First time user, problem with toplev.py levels (-l)

First time user, problem with toplev.py levels (-l)

[Luke Gorrie]

Howdy!

I'm a first-time user of toplev.py today. The -l option isn't working
for me, and I'm wondering if someone can help me understand why?

I've tested with a few different programs under test and combinations
of -l<n>, -v, and -d. No luck yet. Here's an example below:

$ uname -a
Linux chur 3.11.0-8-generic #15-Ubuntu SMP Fri Sep 20 04:11:26 UTC
2013 x86_64 x86_64 x86_64 GNU/Linux
$ perf --version
perf version 3.11.1
$ git log | head -1
commit c765cf1ae73efce887468abf4c7a3f4fbe549763
$ ./toplev.py -v -d -l4 sleep 5
WARNING: HT enabled
Measuring multiple processes/threads on the same core may not be reliable.
perf stat -x, -e '{r2c2,r100030d,r19c,r10e,cycles}' sleep 5
Frontend Bound:                                 10.11% below
        This category reflects slots where the Frontend of the
processor undersupplies
        its Backend.
Bad Speculation:                                 9.05% below
        This category reflects slots wasted due to incorrect speculations, which
        include slots used to allocate uops that do not eventually get
retired and
        slots for which allocation was blocked due to recovery from
earlier incorrect
        speculation. For example, wasted work due to miss-predicted branches is
        categorized under the Bad Speculation category
Backend Bound:                                  59.36% above
        This category reflects slots where no uops are being delivered
due to a lack
        of required resources for accepting more uops in the Backend
of the pipeline.
Retiring:                                       21.48% below
        This category reflects slots utilized by good uops i.e.
allocated uops that
        eventually get retired.

any tips would be much appreciated.

Cheers,
-Luke

Re: First time user, problem with toplev.py levels (-l)

[Andi Kleen]

Luke Gorrie <luke@snabb.co> writes:

> I'm a first-time user of toplev.py today. The -l option isn't working
> for me, and I'm wondering if someone can help me understand why?
>
> I've tested with a few different programs under test and combinations
> of -l<n>, -v, and -d. No luck yet. Here's an example below:

What CPU do you have (head -n20 /proc/cpuinfo) ? 

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only

Re: First time user, problem with toplev.py levels (-l)

[Luke Gorrie]

On 25 September 2013 17:24, Andi Kleen <andi@firstfloor.org> wrote:
> What CPU do you have (head -n20 /proc/cpuinfo) ?

$ head -n20 /proc/cpuinfo
processor       : 0
vendor_id       : GenuineIntel
cpu family      : 6
model           : 45
model name      : Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz
stepping        : 7
microcode       : 0x70d
cpu MHz         : 1200.000
cache size      : 20480 KB
physical id     : 0
siblings        : 16
core id         : 0
cpu cores       : 8
apicid          : 0
initial apicid  : 0
fpu             : yes
fpu_exception   : yes
cpuid level     : 13
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge
mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe
syscall nx pdpe1gb rdtscp lm constant_tsc arch_perfmon pebs bts
rep_good nopl xtopology nonstop_tsc aperfmperf eagerfpu pni pclmulqdq
dtes64 monitor ds_cpl vmx smx est tm2 ssse3 cx16 xtpr pdcm pcid dca
sse4_1 sse4_2 x2apic popcnt tsc_deadline_timer aes xsave avx lahf_lm
ida arat epb xsaveopt pln pts dtherm tpr_shadow vnmi flexpriority ept
vpid

Re: First time user, problem with toplev.py levels (-l)

[Andi Kleen]

On Wed, Sep 25, 2013 at 05:29:49PM +0200, Luke Gorrie wrote:
> On 25 September 2013 17:24, Andi Kleen <andi@firstfloor.org> wrote:
> > What CPU do you have (head -n20 /proc/cpuinfo) ?
> 
> $ head -n20 /proc/cpuinfo
> processor       : 0
> vendor_id       : GenuineIntel
> cpu family      : 6
> model           : 45
> model name      : Intel(R) Xeon(R) CPU E5-2650 0 @ 2.00GHz

toplev doesn't have a detailed model for SandyBridge-E currently,
that is why only level 1 is enabled.

You could force snb in the code for jkt (the table at line 175+), but some
of the estimated latencies would be a bit off and it would not 
support NUMA.

-Andi

Re: First time user, problem with toplev.py levels (-l)

[Luke Gorrie]

On 25 September 2013 17:54, Andi Kleen <andi@firstfloor.org> wrote:
> toplev doesn't have a detailed model for SandyBridge-E currently,
> that is why only level 1 is enabled.

Ah, I see.

> You could force snb in the code for jkt (the table at line 175+), but some
> of the estimated latencies would be a bit off and it would not
> support NUMA.

Thanks for the tip. I tried FORCECPU=snb and now it appears I at least
have something to play with :-).

Cheers,
-Luke