Pro vs. E02 - Some Benchmarks

Re: Pro vs. E02 - Some Benchmarks

JoeyPogoPlugE02 — Sat, 21 Mar 2015 23:28:12 -0500

Thanks from me too. I don't suppose you have a watt meter on hand would you? No matter what other devices I see out there, I keep coming back to the idea something can run on 5 watts or less.

By way of comparison, my older DISH network 311 receiver takes up 20 watts even when it's supposedly off. Then a couple watts higher when "on".

So i still have plans for my E02 to be the heartbeat of a home cloud. And way less heat than any thin client I've tested. Thin Clients can get warm!

Re: Pro vs. E02 - Some Benchmarks

LeggoMyEggo — Thu, 19 Mar 2015 19:19:58 -0500

Gravelrash Wrote:
-------------------------------------------------------
> doesnt surprise me tho, when its a single core
> arm5te versus an armv6 dual core with faster
> memory.

For some strange reason I thought the Kirkwoods always performed better than the Oxnas and judging by the level of support for Oxnas - barely any and if it weren't for the work of people of this board then there would be none - it was destined for the ashheap. I guess that was wrong.

Re: Pro vs. E02 - Some Benchmarks

bodhi — Thu, 19 Mar 2015 19:17:44 -0500

Gravelrash Wrote:
-------------------------------------------------------
> thanks for the benchmarks. gives me something to
> compare mine against.....
> doesnt surprise me tho, when its a single core
> arm5te versus an armv6 dual core with faster
> memory.

Yes, dual-core the reason.

Re: Pro vs. E02 - Some Benchmarks

Gravelrash — Thu, 19 Mar 2015 18:51:44 -0500

thanks for the benchmarks. gives me something to compare mine against.....
doesnt surprise me tho, when its a single core arm5te versus an armv6 dual core with faster memory.

Re: Pro vs. E02 - Some Benchmarks

LeggoMyEggo — Thu, 19 Mar 2015 18:34:24 -0500

Just did some memory benchmarks. Here they are...still shaking head at the extreme disparity:

Pogoplug E02

root@pogoplug:~# sysbench --test=memory --memory-block-size=512K --memory-total-size=512M run
sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1

Doing memory operations speed test
Memory block size: 512K

Memory transfer size: 512M

Memory operations type: write
Memory scope type: global
Threads started!
Done.

Operations performed: 1024 (  723.00 ops/sec)

512.00 MB transferred (361.50 MB/sec)


Test execution summary:
    total time:                          1.4163s
    total number of events:              1024
    total time taken by event execution: 1.4137
    per-request statistics:
         min:                                  1.38ms
         avg:                                  1.38ms
         max:                                  1.62ms
         approx.  95 percentile:               1.38ms

Threads fairness:
    events (avg/stddev):           1024.0000/0.00
    execution time (avg/stddev):   1.4137/0.00

Pogoplug Pro

root@debian:~# sysbench --test=memory --memory-block-size=512K --memory-total-size=512M run
sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 1

Doing memory operations speed test
Memory block size: 512K

Memory transfer size: 512M

Memory operations type: write
Memory scope type: global
Threads started!
Done.

Operations performed: 1024 ( 2086.90 ops/sec)

512.00 MB transferred (1043.45 MB/sec)


Test execution summary:
    total time:                          0.4907s
    total number of events:              1024
    total time taken by event execution: 0.4863
    per-request statistics:
         min:                                  0.46ms
         avg:                                  0.47ms
         max:                                  1.46ms
         approx.  95 percentile:               0.48ms

Threads fairness:
    events (avg/stddev):           1024.0000/0.00
    execution time (avg/stddev):   0.4863/0.00

Pro vs. E02 - Some Benchmarks

LeggoMyEggo — Thu, 19 Mar 2015 18:15:35 -0500

I picked up a bricked E02 on eBay and was able to fix it with serial console. I thought it the appropriate time to compare the performance with my Pro. Here are some benchmarks for both devices using a Sandisk 16GB Ultra Fit 3.0 Flash Drive on both and then a 320gb 7200rpm Seagate 2.5 drive in Esata enclosure for the Pro:

Pogoplug E02 USB with latest u-boot and 3.17.0 rootfs:

root@pogoplug:~# uname -a
Linux 3.17.0-kirkwood-tld-1 #4 PREEMPT Wed Oct 22 00:55:17 PDT 2014 armv5tel GNU/Linux

root@pogoplug:~# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   528 MB in  2.00 seconds = 263.77 MB/sec
 Timing buffered disk reads:  86 MB in  3.07 seconds =  28.02 MB/sec

root@pogoplug:~# sysbench --test=cpu --cpu-max-prime=2000 --num-threads=2 run
sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 2

Doing CPU performance benchmark

Threads started!
Done.

Maximum prime number checked in CPU test: 2000


Test execution summary:
    total time:                          102.3354s
    total number of events:              10000
    total time taken by event execution: 204.6112
    per-request statistics:
         min:                                 10.22ms
         avg:                                 20.46ms
         max:                                 40.25ms
         approx.  95 percentile:              20.30ms

Threads fairness:
    events (avg/stddev):           5000.0000/0.00
    execution time (avg/stddev):   102.3056/0.02

Pogoplug Pro @850Mhz USB with latest u-boot and 3.18.5 rootfs:

root@debian:~# uname –a 
Linux debian 3.18.5-oxnas-tld-1 #2 SMP PREEMPT Thu Feb 12 11:17:40 PST 2015 armv6l GNU/Linux

root@debian:~# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   506 MB in  2.00 seconds = 252.62 MB/sec
 Timing buffered disk reads:  84 MB in  3.06 seconds =  27.46 MB/sec

root@debian:~# sysbench --test=cpu --cpu-max-prime=2000 --num-threads=2 run
sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 2

Doing CPU performance benchmark

Threads started!
Done.

Maximum prime number checked in CPU test: 2000


Test execution summary:
    total time:                          59.7179s
    total number of events:              10000
    total time taken by event execution: 119.4018
    per-request statistics:
         min:                                 11.70ms
         avg:                                 11.94ms
         max:                                 44.25ms
         approx.  95 percentile:              12.46ms

Threads fairness:
    events (avg/stddev):           5000.0000/29.00
    execution time (avg/stddev):   59.7009/0.00

.
Pogoplug Pro @ 850Mhz SATA with latest u-boot and 3.18.5 rootfs:

root@debian:~# uname -a
Linux debian 3.18.5-oxnas-tld-1 #2 SMP PREEMPT Thu Feb 12 11:17:40 PST 2015 armv6l GNU/Linux

root@debian:~# hdparm -tT /dev/sda

/dev/sda:
 Timing cached reads:   512 MB in  2.01 seconds = 255.36 MB/sec
 Timing buffered disk reads: 266 MB in  3.00 seconds =  88.54 MB/sec

root@debian:~# sysbench --test=cpu --cpu-max-prime=2000 --num-threads=2 run
sysbench 0.4.12:  multi-threaded system evaluation benchmark

Running the test with following options:
Number of threads: 2

Doing CPU performance benchmark

Threads started!
Done.

Maximum prime number checked in CPU test: 2000


Test execution summary:
    total time:                          59.7617s
    total number of events:              10000
    total time taken by event execution: 119.4815
    per-request statistics:
         min:                                 11.70ms
         avg:                                 11.95ms
         max:                                 57.55ms
         approx.  95 percentile:              12.46ms

Threads fairness:
    events (avg/stddev):           5000.0000/32.00
    execution time (avg/stddev):   59.7407/0.01

I am not sure why the E02 is not doing as well as the Pro. I thought even overclocked the Pro would barely manage to reach an E02's sysbench . Wrong assumption? Maybe I am doing something wrong? Comments appreciated.

Also, if you wanna compare sysbench numbers to some more modern SOC's, look here: http://wiki.glidernet.org/cpu-boards