lkml.org 
[lkml]   [2022]   [Jun]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
SubjectRe: [BUG REPORT] perf tools: x86_64: weight column displays odd memory latency data
Date
Hi Kan,

Thanks for the reply and sharing your output.

However, I have a couple of questions since the output is not what I have expected and appears odd.

* Are you using the OpenMP version? I ask because I don't observe any openMP symbols in your output
* Does your computer have enough ram for cg.D? (Your output does not include the iteration results and the perf file is smaller than expected)
I observe a similar issue with cg.C.x (cg.C memory requirements are around 8GB)


Regards,
Tyson

From: Liang, Kan <kan.liang@linux.intel.com>
Sent: Saturday, 4 June 2022 12:03 AM
To: Tyson Thomas <tyson.thomas@sydney.edu.au>; linux-kernel@vger.kernel.org <linux-kernel@vger.kernel.org>; linux-perf-users@vger.kernel.org <linux-perf-users@vger.kernel.org>
Subject: Re: [BUG REPORT] perf tools: x86_64: weight column displays odd memory latency data
 


On 6/1/2022 1:03 AM, Tyson Thomas wrote:
> Hi Kan,
>
> Thanks for the reply.
>
> The exact command used was: perf mem report
>
> The columns are in order: Overhead, Samples, Local Weight, Memory Access.
> The results I sent before are: Overhead, Local Weight and Memory.
>
> So, to update the table from before, each tuple has a sample of 1, so it is:
>
>     0.02%,1 ,62515 ,L1 or L1 hit
>     0.02%,1 ,54048 ,L1 or L1 hit
>     0.02%,1 ,52206 ,L1 or L1 hit
>     0.02%,1 ,49831 ,L1 or L1 hit
>     0.02%,1 ,49056 ,Local RAM or RAM hit
>     0.01%,1 ,40666 ,LFB or LFB hit
>     0.01%,1 ,38080 ,L1 or L1 hit
>     0.01%,1 ,36772 ,L1 or L1 hit
>
> So, the weights are absurdly high here.
>
> I'll give a reply on perf report -D shortly.
>
> Regards,
> Tyson
>
>
>
> From: Liang, Kan <kan.liang@linux.intel.com>
> Sent: Wednesday, 25 May 2022 12:58 AM
> To: Tyson Thomas <tyson.thomas@sydney.edu.au>; linux-kernel@vger.kernel.org <linux-kernel@vger.kernel.org>; linux-perf-users@vger.kernel.org <linux-perf-users@vger.kernel.org>
> Subject: Re: [BUG REPORT] perf tools: x86_64: weight column displays odd memory latency data
>  
>
>
> On 5/22/2022 6:14 PM, Tyson Thomas wrote:
>> Hi Kan, Linux-Perf Team
>>
>> I have observed some odd behaviour within perf when using perf-mem. Specifically the reported latency under the weight column appears to be unreasonably high.
>>
>> Here is a given sample from a recent test, I find that some of the  latencies are close to an unsigned short and I cannot seem to be understand why that would be outside of it being an issue with the perf events.
>>
>> This can be replicated using a NAS benchmark, specifically cg.D.


I tried the cg.D benchmark on my machine, but I cannot reproduce the
issue. The weight looks reasonable.

$ perf mem record ./bin/cg.D.x


  NAS Parallel Benchmarks (NPB3.4-OMP) - CG Benchmark

  Size:     1500000
  Iterations:                    100
  Number of available threads:     4

^C[ perf record: Woken up 118 times to write data ]
[ perf record: Captured and wrote 30.108 MB perf.data (438307 samples) ]

$ perf mem report --stdio
# To display the perf.data header info, please use
--header/--header-only options.
#
#
# Total Lost Samples: 0
#
# Samples: 209K of event 'cpu/mem-loads,ldlat=30/Pu'
# Total weight : 56835744
# Sort order   :
local_weight,mem,sym,dso,symbol_daddr,dso_daddr,snoop,tlb,locked,blocked,local_ins_lat,p_stage_cyc
#
# Overhead       Samples  Local Weight  Memory access             Symbol
                            Shared Object         Data Symbol >
# ........  ............  ............  ........................
................................  ....................  ............>
#
      0.01%            66  53            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.01%            53  58            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.01%            59  52            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%            53  53            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%            43  65            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%            50  55            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%            51  50            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%            46  55            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%            49  51            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2496          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%            48  52            L1 or L1 hit              [.]
randlc_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2441          Local RAM or RAM hit      [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2420          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2415          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2403          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2399          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2390          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2385          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2385          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2378          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>
      0.00%             1  2358          LFB or LFB hit            [.]
sparse_                       cg.D.x                [.] 0x00007f>

Thanks,
Kan

>>
>> I observe the following results in perf mem report (just getting the top 10 results)
>>
>
>
> Could you please show me the exact perf command used?
>
> With my perf mem report,
> The first column is the Overhead.
> The second column is the number of samples.
> The third column is the weight.
> The fourth is the Memory access.
>
> Seems like the weight is missed?
>
> Could you please check the perf report -D?
> It will dump the weight for each Sample.
> Does it look correct?
>
> Thanks,
> Kan
>
>>     0.02% ,62515 ,L1 or L1 hit
>>     0.02% ,54048 ,L1 or L1 hit
>>     0.02% ,52206 ,L1 or L1 hit
>>     0.02% ,49831 ,L1 or L1 hit
>>     0.02% ,49056 ,Local RAM or RAM hit
>>     0.01% ,40666 ,LFB or LFB hit
>>     0.01% ,38080 ,L1 or L1 hit
>>     0.01% ,36772 ,L1 or L1 hit
>>     0.01% ,36729 ,LFB or LFB hit
>>     0.01% ,27101 ,LFB or LFB hit
>>
>> Is it possible for someone to shed some light on this or am I misunderstanding how the weight column is used here?
>> This appears to have been an issue on 5.4, 5.10 and 5.15. I am looking into seeing if it is still present in 5.17 and 5.18.
>>
>> I've also tried this on different Intel CPUs such as Intel Xeon 6230, i5-1135G7, Intel Xeon 6330
>>
>> Any insight or help would be appreciated,
>> Tyson

\
 
 \ /
  Last update: 2022-06-08 03:39    [W:0.074 / U:0.076 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site