Re: [PATCH] perf trace: Simplify '--summary' output

From: Ingo Molnar
Date: Wed Nov 13 2013 - 06:14:50 EST



* Pekka Enberg <penberg@xxxxxx> wrote:

> On 11/12/13 11:40 PM, Ingo Molnar wrote:
> >So if you prefer unit-less lines that's defensible, perhaps output the
> >unit somewhere else:
> >
> > syscall calls min avg max stddev
> > (msec) (msec) (msec) (%)
> > --------------- -------- -------- -------- -------- ------
> > sendmsg 2 0.002 0.005 0.008 55.00
> > recvmsg 2 0.002 0.003 0.005 44.00
> > epoll_wait 1 0.000 0.000 0.000 0.00
> >
> >or so?
>
> Looks good. I'll make a patch later today unless someone else beats
> me to it.

Maybe this variant is even better:

syscall calls min avg max stddev
(msec) (msec) (msec) (%)
--------------- -------- -------- -------- -------- ------
sendmsg 2 0.002 0.005 0.008 55.0%
recvmsg 2 0.002 0.003 0.005 44.0%
epoll_wait 1 0.000 0.000 0.000 0.0%

Note how we can squeeze the % into that column by using only single digit
precision percentage - double digit is pretty pointless IMO.

Also note that the different nature of the stddev column stands out better
if there's four more horizontal spaces separating it from the other
colums.

But maybe it's just me.

Btw., the msec colums seem a bit narrow, they allow up to 9999.999msec
values which is 10 seconds. If any delay is larger than that then what
happens, do the values get unaligned in an ugly fashion like they so often
do in 'vmstat'? Maybe space for one or two digits more would give it more
breathing space:

syscall calls min avg max stddev
(msec) (msec) (msec) (%)
--------------- -------- --------- --------- --------- ------
sendmsg 2 0.002 0.005 0.008 55.0%
recvmsg 2 0.002 0.003 0.005 44.0%
epoll_wait 1 0.000 0.000 0.000 0.0%

This still looks directly pasteable into email and commit logs and looks
nicely compact on my terminals as well.

Thanks,

Ingo
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/