stat changes, but your branch cleanly rebased for me
thanks,
jirka
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
