Difference between revisions of "RaspberryPiPerformance"

From eLinux.org
Jump to: navigation, search
(GPU)
(Transferred to RPi Performance, intentionally blanking and redirecting)
 
(3 intermediate revisions by 2 users not shown)
Line 1: Line 1:
=CPU=
+
#REDIRECT [[RPi Performance]]
==Linpack==
 
 
 
The Arm has been tested using the linpack benchmark from [http://www.netlib.org/benchmark/linpackc.new], built with gcc with -O3 (Optimisation level 3). Run with array size 200.
 
 
 
With software floating point
 
 
 
===Source===
 
[http://www.netlib.org/benchmark/linpackc.new]
 
 
 
===Compile/Run===
 
<pre>
 
cc -O3 -o linpack linpack.c -lm
 
  linpack.c: In function ‘main’:
 
  linpack.c:69: warning: return type of ‘main’ is not ‘int’
 
./linpack
 
  Enter array size (q to quit) [200]: 200
 
</pre>
 
 
 
 
 
===Results===
 
Crippled
 
<pre>
 
Memory required:  315K.
 
 
 
LINPACK benchmark, Double precision.
 
Machine precision:  15 digits.
 
Array size 200 X 200.
 
Average rolled and unrolled performance:
 
 
 
    Reps Time(s) DGEFA  DGESL  OVERHEAD    KFLOPS
 
      2  0.53  92.45%  1.89%  5.66%  5493.333
 
      4  1.07  92.52%  2.80%  4.67%  5385.621
 
      8  2.12  92.45%  2.36%  5.19%  5466.003
 
      16  4.24  92.45%  2.83%  4.72%  5438.944
 
      32  8.49  92.11%  2.71%  5.18%  5459.213
 
      64  16.98  92.05%  2.89%  5.06%  5452.440
 
</pre>
 
 
 
Hardware floating point (-mfloat-abi=softfp)
 
<pre>
 
Memory required:  315K.
 
LINPACK benchmark, Double precision.
 
Machine precision:  15 digits.
 
Array size 200 X 200.
 
Average rolled and unrolled performance:
 
 
 
    Reps Time(s) DGEFA  DGESL  OVERHEAD    KFLOPS
 
      8  0.51  90.20%  3.92%  5.88%  22888.889
 
      16  1.02  89.22%  4.90%  5.88%  22888.889
 
      32  2.05  90.24%  3.41%  6.34%  22888.889
 
      64  4.08  91.42%  2.94%  5.64%  22829.437
 
    128  8.16  91.54%  2.94%  5.51%  22799.827
 
    256  16.31  91.35%  2.76%  5.89%  22903.800
 
</pre>
 
 
 
==Whetstone/Dhrystone==
 
 
 
All code compiled with gcc options -float-abi=softfp -O3
 
 
 
===Source===
 
Code for these tests can be found here http://www.rowley.co.uk/arm/whet_dhry.zip.
 
Or if 404 this code might be analogous http://freespace.virgin.net/roy.longbottom/benchnt.zip
 
 
 
 
 
===Compile/Run===
 
<pre>
 
?
 
</pre>
 
 
 
 
 
===Results===
 
Dhrystone
 
<pre>
 
Microseconds for one run through Dhrystone: 1.2
 
 
 
Dhrystones per Second: 809061.5
 
</pre>
 
 
 
 
 
Whetstone Crippled
 
<pre>
 
Loops: 1000, Iterations: 10, Duration: 24 sec.
 
 
 
C Converted Double Precision Whetstones: 41.7 MIPS
 
</pre>
 
 
 
Rebuilding the Whetstone test code with 'gcc -mfpu -float-abi=softfp' gives better results:
 
<pre>
 
 
 
Loops: 1000, Iterations: 100, Duration: 106 sec.
 
C Converted Double Precision Whetstones: 94.3 MIPS
 
</pre>
 
 
 
However, the majority of compute time is spent in the SQRT function, which for the above test was built without -mfpu=vfp. Using a library with vfp give the following much improved result :
 
<pre>
 
Loops: 1000, Iterations: 100, Duration: 15 sec.
 
C Converted Double Precision Whetstones: 666.7 MIPS
 
</pre>
 
 
 
==OpenSSL==
 
 
 
===Source===
 
[http://www.openssl.org/source/]
 
 
 
===Compile/Run===
 
<pre>
 
openssl version;
 
openssl speed;
 
</pre>
 
 
 
===Results===
 
<pre>
 
OpenSSL 0.9.8o 01 Jun 2010
 
built on: Thu Aug 26 18:56:26 UTC 2010
 
options:bn(64,32) md2(int) rc4(ptr,int) des(idx,risc1,4,long) aes(partial) blowfish(idx)
 
compiler: gcc -fPIC -DOPENSSL_PIC -DZLIB -DOPENSSL_THREADS -D_REENTRANT -DDSO_DLFCN -DHAVE_DLFCN_H -DL_ENDIAN -DTERMIO -O2 -Wa,--noexecstack -g -Wall
 
available timing options: TIMES TIMEB HZ=100 [sysconf value]
 
timing function used: times
 
The 'numbers' are in 1000s of bytes per second processed.
 
type            16 bytes    64 bytes    256 bytes  1024 bytes  8192 bytes
 
md2                148.81k      372.18k      624.81k      769.95k      832.90k
 
mdc2                0.00        0.00        0.00        0.00        0.00
 
md4                615.30k    2468.76k    7612.19k    16707.01k    28104.86k
 
md5                380.13k    1501.12k    4800.77k    11312.81k    21682.77k
 
hmac(md5)        1022.28k    3480.23k    9587.80k    17492.25k    25441.78k
 
sha1              303.72k    1092.39k    3106.50k    6302.57k    9852.39k
 
rmd160            244.29k      849.04k    2414.53k    4747.26k    7513.00k
 
rc4              14658.70k    16836.49k    17462.03k    17628.21k    17522.08k
 
des cbc          2913.17k    3221.30k    3289.77k    3360.09k    3367.21k
 
des ede3          1149.87k    1188.59k    1198.46k    1206.00k    1208.25k
 
idea cbc            0.00        0.00        0.00        0.00        0.00
 
seed cbc            0.00        0.00        0.00        0.00        0.00
 
rc2 cbc          2812.71k    3012.02k    3054.19k    3077.82k    3076.12k
 
rc5-32/12 cbc        0.00        0.00        0.00        0.00        0.00
 
blowfish cbc      6091.32k    7007.89k    7250.62k    7288.21k    7163.88k
 
cast cbc          5068.25k    6020.03k    6345.71k    6367.64k    6260.44k
 
aes-128 cbc      3205.76k    3497.72k    3616.00k    3652.49k    3665.85k
 
aes-192 cbc      2730.65k    2981.88k    3073.20k    3102.38k    3111.86k
 
aes-256 cbc      2383.90k    2596.12k    2659.91k    2702.13k    2732.50k
 
camellia-128 cbc    0.00        0.00        0.00        0.00        0.00
 
camellia-192 cbc    0.00        0.00        0.00        0.00        0.00
 
camellia-256 cbc    0.00        0.00        0.00        0.00        0.00
 
sha256            679.98k    1629.47k    2905.43k    3708.32k    4175.45k
 
sha512              41.02k      163.83k      232.63k      318.20k      353.81k
 
aes-128 ige      3089.03k    3579.08k    3698.68k    3689.14k    3578.18k
 
aes-192 ige      2641.68k    3019.45k    3111.38k    3144.95k    3035.70k
 
aes-256 ige      2334.50k    2632.35k    2705.04k    2735.69k    2687.74k
 
                  sign    verify    sign/s verify/s
 
rsa  512 bits 0.013747s 0.001193s    72.7    838.4
 
rsa 1024 bits 0.063481s 0.002742s    15.8    364.7
 
rsa 2048 bits 0.321250s 0.007378s      3.1    135.5
 
rsa 4096 bits 1.805000s 0.022528s      0.6    44.4
 
                  sign    verify    sign/s verify/s
 
dsa  512 bits 0.011690s 0.013597s    85.5    73.5
 
dsa 1024 bits 0.027233s 0.031683s    36.7    31.6
 
dsa 2048 bits 0.073897s 0.087304s    13.5    11.5
 
</pre>
 
 
 
=GPU=
 
The RaspberryPi appears to handle h264 1080p movie from USB to HDMI at least 4MB/s.
 
The Admin "JamesH" said it would handle "basically 1080p30, high profile, >40Mb/s."
 
 
 
 
 
 
 
==3DMarkMobile ES 2.0==
 
 
 
===Source===
 
?
 
 
 
===Compile/Run===
 
<pre>
 
?
 
</pre>
 
 
 
===Results===
 
<pre>
 
?
 
</pre>
 
 
 
[[Category: RaspberryPi]]
 
 
 
=IO=
 
 
 
==USB buss==
 
*All IO uses the same bus so the combination of all IO can not exceed the the bus speed of an as yet hypothetical 60MB/s
 
==SD card==
 
*TODO test
 
===Compile/Run===
 
<pre>
 
dd if=/dev/zero of=~/test.tmp bs=100K count=1024
 
dd if~/test.tmp of=/dev/null bs=100K count=1024
 
rm ~/test.tmp
 
</pre>
 
===Results===
 
* Depends on SD card used http://elinux.org/RaspberryPiBoardVerifiedPeripherals#SDHC_cards
 
<pre>
 
?maybe 15MB/s?
 
</pre>
 
 
 
==NIC==
 
*TODO test with wget, curl, etc
 

Latest revision as of 09:08, 28 January 2012

Redirect to: