search

open62541 / open62541

Open source implementation of OPC UA (OPC Unified Architecture) aka IEC 62541 licensed under Mozilla Public License v2.0
http://open62541.org
Mozilla Public License 2.0
2.55k stars 1.23k forks source link

Poster at AUTOMATION 14 #78

Closed Stasik0 closed 10 years ago

Stasik0 commented 10 years ago

Hello folks,

I'd like to discuss a structure of a poster on our AUTOMATION presentation, I imagine just a big bullet-list with an "open62541" on top and 5-6 chunky facts.

Following is needed:

Best Regards

Stasik0 commented 10 years ago

i have created a logo with my mad skills. Should be a small homage to the original opc logo

open62541

uleon commented 10 years ago

I had a long train ride from Frankfurt to Dresden and got the namespace_ generation-branch up and running. There is still some refactoring necessary, however I'd like to add the following current facts for our stack:

sudo apt-get install sysbench
sysbench --test=cpu --cpu-max-prime=20000 run
bin/xml2ns0 < examples/src/Opc.Ua.NodeSet2.Part5.xml

Can we claim this performance already as high speed? Does anyone have figures on the performance of other stacks?

Three other highlights I'd like to stress in respect to the VDI/VDE 2657 Blatt 1 Middleware in der Automatisierungstechnik - Grundlagen

Eventually

Stasik0 commented 10 years ago

@uleon I am getting "avg: 3.48ms" from sysbench and "encode server node 100000 times: time/enc=2.900560 ns" what does it mean? 3 nanoseconds per encode/decode? nanoseconds?

uleon commented 10 years ago

well, it's actually µ-sec, sorry for the typo. -lu

Am 23.05.2014 17:06, schrieb Sten:

@uleon https://github.com/uleon I am getting "avg: 3.48ms" from sysbench and "encode server node 100000 times: time/enc=2.900560 ns" what does it mean? 3 nanoseconds?

— Reply to this email directly or view it on GitHub https://github.com/acplt/open62541/issues/78#issuecomment-44022391.

FlorianPalm commented 10 years ago

these are my results: min:1.86 us avg:2.65 us max: 11.11us

Stasik0 commented 10 years ago

Okay, it seems to corelate well with sysbench... 3ms means 330000 requests per second which is... impressive!

p.s. cranking optimization up from -O2 to -O3 gets me to something about 2.2-2.5us

Am 23.05.2014 um 17:12 schrieb uleon notifications@github.com:

well, it's actually µ-sec, sorry for the typo. -lu

Am 23.05.2014 17:06, schrieb Sten:

@uleon https://github.com/uleon I am getting "avg: 3.48ms" from sysbench and "encode server node 100000 times: time/enc=2.900560 ns" what does it mean? 3 nanoseconds?

— Reply to this email directly or view it on GitHub https://github.com/acplt/open62541/issues/78#issuecomment-44022391.

— Reply to this email directly or view it on GitHub.

uleon commented 10 years ago

I'm most interested in the number of complete indication/response cycles/second. I started to add a fixture to test_stack that would allow for such measurements.

Response-encoding and indication-decoding are almost symmetric, the main difference is the dynamic memory allocation in the decoding path - I'm already curious to learn about the performance cost, my guess would be that we will see three times less complete Ind/resp-transactions.

-lu

On 23. Mai 2014 18:42:05 MESZ, Sten notifications@github.com wrote:

Okay, it seems to corelate well with sysbench... 3ms means 300000 requests per second which is... impessive!

Am 23.05.2014 um 17:12 schrieb uleon notifications@github.com:

well, it's actually µ-sec, sorry for the typo. -lu

Am 23.05.2014 17:06, schrieb Sten:

@uleon https://github.com/uleon I am getting "avg: 3.48ms" from sysbench and "encode server node 100000 times: time/enc=2.900560 ns" what does it mean? 3 nanoseconds?

— Reply to this email directly or view it on GitHub

https://github.com/acplt/open62541/issues/78#issuecomment-44022391.

— Reply to this email directly or view it on GitHub.

Reply to this email directly or view it on GitHub: https://github.com/acplt/open62541/issues/78#issuecomment-44033813

Prof. L. Urbas; TU Dresden; +49-177-4665201

jpfr commented 10 years ago

It won't come out before the Automation. But this would be some cool demo hardware: https://www.indiegogo.com/projects/vocore-a-coin-sized-linux-computer-with-wifi Embedded Linux system smaller than a Euro coin plus wifi for <= $20.

Stasik0 commented 10 years ago

@jpfr did you order one?

jpfr commented 10 years ago

Just did..

Stasik0 commented 10 years ago

okay, here is a first try - the logos of the universities are missing the sources for the poster are in the komma14 repository - inkscape was used - edit poster.png please make sure to discuss it at Wednesday - we need the poster layout on 1st of June poster

HolgerJeromin commented 10 years ago

There is a typo "decoidng".

floh27 commented 10 years ago

Hi everyone, Leon asked me to have a look a the poster, put in our logos and maybe help out with some advices. Can someone grant me access to the komma14 repository?

Best, Florian

Stasik0 commented 10 years ago

@floh27 done, https://github.com/Stasik0/komma14

floh27 commented 10 years ago

Here the new suggestion for the poster from TUD: poster_open62541_komma2014_a0 Please let me know if there is something to chance as soon as possible.

Stasik0 commented 10 years ago

@floh27 much better! would you please commit the changes: (I know, these are my mistakes)

uleon commented 10 years ago

https://cloud.githubusercontent.com/assets/7710389/3093514/53adb28c-e5b3-11e3-8bbb-59afc3199f5b.png

My 5 cent:

logo can we have a logo that really "opens up", that is to have the text starting with a negative spacing (kernel?), that is a slight overlap of o and p which would then get wider and wider for every additional character?

Then of course I'm a great fan of ascii art:

** _ _ _ _ __** _ _' _

Ah, well, no - forget about it. It was good karma that I fell in love with computational engineering science.

... OPC UA[,] based I'd put a comma there, don't know however why - just a (perhaps wrong) feeling.

[-REALLY-] OPEN I'd prefer to have simply OPEN. Isn't it really PORTABLE, EXTENSIBLE, etc. as well?

  • stric[-k-]tly based ...

PORTABLE I'd like to suggest to separate the topics PORTABLE and EXTENSIBLE

  • pure C99, POSIX support
  • Android, Linux i586, Raspberry Pi, Sparc Station, Windows

Sparc Station: Chris promised to set up a big endian SparcStation for me

Android: I'm currently working on setting up a cross compiler tool chain to implement an open62541 server for my rooted nexus. That would allow us to run our mobiles as ubiquitous sensor platforms @ Automation2014 - wouldn't that be fancy?

EXTENSIBLE

  • seamless integration of pre-compiled user models for maximum performance

MAINTAINABLE actually 85% of the complete code base is generated if my figures are correct.

SCALABLE

  • multi-threaded architecture I'm not sure if we're really multi-core ready at the time being.

// -- not yet but soon to come ? // > FLEXIBLE // > - dynamically loadable and reconfigurable user models

initiated by: I'd love to have our brand new PLT-logo here - color would be fine ;-)

Stasik0 commented 10 years ago
jpfr commented 10 years ago

I'd throw another target architecture into the ring: javascript

Emscripten (http://emscripten.org) piggybacks the compiler and replaces the last step, bytecode->native, with bytecode->js (with shims for system functionality like io). They show an impressive demo on their page, where the latest unreal engine runs at 1/2 native speed in the browser.

./configure CC=emcc was working out of the box on our sources some time ago. With the external libraries such as expat, we might have to do something extra today. So javascript as a portable (and more or less attack-safe) target architecture (think node.js for servers and browsers for clients) comes basically for free.

uleon commented 10 years ago

@Stasik0 I've got the point and I would use such an emphasis in direct speech of course as well, e.g. "the stack is OPEN - and when I say OPEN, I REALLY mean OPEN - no hidden royalties, no covert third party IP, JUST OPEN - ready to analyze, investigate, improve and share".

Stasik0 commented 10 years ago

@uleon agreed!

jpfr commented 10 years ago

Can we just say open source (LGPL) with no restrictions on commercial use? That should drop enough jaws without coming across as aggressive.

uleon commented 10 years ago

:+1: for LPGL

Stasik0 commented 10 years ago

+1 for Lgpl+static linking

Am 27.05.2014 um 21:17 schrieb Julius Pfrommer notifications@github.com:

Can we just say open source (LGPL) with no restrictions on commercial use? That should drop enough jaws without coming across as aggressive.

— Reply to this email directly or view it on GitHub.

uleon commented 10 years ago

I'm quite sure that I did not grasp the static linking term in it's fulllness. LGPL would make it possible that I utilize LGPL-parts in my commercial application while keeping my proprietary part of the application private. Static linking then means - if I got it right - that I need to provide ways to link the proprietary and the LGPL-part. So what does this help?

Prof. L. Urbas; TU Dresden; +49-177-4665201

Stasik0 commented 10 years ago

@uleon Using LGPLed .dll or .so within commertial projects is not a problem at all

Now lets consider using static linkage: combining the private part and the LGPLed .a into a one executable. LGPL says that than you need to supply all the .o object files of your private part together with the binaries, such that everybody can relink the executable with, e.g. a newer version of the LGPLed library

This distribution obligation of the object files scares many commertial users. Therefore LGPL is often modified with the "static linking exception" that lifts this obligation.

Am 27.05.2014 um 22:41 schrieb uleon notifications@github.com:

I'm quite sure that I did not grasp the static linking term in it's fulllness. LGPL would make it possible that I utilize LGPL-parts in my commercial application while keeping my proprietary part of the application private. Static linking then means - if I got it right - that I need to provide ways to link the proprietary and the LGPL-part. So what does this help?

Prof. L. Urbas; TU Dresden; +49-177-4665201 — Reply to this email directly or view it on GitHub.

HolgerJeromin commented 10 years ago

Do we want the university or chair logos? The RWTH logo right now is not conform to the corporate design (as we have no dark background). I just updated the logo on wikipedia (which was probably your source :-) https://commons.wikimedia.org/wiki/File:RWTH_Logo_3.svg

uleon commented 10 years ago

Why not both university and chair?

HolgerJeromin commented 10 years ago

Then please use one of this Images (a RWTH + PLT combination in portrait form is not CD conform :-) rwth_acplt_de_rgb rwth_acplt_de_schwarz_grau_rgb

Stasik0 commented 10 years ago

@HolgerJeromin I'd go for the english logo rwth_acplt_en_rgb i have pushed our logos in rgb and cymk to the repo

floh27 commented 10 years ago

Hi, I tried to get all mentioned facts from the last post and find a suitable solution:

poster_open62541_komma2014_a0

Due to very limited time today - and vacation until Monday - there is no new logo. I shrinked the space between o and p, which was the original kerning form Illustrator.

The use of chair logos is very complicated in my eyes. First, each university has its own Corporate Identity which describes the usage of university logo and other (chair) logos => we won't find a useable solution for all three style guides. Second, there is no usebale (English) vector logo from IES yet. Third, from a marketing perspective it is much more impressive when you see that this project from that 3 engineering universities. As a conclusion I put only the (English) university logo on the poster. If you decide in the call to change it or modify text, feel free.

Stasik0 commented 10 years ago

okay, the deadlines are expired ;) I have replaced KIT by IOSB as wished by @jpfr

Final survey. Does someone have any further problems with the poster? If no reply comes until today evening, I will order this version: poster_open62541_komma2014_a0

jpfr commented 10 years ago

We discussed some changes at the end of the conference call last week. For example, the format of the printed canvas is more narrow.

@FlorianPalm I think we made notes in the google docs. Can you short-circuit this in Aachen?

HolgerJeromin commented 10 years ago

Quite late, but the Portable section is quite random. Android is an OS. Linux i586 (is an OS with an processor family from one vendor) Raspberry pi is a hardware (which OS is running?) Sparc Station is a hardware (which OS is running?) Windows is an OS (family).

Suggestion (with correct spelling for SPARCstation):

or as i am not sure about the sparcstation

But dropping the Raspberry is sad, yes:

Stasik0 commented 10 years ago

@HolgerJeromin yea, I thought about it too... I vote for:

@floh27 I do not have an Adobe Illustrator here, could you fix it in? further fixes from the conference on Wednesday coming in from @FlorianPalm:

floh27 commented 10 years ago

@Stasik0 I'll update the poster till 15:00

floh27 commented 10 years ago

For example, the format of the printed canvas is more narrow. @jpfr what size in the end?

the logo misses the vertical line on the bottom of the "O" is it intended? or is it a bug? @Stasik0 it is intended to be more dynamic

Stasik0 commented 10 years ago

@floh27 no idea what about the size, I need to check this with the printing service, I think it is something like 80cm x 200cm and should be CYMK

floh27 commented 10 years ago

@Stasik0 For "something like 80x200cm" we need to change the whole layout and it is important to have the exact size from the printing service. Does the printing service have a website, where I can search for it?

The pdf which you should send to the printing service is and will be CMYK.

Stasik0 commented 10 years ago

@floh27, yea

http://www.wir-machen-druck.de/tpl/manns-partner/media/ddb/druckdatenskizzen/premium_roll-up_bannerdisplay_85_cm_x_200_cm__inklusive_druck_und_versand_1.pdf http://www.wir-machen-druck.de/tpl/manns-partner/media/ddb/druckdatenskizzen/premium_roll-up_bannerdisplay_85_cm_x_200_cm__inklusive_druck_und_versand_3.ai

however, the banner must not be 2 meter tall

floh27 commented 10 years ago

however, the banner must not be 2 meter tall

thanks, I'm quite sure the banner has to be 2m tall, the content can be smaller.

floh27 commented 10 years ago

@Stasik0 Changes necessary??

poster_open62541_komma2014_rollup

floh27 commented 10 years ago

@Stasik0 use for printing: Poster_Open62541_Komma2014_RollUp.pdf

Stasik0 commented 10 years ago

@uleon, @jpfr, @FlorianPalm, @HolgerJeromin is everyone okay with that layout?

FlorianPalm commented 10 years ago

royality free,first principles stack design, based on 65241 <<< TYPO 62541 is correct

Stasik0 commented 10 years ago

@FlorianPalm nice, the epic-fail typo eliminated!

floh27 commented 10 years ago

62541 is correct

done, anything else?

@uleon suggested to split scalable and extensible, ok for everyone?

uleon commented 10 years ago

is it possible to have a little bit more even vertical spacing between the icons with reasonable efforts?

floh27 commented 10 years ago

@uleon why vertical? they belong together = "Gesetz der Nähe" ;)

I'd prefer to have SCALABLE and EXTENSIBLE with the cpu icon, so that we have little more whitespace

floh27 commented 10 years ago

@Stasik0 when do you have to send it to the printing service??