Project

General

Profile

Wt embedded » History » Version 13

Koen Deforche, 05/27/2010 07:08 AM

1 1 Pieter Libin
h1. Wt embedded
2
3
{{toc}}
4
5
Find here information on running Wt in resource constrained embedded systems: performance, code size, memory usage, and other info.
6
7
h2. General
8
9
Wt can easily be built for and deployed on embedded POSIX systems, such as embedded linux.
10
11
h3. Cross-building
12
13
Using CMake with a cross compilation environment: to be completed...
14
15 8 Koen Deforche
Instructions for cross compiling with cmake can be found on the "CMake Wiki":http://www.cmake.org/Wiki/CMake_Cross_Compiling.
16 1 Pieter Libin
17
h3. Optimizing executable size
18
19 8 Koen Deforche
Points to consider when optimizing the executable size.
20 1 Pieter Libin
21 8 Koen Deforche
For building boost:
22
* Use static build of boost, which allows the linker to strip away unused symbols
23
* Use the following compile flags for boost:
24
** @-fvisibility=hidden -fvisibility-inlines-hidden@: to avoid exporting symbols in the executable
25
** @-ffunction-sections -fdata-sections@: to allowing fine-grained garbage collection of unused functions/data
26
27
For building Wt:
28
* Choose build-type @MinSizeRel@
29
* Extra compile falgs (@CMAKE_CXX_FLAGS@)
30
** @-fvisibility=hidden -fvisibility-inlines-hidden@: to avoid exporting symbols in the executable
31
** @-ffunction-sections -fdata-sections@: to allowing finegrained garbage collection of unused functions/data
32
** @-DHAVE_GNU_REGEX@: to avoid the dependency on libboost_regex, when building on a system that is based on glibc or uClibc
33
** @-DWT_NO_LAYOUT@: to avoid pulling in the Wt's layout managers, if you are not using any WLayout classes
34
** @-WT_NO_SPIRIT@: to avoid depending on spirit to parse locale and cookies (if you don't need that)
35
** @-DWT_NO_XSS_FILTER@: to avoid the extra (runtime) overhead of XSS filtering, usually not relevant for a trusted embedded platform
36 1 Pieter Libin
* Build static libraries (for libwt.a and libwthttp.a)
37 8 Koen Deforche
** in CMake: @SHARED_LIBS:BOOL=OFF@
38 1 Pieter Libin
* Disable build options you don't need and introduce extra dependencies (libz, openssl ?)
39 8 Koen Deforche
* Further tune your linker command:
40
** Append @-v@ to the linker command used by CMake to see the raw @collect2@ command-line.
41
** By default, shared/static libraries is all-or-nothing with CMake. However, you probably want to use system-wide versions of libstdc++, libm and libc depending on other applications on your device.
42
*** Use -Bdynamic in front of libraries you wish to link dynamically against
43
** There are some other flags that you need to use to make sure the linker does not keep unused symbols:
44
*** Remove @-export-dynamic@
45
*** Add @--gc-sections@
46 9 Koen Deforche
* Strip your binary using @strip -s@.
47 1 Pieter Libin
* Optionally, when available for your platform, you may want to compress the size of your binary using the "Ultimate Packer for eXecutables (upx)":http://upx.sourceforge.net/. This typically reduces executable size further by 60-70%, without noticable run-time performance hits.
48
49
h3. Measuring performance
50
51
To report the run-time performance of Wt on a particular embedded platform, you must connect to the device using a local area connection (through at most one switch), and measure the time between transmission and reception of packets (using a packet sniffer). For the measurements, we use two examples that are included in the Wt distribution: "hello":http://www.webtoolkit.eu/wt/examples/hello/hello.wt (as an example of a minimal application), and "composer":http://www.webtoolkit.eu/wt/examples/composer/composer.wt (as an example of a simple, yet functional, application).
52
53
We propose to measure the time to create a new session, and the time of a small event.
54
55
56
h4. Runtime: new session
57
58 13 Koen Deforche
Wt starts a new session by serving a small page to determines browser capabilities, and then trigger a second call to get the "main page", that has all visible content. To compare the relative performance for a particular platform, you should measure this "load" time, as the total duration of these two requests. You should measure the time from sending the first request, to sending the third request. The third request is either a GET request for auxiliary content (CSS or images), a GET request to a Wt resource, or a POST request to load invisible content in the background.
59 1 Pieter Libin
60
61
h4. Runtime: event
62
63
We estimate the time needed to process a small event, such as a click on the "Greet me" button in hello, and "Save now" in composer, by measuring the total time for the packet exchange triggered by such an event.
64
65
66
h4. Memory usage: basis
67
68
Measuring memory usage is a tricky thing, since code and read-only data memory used by shared libraries is effectively shared between processes, while writable data segments are obviously private to each process.
69
70 13 Koen Deforche
Therefore, we use @pmap@ to study the memory in different segments. The basis RAM usage is divided between read-only segments, and writable segments. Only the latter are really constrained by physical RAM. We get the total writable size. by summing the size of all writable segments, indicated by pmap with a *w*. The total size reported by pmap and top, minus the size of all writable segments is then the read-only RAM usage. Thus, this number includes shared libraries, and thus overestimates actual RAM usage.
71 1 Pieter Libin
72
h4. Memory usage: per session
73
74
Compare the memory usage after starting 10 sessions with base memory usage, and divide the difference by 10 to estimate the memory used by a single session.
75
76
h2. Platforms
77
78
h3. ARM926EJ-S
79
80
h4. Processor features
81
82
* Clock-speed: 200 MHz
83
* Linux BogoMIPS: 89.70
84
* Caches: 8K instruction, 8K data
85
86 8 Koen Deforche
Configurations are ordered chronically, latest first.
87 1 Pieter Libin
88 8 Koen Deforche
h4. Config 2: minimal (16/03/2010)
89 1 Pieter Libin
90 8 Koen Deforche
h5. Setup
91
92
* *Wt version:* git (16/03/2010, >= Wt 3.1.1)
93
* *Target system:* Linux uclibc 2.6.23
94
* *Build environment:* buildroot, arm-linux-gcc 4.2.1
95
* *Options:* without multi-threading, libz and OpenSSL
96 11 Koen Deforche
* *Build type:* full static build, except for: libstdc++, libc, and libm
97 8 Koen Deforche
* *Runtime settings:* ./app.wt --docroot . --http-address 0.0.0.0 --no-compression
98
99
h5. Performance results
100
101
*Code size and RAM usage (in KBytes)*
102
|_.Program|_.Code size (strip)|_.Code size (strip + upx)|_.RAM: basis † (read-only)|_.RAM: basis (writable)|_.RAM: per session|
103 9 Koen Deforche
| hello| 1.214  | 362 | 2.544 | 228 | 14.8 |
104
| composer| 1.462  | 420 | 2.796 | 232 | 83.6 |
105 8 Koen Deforche
106
† includes shared libraries !
107
108
*Runtime-performance*
109
|_.Program  |_.New session (http) |_.Event (http)|
110 9 Koen Deforche
| hello | 0.26 s | 0.07 s |
111 8 Koen Deforche
|composer| 0.69 s | 0.08 s |
112
113
h4. Config 1: minimal (18/03/2008)
114 1 Pieter Libin
115
h5. Setup
116
117
* *Wt version:* CVS-snapshot 18/03/08
118
* *Target system:* Linux uclibc 2.6.23
119
* *Build environment:* buildroot, arm-linux-gcc 4.2.1
120
* *Options:* with multi-threading, but without libz and OpenSSL
121
* *Build type:* full static build, except for: libc, libpthread, libdl, libstdc++, and libm
122
* *Build settings:* MinSizeRel, -DHAVE_GNU_REGEX
123
* *Runtime settings:* ./app.wt --docroot . --http-address 0.0.0.0 --threads=2 --no-compression
124
125
h5. Performance results
126 3 Pieter Libin
127 4 Pieter Libin
*Code size and RAM usage (in KBytes)*
128 6 Pieter Libin
|_.Program|_.Code size (strip)|_.Code size (strip + upx)|_.RAM: basis † (read-only)|_.RAM: basis (writable)|_.RAM: per session|
129 5 Pieter Libin
| hello| 1.130  | 304 | 2.580 | 372 | 28|
130
| composer| 1.265  | 332 | 2.712 | 372 | 126|
131 1 Pieter Libin
132 3 Pieter Libin
† includes shared libraries !
133 1 Pieter Libin
134 5 Pieter Libin
*Runtime-performance*
135 7 Pieter Libin
|_.Program  |_.New session (http) |_.Event (http)|
136 5 Pieter Libin
| hello | 0.58 s | 0.15 s |
137
|composer| 1.8 s | 0.15 s |