Linux Motion Video I/O
[Most Recent Entries]
[Calendar View]
[Friends]
Below are the 20 most recent journal entries recorded in
vlnv's LiveJournal:
[ << Previous 20 ]
| Saturday, March 8th, 2008 | | 10:46 pm |
RMS MOSCOW MARCH 2008 we will place images, video, audio materials and links for same here Счастье есть - его не может не быть :-) На фото мой residual self image после выброса эндорфинов, вызванного нахождением в ауре RMS.  спасибо друзьям ella@ella.su за фото | | Thursday, March 6th, 2008 | | 2:15 pm |
RMS visit debriefing / по итогам визита
Richard Stallman's (RMS) visit to Russia happened in full accordance with being at the right time, at the right place rule. The meetings with students, media, Free Software users, developers and Russian Govt. officials - went with high degree of enthusiasm, mutual compassion and internal alignment to the goals of Freedom and Progress.
Визит Ричарда Сталлмана (RMS) в Россию получился в нужное время, в нужном месте. Прошли встречи со студентами, журналистами, пользователями и разработчиками СПО, представителями Правительства РФ - все встречи были полны взаминого энтузиазми, симпатии и приверженности ценностям свободы и прогресса.
Мы выложим видео запись всех выступлений RMS в ближайшее время (8 часов).
| | Friday, February 29th, 2008 | | 1:16 pm |
к вопросу об AI
Я не верю, что на существующей базе (дискретные логические элементы/кремний) возможна реализация интеллекта. Возможно - квантовые компьютеры позволят к этому подойти. Разум и жизнь сама по себе существенно квантово-механические явления.
Разум предполагает решение большого количества NP-полных задач, которые дискретно-логические компьютеры решать не могут. Покажите мне систему для решения NP задач - тогда можно будет - может быть - думать, что на их основе можно создать AI.
Заметим, что когда было обнаружено явление радиоактивности мало кто всерьез задумывался о военных применения. По-моему сама идея создания атомной бомбы за 10 лет до ее создания всерьез никем не воспринималась в принципе. Да и компьютера за 15 лет до его создания, да и персонального компьютера за год до его создания.
А вот тема о роботах/терминаторах обсуждается в научной и особенно художественной литературе, кино просто как общее место. В массовом сознании проблем с созданием AI нет. Только вот AI за последние 50 лет, когда казалось, что вот-вот - никак не просматривается.
Может быть технология только начинает подходить к созданию искуственного инстинкта, вроде насекомых, но никак не интеллекта. | | Thursday, February 28th, 2008 | | 6:54 am |
программа визита Ричарда Сталлмана в Москву март 2008
3 марта
11:00 Посещение ИСП РАН
16:20 Лекция в МГУ, ВМК, аудитория П14, ответы на вопросы
4 марта
15:00 Визит и встреча с сотрудниками Физтех, Долгопрудный
17:05 Лекция в большой аудитории
5 марта
11:00-13 Теле и Интернет конференция и пресс-конференция CNEWs с участием LinuxCenter и А.Поносова
18:00 Лекция в Институте Философии РАН на тему "Copyright vs Community in the Age of Computer Networks" ("Авторское право и народ - противостояние в век компьютерных сетей"
ЖДЕМ ВСЕХ СТОРОННИКОВ И ЭНТУЗИАСТОВ СВОБОДНОГО ПО в первую очередь в МГУ, на ВМК,в понедельник, а также на лекции в Институте Философии в среду. | | Tuesday, February 26th, 2008 | | 6:35 pm |
schedule for RMS visit to Russia, March 2008 March 311:00 Meeting at the Institute for System Programming of RAS.
Discussions with people of the institute, topics such as GCC optimization for Itanium; GCC for power optimization; Linux verification. Hosted by Viktor Petrovich Ivannikov, a corresponding member of the Russian Academy of Sciences.16.20. Lecture at the Moscow State University, Computer Science Dept. (VMK) Google Earth view
Google Earth KMZ (location file)
March 4
15.00 Visit Moscow Institute of Physics and Technology (PhysTech). Meeting with employees and Ph.D.students.
17.05. Lecture on the topic of "Free Software - Ethics and Practice"
March 5
11.00-13.00 Internet-conference and press conference
18.00 Lecture in Institute of Philosophy of Russian Academy of Science.
| | Thursday, February 14th, 2008 | | 9:26 pm |
RMS will visit Russia early March 2008
Мы рады сообщить, что визит творца GNU, основателя FSF Ричарда Сталлмана в Россию запланирован со 2го по 6е марта 2008 г.
Программа визита сейчас уточняется.
We are glad to announce that creator of GNU, FSF founder Richard Stallman (R.M.S.) will visit Russia on March 2nd to 6th 2008.
The visit program is being finalized now. | | Monday, January 7th, 2008 | | 9:12 am |
RMS visit to Russia
I want to update all Freedom Fighters of the World that the process of arranging RMS visit to Russia in Q1 2008 is moving along into specifics. We should be able to announce exact days very soon.
Хочу сообщить Прогрессивному Человечеству - организация визита RMS продолжается и мы уже прорабатываем конкретные детали - очень скоро будут известны точные даты визита. | | Monday, October 22nd, 2007 | | 3:03 pm |
R.M.S. may visit Russia in 2008
Мы счастливы жить в одно время с человеком, оказавшим ключевое воздействие на IT индустрию - с человеком, который убедительно доказал, что и один в поле воин. Еще более нам повезло в том плане, что R.M.S. активно пропагандирует идеи Свободного ПО и ставит вопросы о моральности и правомочности современных Западных подходов к вопросам лицензирования ПО, что так актуально для России, находящейся на перепутии по этим жизненным вопросам.
Поэтому возникла идея пригласить Ричарда Сталлмана приехать в Россию в начале 2008 года. Несмотря на свою плотную программу он пообещал найти время и спланировать этот визит в Россию. Депутат гос Думы В.И.Алкснис с большим интересом отнесся к данному проекту, пообещал оказать организационно-информационную поддержку, что дает нам уверенность в успехе данного начинания.
Приглашаем к сотрудничеству всех заинтересованных в организации проведения встреч и мероприятий с участием создателя концепции и воплощения идеи СПО/GNU Ричарда Сталлмана. | | Wednesday, October 3rd, 2007 | | 1:12 am |
Long live RMS and GNU/Linux Why we should use GNU/Linux name for our favourite operating system."People willing to trade their freedom for temporary security and money deserve neither and will lose both"
And with GNU/Linux you get both better security and it costs less in terms of money. Why would then anybody trade their freedom? And in Russia, in view of criminal code article 146 "freedom" means literrally - freedom to stay out of jail. "Thank you" Microsoft and russian "liberal democrats" for that "improvement". Close affiliation with West results in more ways for a Russian person to get convicted.
With all my respect and admiration to Linus, the ultimate moral authority as well as an example that a single person can change the world for the better singlehandedly - is certainly Richard Stallman - the arch rival of bILL gATES.
| | Monday, September 10th, 2007 | | 5:17 am |
LML26413 breakthrough Had I found this earlier - a week of frustration and suffering could be spared :-) I've almost built an adapter to hook up our MG1264 carrier board to Comet-3... at least I'm lucky that you've told me about the importance of product_conf setting before I did that much micro-soldering and board cutting... LML26413 is built with 16Mb SDRAM. For no particular reason, but if it is better to use 32Mb SDRAM we'll change the BOM to reflect it - what is the benefit of 32Mb SDRAM? ======================================== =================================== As soon as I've changed product_conf to 17 => !!!!! I GOT LML26413 FULLY OPERATIONAL !!!!! Both video/audio capture and playback are DVD-like quality. WOW! It was my birthday on the 8th, so you waited till 9th for the gift :-) Thank you! God bless you and everybody who helped. Current Mood:  grateful | | Wednesday, August 15th, 2007 | | 7:58 pm |
LML26413 hardware works!
LML26413 hardware tested - PCI controller, Mobilygen MG1264 registers access, audio I/O, video I/O (digital audio/video loopback).
Now all we have to complete is QHAL integration and soft wiring of raw audio/video I/O. | | Saturday, March 17th, 2007 | | 11:22 pm |
Freedom == energy
Human being can imagine clearly only something one had already experienced in life. Therefore our limits in understanding the Universe are set by the society organization. In order to invent nuclear weapon people must live in the society complicated enough to accommodate understanding quantum physics. No matter what human does - it is anthropomorphic. Therefore the progress in understanding the Universe is accompanied with correlated complexity of the society. Society has to model the universe.
Therefore the following theorem applies - We should not be concerned about regression towards "Dark Ages"
Proof:
Since only liberal enough society would be able to produce all "power toys" that governments that wish to simplify and deprive human beings of freedom would also get reduction in their access to knowledge and technology.
US are not good on that account - there is very pronounced tendency to level everybody, but that leads to less social interaction, that leads to less depth in understanding the Universe.
So, as opposed to what supposedly Protestants believe in - that wealth is sign of Gods predisposition towards that person, I would say that it's not wealth - but maximum aggregated freedom of life for the given society structure. Freedom of the majority implies reasonable economic condition - free human being would not tolerate staying poor, therefore forcing the structure of the society to adjust accordingly.
The slavery in the original US shows that this line of reasoning about freedom did not quite prevail then. I even feel bad when I'm junking computers, since I feel that Universe becomes less complex with that one gone. Let alone turning human beings into robots.
Certainly some amount energy can be produced by reducing freedom - but that energy would soon be gone. Ultimate example - something can be gained by putting every person in the society to jail. But that less free - frozen - structure would very soon stop producing much in terms of actual goods, as well as understanding the Universe. If everybody is in jail, who would have the benefit of thinking about nuances of the way Universe functions.
Nothing in this model implies that people should have uniform say in the rule of the society (democracy), but on the other hand structures that are created in order to replace simple "democratic" organization, should at the same time improve freedom. So - the structure of the jail is not energetically viable structure. Western democracy, as long as it is not limiting freedoms is OK, but the constitutional monarchy or anything new (like communist China or late Soviet Union style) can be equally complex, and provided human freedom is looked after are just fine as well.
So, what are the practical implications for us? K.I.S.S - LINUX RULEZ! WORLD DOMINATION BY 2010! | | 12:21 am |
printf style output in VHDL One very useful package for VHDL is called PCK_FIO, it is available from http://www.easics.com/webtools/freesicsor from LML site: PCK_FIO-2002.7.tar.gzYou need to build the package and library (files PCK_FIO_1993.vhd and PCK_FIO_1993_BODY.vhd). I have used name printf for the library. Then it's very simple to use: library std;
use std.textio.all;
library printf;
use printf.pck_fio.all;
architecture ...
file output : text open write_mode is "assertTst.lst";
...
process ...
variable l : line;
...
fprint(output,l, "Tc=%3d ns A=%b B=%b x=%b\n",fo(NOW), fo(A), fo(B),fo(x));
Produces this output: Tc= 0 ns A=U B=U x=U
Tc= 0 ns A=U B=U x=0
Tc= 0 ns A=0 B=0 x=0
Tc= 1 ns A=0 B=0 x=1
Tc= 1 ns A=0 B=1 x=1
Tc= 1 ns A=1 B=1 x=1
Tc= 2 ns A=1 B=1 x=0
Tc= 2 ns A=0 B=0 x=0If you need to output to the console, comment out the "file output ... " line (VHDL standard environment has file 'output' already opened as standard output). These common types are supported for output: std_logic std_ulogic bit
std_logic_vector std_ulogic_vector bit_vector
integer unsigned signed
boolean
character
string
time | | Friday, March 16th, 2007 | | 3:43 am |
VHDL assert strange 'feature'
Here is VHDL problem. First, I thought it was ModelSim's problem, but then I've got exactly same results in Active-HDL I figured out that ModelSim is a mature product and it's my misunderstanding of VHDL (or it is the broken way VHDL standard is)
Here is the problem. Suppose we have 2 signals A and B. We are asserting that when B gets active A is already active for at least tS, i.e. we're checking that when B is set A is already set for specified setup time. We also want code to work for special case of tS := 0 ns.
entity assertTst is
end;
architecture sim of assertTst is
signal A : bit := '0';
signal B : bit := '0';
signal x : bit;
constant tS : time := 0 ns;
begin
checkAbeforeB : process(B)
begin
if B'event and B='1' then
assert A='1' and A'last_event >= tS;
end if;
end process;
x <= '1' after 1 ns;
B <= x;
A <= x and B;
end;
This is how the result looks in simulation time:
time delta x A B
0 +0 0 0 0
1 +0 1 0 0
1 +1 1 0 1
1 +2 1 1 1
So, it happens that although A and B are both getting '1' value at the end of simulation time 1, i..e. at delta +2, we do have B goint to '1' ahead of 'A' at delta +1. Why should assertion use the intermediate delta-time values, as opposed to final? Probably need to look for some 'stable attribute.
After some investigation...
Reduced it even further - still same issue. I wonder about assertions - they are useless arnt' they - if they trigger on intermediate results.
entity assertTst is
end;
architecture sim of assertTst is
signal A : bit := '0';
signal B : bit := '0';
signal x : bit;
begin
assert (A='0' and B='0') or (A='1' and B='1')
report "A,B not same";
x <= '1' after 1 ns;
B <= x;
A <= x and B;
end;
Well, figured out the workaround, but I need to investigate - I just can't believe the VHDL standard has this crazy behavior specified:
assert not (A'stable and B'stable) or (A='0' and B='0') or (A='1' and B='1')
report "A,B not same";
But this is really strange...
| | Saturday, March 3rd, 2007 | | 5:23 pm |
LML draft for VHDL coding style (standard) Whereas in C/C++/Perl/Pascal/etc. we have object talking to each other by the way of calling functions/methods, in VHDL we have components talking to each other by the way of signals. Apparently - entity is same as class (some storage and some inputs (methods) and some output (method return values or globals)). An instance of an entity is same as class instance - object. We are trying to base it on our C++ coding standard, therefore: 1. same2. Class Name --> Entity Name, or Process Name 2a. Architecture name should start with : RTL, SIM, then optional underscore and specifics Example: RTL, SIM, RTL_pipelined, RTL_async 3. attributes and operations --> variables 4. constants and labels - UPPER_CASE 6. global variables --> port_names and signal_names. 6a. use same name if possible for port name and signal connected to that port, if collision happens, add _sig suffix to signal name 6b. use _n for active love signal and por names 6c. use port_NAME_n for external ports - i.e. physical ports on the FPGA or ASICs 6d. use clk_ prefix for clocks and rst_ prefix for resets 7-10 same
11. line length limited to 72 chars 12. for entity ports the order: a) inputs: clk, rst, enable, address/data, control b) inputs: same order - clk, rst, enable, address/data, control c) for multi-section entities - repeat enable/address/data/control sequence for each section 13. port_name => signal, do not use positional assignments
14. TYPE and SUBTYPE names - UPPERCASE ====================== Googling on the subject gives several references: - CS Univ. of Alberta - based on Xilinx coding style- ALSE-FR coding styleWould love to hear comments. | | 5:07 pm |
LML C++ / C coding style (standard)
This is to be used for all C++, C and reasonbly applied to Perl or any other language.
1) Avoid abbreviations in names, rather have long, but easy to understand names.
2) Class names shall begin with Upper Case Letter and contain lower case.
Underscores and abbreviations to be avoided as much as possible.
Example: FileOperations
3) Class attributes and operations shall begin with lowercase and contain an Upper Case:
Example: xResolution, busNumber, colorEncoding
Exception: loop variables like: i, j, k
4) #define constants, all upper case, with words separated with "_"
Example: #define CHIP_ID 0x3876
4a) const constants (C++), shall also follow rule #4:
Example: const int CHIP_IDC=0x3876;
5) In C++ "const constants" shall be used instead of "#define constants"
6) global variables shall be lowercase, with words separated with "_"
Example: video_stream
NOTE: functions (even non static) shall not follow rule #6, but rule #3 instead
7) one operator per line
8) indent 2 level of spaces, no tabs
9) put { and } on a separate line, at the parent level
10) language keywords shall be lowercase (class, switch, for, if, etc.)
Example:
for(i=1;i<10;i++)
{
sumValue += i;
invSumValue += 100/i;
}
99. Linux Media Labs Copyright (default) or
GNU GPL Copyright shall be at the beginning of every file.
Long form is preferable, short form is acceptable for time/space
critical interpreted scripts. Those can be extracted from
SVN: LMLSW/GNUnotice
| | 4:49 am |
design to interface with async RAM connected to sync bus I wonder if there is a generic solution to the following problem: - We have RAM (ADDR(),RD,WR,CS: in; DATA: inout) interface
- We have sync. bus with CLK, WR_EN,ADDR, inout DATA
- We need a design to interface that sync. bus (like PCI) to that async RAM
It seems that it should not be that difficult, but certainly would require couple registers and state machine. | | Sunday, February 11th, 2007 | | 5:06 pm |
notes about Altera QuartusII LPT setup For /dev/parport0: - edit /etc/sysconfig/modules/udev-stw.modules to avoid usage of ppdev by lp (do vmware can use it) #!/bin/sh
#for i in nvram floppy parport_pc parport lp;do
for i in nvram floppy ppdev;do
modprobe $i >/dev/null 2>&1
done
- set permissions on /dev/parport0, file /etc/udev/rules.d/50-udev.rules to allow read/write by all: KERNEL=="parport*", GROUP="lp", MODE="0666" For WXP, in ... quartus60/drivers/i386 run: ntdriver.exe install
nnlpt /i
reboot, add byteblaster-II on LPT, it should be useable (run 'audo-detect' from quartus) | | Sunday, October 1st, 2006 | | 5:56 am |
LMLM4401 and MEZ4M2DK are close to reality
MEZ4M2DK is build, MG1264 soldered in and I'm working on assembly - seems most difficult part is over.
LMLM4401 is assembled (2 pcs) and I pray to God that they work fine, since then we can build 10-100 pcs and hope they would sell quick. But there is a task before sales can really start (unless Windows XP) - code the Linux driver. But that requires M2DK based PCI LA project to commence.
And the show is just in one week. That should be very tough, yet productive week. | | Monday, September 18th, 2006 | | 2:37 pm |
|
[ << Previous 20 ]
|
![[info]](http://p-stat.livejournal.com/img/userinfo.gif){kind=small}
vlnv's journal