Hi Alex,

Thank you for your reply.

I don't think there there is a problem with the notion of time in the 
simulator, as we are using GEM5, which is widely used simulator in the research 
and industrial community. Secondly, we used the same GEM5 simulator on some 
other cloudsuite benchmarks, and it works fine on them.  The libraries used on 
ARM are also the same as used by you on x86 (standard debian packages), as we 
followed the docker script.

I think there might be a problem with the output as the one I posted was taken 
from the terminal, as printed by the benchmark after it is finished. The 
contents of resultsX.log is as follows

Total: connections 25 requests 422 replies 422 test-duration 332.020 s
Number of connected connections is currently = 0

Connection rate: 0.1 conn/s (13280.8 ms/conn, <=25 concurrent connections)
Connected connection rate: 0.1 conn/s (13280.8 ms/conn, <=25 concurrent 
connected connections)
Connection time [ms]: min 30006.3 avg 128810.8 max 320019.9 median 99999.5 
stddev 104499.5
Connection time [ms]: connect 0.0
Connection length [replies/conn]: 16.880

Response time (no timeouts) [ms]: 0.0
Response time (only timeouts) [ms]: 0.0
Response time (all) [ms]: 128810.8
Request rate: 1.3 req/s (786.8 ms/req)
Request size [B]: 131.0

Overall reply rate: 1.3 replies/sec
Reply rate [replies/s]: min 0.4 avg 1.3 max 8.2 stddev 1.5 (66 samples)
Reply time [ms]: response 0.0 transfer 0.0
Reply size [B]: header 164.0 content 178.0 footer 0.0 (total 342.0)
Reply status: 1xx=0 2xx=0 3xx=0 4xx=422 5xx=0

CPU time [s]: user 0.03 system 1.08 (user 0.0% system 0.3% total 0.3%)
Net I/O: 0.6 KB/s (0.0*10^6 bps)

Errors: total 0 client-timo 0 socket-timo 0 connrefused 0 connreset 0
Errors: fd-unavail 0 addrunavail 0 ftab-full 0 addrinuse 0 other 0
Unconnected timeouts 0 connected timeouts 0 total (0) client_timeouts 0


Session rate [sess/s]: min 0.00 avg 0.08 max 0.60 stddev 0.16 (25/25)
Session: avg 1.00 connections/session
Session lifetime [s]: 128.8
Session failtime [s]: 0.0
Session length histogram: 0 0 0 0 0 0 0 6 0 0 4 0 0 0 0 5 0 5 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 5


If you look at the results.log, the response time is reported as zero, both in 
the one that I have posted and the one that you provided in your last email.

Can you please let us know what is the difference in response time and reply 
time, like what do they actually mean ?

Regards
Yasir

From: Alexandros Daglis
Sent: 22 January 2017 17:08
To: Qureshi Yasir Mahmood <[email protected]>
Cc: [email protected]
Subject: Re: Media Streaming Benchmark Results

Hi Yasir,

The reply time should not be zero. It could be a problem related to the notion 
of time within the simulator (or the libraries used for getting time in ARM - 
we haven't tried running it on ARM ourselves).
If you have access to a real ARM platform, I'd suggest you first try to run the 
workload there, to separate the implications that might arise from the ISA and 
the simulation platform.
You can safely ignore the message about the descriptors - it prints at the very 
beginning of the execution, before the clients start sending requests.

As a final remark, I don't recognize the output format that you posted, so it 
might just be a problem with the output rather than the actual measurement. The 
clients create output files named resultX.log, the contents of which should 
look like the following:

-------------------------------------------------------------------------------------------
Total: connections 25 requests 422 replies 422 test-duration 332.046 s
Number of connected connections is currently = 0

Connection rate: 0.1 conn/s (13281.9 ms/conn, <=25 concurrent connections)
Connected connection rate: 0.1 conn/s (13281.9 ms/conn, <=25 concurrent 
connected connections)
Connection time [ms]: min 30012.5 avg 128838.6 max 320137.8 median 99999.5 
stddev 104516.5
Connection time [ms]: connect 0.0
Connection length [replies/conn]: 16.880

Response time (no timeouts) [ms]: 0.0
Response time (only timeouts) [ms]: 0.0
Response time (all) [ms]: 128838.6
Request rate: 1.3 req/s (786.8 ms/req)
Request size [B]: 137.0

Overall reply rate: 1.3 replies/sec
Reply rate [replies/s]: min 0.4 avg 1.3 max 8.0 stddev 1.5 (66 samples)
Reply time [ms]: response 0.1 transfer 1.6
Reply size [B]: header 294.0 content 3047157.0 footer 0.0 (total 3047451.0)
Reply status: 1xx=0 2xx=422 3xx=0 4xx=0 5xx=0

CPU time [s]: user 0.68 system 4.01 (user 0.2% system 1.2% total 1.4%)
Net I/O: 3782.4 KB/s (31.0*10^6 bps)

Errors: total 0 client-timo 0 socket-timo 0 connrefused 0 connreset 0
Errors: fd-unavail 0 addrunavail 0 ftab-full 0 addrinuse 0 other 0
Unconnected timeouts 0 connected timeouts 0 total (0) client_timeouts 0


Session rate [sess/s]: min 0.00 avg 0.08 max 0.60 stddev 0.16 (25/25)
Session: avg 1.00 connections/session
Session lifetime [s]: 128.8
Session failtime [s]: 0.0
Session length histogram: 0 0 0 0 0 0 0 6 0 0 4 0 0 0 0 5 0 5 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 5

-------------------------------------------------------------------------------------------
Regards,
Alex

On 19 Jan 2017, at 10:07, Qureshi Yasir Mahmood 
<[email protected]<mailto:[email protected]>> wrote:

Hi All,

I am running the media streaming benchmark on a simulator platform. At the end 
of the benchmark I get the following results

Total connections = 100
Total errors = 0
Percentage failure = 0
Benchmark succeeded for 500 sessions
Maximum limit for number of sessions too low.
Requests: 1688
Replies: 1688
Reply rate: 1.30
Reply time: 0
Net I/O: 2.4
httperf: maximum number of open descriptors = 0
httperf: maximum number of open descriptors = 0
Backtrace:
httperf[/root/files/videoperf/httperf.c:280]
httperf: maximum number of open descriptors = 0
httperf: maximum number of open descriptors = 0
Backtrace:
httperf[/root/files/videoperf/httperf.c:280]
httperf: maximum number of open descriptors = 0
httperf: maximum number of open descriptors = 0
Backtrace:
httperf[/root/files/videoperf/httperf.c:280]
httperf: maximum number of open descriptors = 0
httperf: maximum number of open descriptors = 0
Backtrace:
httperf[/root/files/videoperf/httperf.c:280]

I have two questions in this regard,

1.       The results show the reply rate=1.3 and total replies=1688, but reply 
time=0. Is this correct ? If yes, can someone please let me know why is the 
reply time=0
2.       Can someone please explain whatdoes "httperf: maximum number of open 
descriptors = 0" mean ?

Thank you.

Regards
Yasir Mahmood
ESL, EPFL

Reply via email to