Posts by Raistmer


log in
1) Message boards : Number crunching : A journey: iGPU slowing CPU processing (Message 1545041)
Posted 7 days ago by Profile Raistmer
check thoroughly that all freq reducing CPU/mobo features are disabled.
These chips seems can't work at full load. Power or temp constrains lead to reduced freqs. For either CPU or GPU part. Cache contention effects can be big too, but if freqs go down effect will be much bigger.
2) Message boards : Number crunching : Volunteer for SETI@home porting to x86 Android platform required (Message 1535129)
Posted 25 days ago by Profile Raistmer
Paths set, now compilation errors:

oinc-analyzeFuncs.o `test -f 'analyzeFuncs.cpp' || echo './'`analyzeFuncs.cpp
analyzeFuncs.cpp:675:2: error: #error Analysis needs to be modified for your FFT
choice.
#error Analysis needs to be modified for your FFT choice.
^
analyzeFuncs.cpp:39:46: error: expected ',' or ';' before 'PACKAGE_STRING'
const char *BOINC_PACKAGE_STRING="libboinc: "PACKAGE_STRING;
^
analyzeFuncs.cpp:96:33: note: #pragma message: ----ooura----
#pragma message ("----ooura----")

matszpk added both FFTS and PFFFT to original MB sources - now need to establish what library actually used in working code...
3) Message boards : Number crunching : Volunteer for SETI@home porting to x86 Android platform required (Message 1535045)
Posted 25 days ago by Profile Raistmer
Added ' ' to paths and got:

if /cygdrive/c/Intel/INDE/Framework/NDK/toolchains/arm-linux-androideabi-4.8/pre
built/windows-x86_64/bin//arm-linux-androideabi-g++ --include ../sah_config.h -D
HAVE_CONFIG_H -I. -I.. -DHAVE_CONFIG_H -DTEXT_UI -DNDEBUG -DCLIENT -D_POSIX_C_
SOURCE=1 --sysroot=c:\Intel\INDE\Framework\NDK\platforms\android-4\arch-arm -Wal
l -O3 -fomit-frame-pointer -DANDROID=1 -I'c:\Intel\INDE\Framework\NDK\sources\cx
x-stl\gnu-libstdc++\4.8\include' -I'/cygdrive/c/Intel/INDE/Framework/NDK/sources
/cxx-stl/gnu-libstdc++/4.8/include' -Ic:\Intel\INDE\Framework\NDK/sources/cxx-st
l/gnu-libstdc++/4.8/libs/armeabi/include -Ic:\cygwin64\home\R\SETI_android\boinc
-api-7.0.36/include -Ic:\Intel\INDE\Framework\NDK\platforms\android-4\arch-arm/u
sr/include -Ic:\cygwin64\home\R\SETI_android\boinc-api-7.0.36/include -DHAVE_CON
FIG_H -DTEXT_UI -DNDEBUG -DCLIENT -I../db -I/ -I//api -I//lib -I//sched -I//db
-I../db -I/ -I//api -I//lib -I//sched -I//db -O2 -Wall -MT seti_boinc-main.o
-MD -MP -MF ".deps/seti_boinc-main.Tpo" -c -o seti_boinc-main.o main.cpp; then m
v -f ".deps/seti_boinc-main.Tpo" ".deps/seti_boinc-main.Po"; else rm -f ".deps/s
eti_boinc-main.Tpo"; exit 1; fi
In file included from main.cpp:46:0:
c:\Intel\INDE\Framework\NDK\sources\cxx-stl\gnu-libstdc++\4.8\include/cstdio:41:
28: fatal error: bits/c++config.h: No such file or directory
#include <bits/c++config.h>
^
compilation terminated.

Looks like "++" in directory name required additional escaping.
And now error is more direct - there is no such file in such directory indeed.

For the other side there are really lots of cstdio includes available:

C:\Intel\INDE\Framework\NDK\sources\cxx-stl\llvm-libc++\libcxx\include\cstdio
C:\Intel\INDE\Framework\NDK\sources\cxx-stl\gnu-libstdc++\4.8\include\cstdio
C:\Intel\INDE\Framework\NDK\sources\cxx-stl\gnu-libstdc++\4.8\include\tr1\cstdio
C:\Intel\INDE\Framework\NDK\sources\cxx-stl\gnu-libstdc++\4.6\include\cstdio
C:\Intel\INDE\Framework\NDK\sources\cxx-stl\gnu-libstdc++\4.6\include\tr1\cstdio
C:\Intel\INDE\Framework\NDK\sources\cxx-stl\gabi++\include\cstdio
C:\Intel\INDE\Framework\NDK\sources\cxx-stl\stlport\stlport\cstdio
C:\Intel\INDE\Framework\NDK\sources\cxx-stl\system\include\cstdio

Maybe I picked up wrong version?...
4) Message boards : Number crunching : Volunteer for SETI@home porting to x86 Android platform required (Message 1535009)
Posted 25 days ago by Profile Raistmer
Ok, tried to supply both paths:

if /cygdrive/c/Intel/INDE/Framework/NDK/toolchains/arm-linux-androideabi-4.8/pre
built/windows-x86_64/bin//arm-linux-androideabi-g++ --include ../sah_config.h -D
HAVE_CONFIG_H -I. -I.. -DHAVE_CONFIG_H -DTEXT_UI -DNDEBUG -DCLIENT -D_POSIX_C_
SOURCE=1 --sysroot=c:\Intel\INDE\Framework\NDK\platforms\android-4\arch-arm -Wal
l -O3 -fomit-frame-pointer -DANDROID=1 -I/cygdrive/c/Intel/INDE/Framework/NDK/so
urces/cxx-stl/gnu-libstdc++/4.8/include
-Ic:\Intel\INDE\Framework\NDK\sources\cx
x-stl\gnu-libstdc++\4.8\include
-Ic:\Intel\INDE\Framework\NDK/sources/cxx-stl/gn
u-libstdc++/4.8/libs/armeabi/include -Ic:\cygwin64\home\R\SETI_android\boinc-api
-7.0.36/include -Ic:\Intel\INDE\Framework\NDK\platforms\android-4\arch-arm/usr/i
nclude -Ic:\cygwin64\home\R\SETI_android\boinc-api-7.0.36/include -DHAVE_CONFIG_
H -DTEXT_UI -DNDEBUG -DCLIENT -I../db -I/ -I//api -I//lib -I//sched -I//db -I
../db -I/ -I//api -I//lib -I//sched -I//db -O2 -Wall -MT seti_boinc-main.o -MD
-MP -MF ".deps/seti_boinc-main.Tpo" -c -o seti_boinc-main.o main.cpp; then mv -f
".deps/seti_boinc-main.Tpo" ".deps/seti_boinc-main.Po"; else rm -f ".deps/seti_
boinc-main.Tpo"; exit 1; fi
main.cpp:46:18: fatal error: cstdio: No such file or directory
#include <cstdio>

^
compilation terminated.
Makefile:1692: recipe for target 'seti_boinc-main.o' failed

Maybe some other suggestions?

EDIT: and leaving only POSIX-style path leads to the same error.
5) Message boards : Number crunching : Volunteer for SETI@home porting to x86 Android platform required (Message 1534758)
Posted 26 days ago by Profile Raistmer
-Ic:\Intel\INDE\Framework\NDK/sources/cxx-stl/gnu-libstdc++/4.8/include
Need to say that cstdio is in include path:

Directory of c:\Intel\INDE\Framework\NDK\sources\cxx-stl\gnu-libstdc++\4.8\include

Forward slashes rather than backward sloshes?
Or, perhaps more likely, the DOS C: idiom; IME cygwin only groks that for the cd command. Try changing all your C:/s to /cygdrive/c/.


That path would fail on compilation ability checkng already.
Note that NDK used from Intel INDE, that is, windows-compiled toolset that expects usual windows-style directory names (Cygwin not included in INDE at all, I use Cygwin to be able to use configure scripts).
I'll check again regarding slashes but perhaps something another here.
(and already checked if all would be "\", w/o mix - didn't help).
[Perhaps worth to add both variants together]
6) Message boards : Number crunching : Volunteer for SETI@home porting to x86 Android platform required (Message 1534533)
Posted 26 days ago by Profile Raistmer
I'm trying to do Android build in Cygwin + Intel INDE environment.
After gew attempts configure ran OK and make began.
But compilation fails on this statement:

make all-recursive
make[1]: Entering directory '/home/R/SETI_android/setiathome-7.0/seti_boinc'
Making all in client
make[2]: Entering directory '/home/R/SETI_android/setiathome-7.0/seti_boinc/clie
nt'
if /cygdrive/c/Intel/INDE/Framework/NDK/toolchains/arm-linux-androideabi-4.8/pre
built/windows-x86_64/bin//arm-linux-androideabi-g++ --include ../sah_config.h -D
HAVE_CONFIG_H -I. -I.. -DHAVE_CONFIG_H -DTEXT_UI -DNDEBUG -DCLIENT -D_POSIX_C_
SOURCE=1 --sysroot=c:\Intel\INDE\Framework\NDK\platforms\android-4\arch-arm -Wal
l -O3 -fomit-frame-pointer -DANDROID=1 -Ic:\Intel\INDE\Framework\NDK/sources/cxx
-stl/gnu-libstdc++/4.8/include
-Ic:\Intel\INDE\Framework\NDK/sources/cxx-stl/gnu
-libstdc++/4.8/libs/armeabi/include -Ic:\cygwin64\home\R\SETI_android\boinc-api-
7.0.36/include -Ic:\Intel\INDE\Framework\NDK\platforms\android-4\arch-arm/usr/in
clude -Ic:\cygwin64\home\R\SETI_android\boinc-api-7.0.36/include -DHAVE_CONFIG_H
-DTEXT_UI -DNDEBUG -DCLIENT -I../db -I/ -I//api -I//lib -I//sched -I//db -I.
./db -I/ -I//api -I//lib -I//sched -I//db -O2 -Wall -MT seti_boinc-main.o -MD -
MP -MF ".deps/seti_boinc-main.Tpo" -c -o seti_boinc-main.o main.cpp; then mv -f
".deps/seti_boinc-main.Tpo" ".deps/seti_boinc-main.Po"; else rm -f ".deps/seti_b
oinc-main.Tpo"; exit 1; fi
main.cpp:46:18: fatal error: cstdio: No such file or directory
#include <cstdio>
^
compilation terminated.
Makefile:1692: recipe for target 'seti_boinc-main.o' failed
make[2]: *** [seti_boinc-main.o] Error 1

Need to say that cstdio is in include path:

Directory of c:\Intel\INDE\Framework\NDK\sources\cxx-stl\gnu-libstdc++\4.8\include

06/28/2014 05:46 AM <DIR> .
06/28/2014 05:46 AM <DIR> ..
01/22/2014 12:48 PM 2,518 algorithm
01/22/2014 12:48 PM 9,495 array
01/22/2014 12:48 PM 34,101 atomic
06/28/2014 05:45 AM <DIR> backward
06/28/2014 05:46 AM <DIR> bits
01/22/2014 12:48 PM 45,616 bitset
01/22/2014 12:48 PM 1,620 cassert
01/22/2014 12:48 PM 1,318 ccomplex
01/22/2014 12:48 PM 2,409 cctype
01/22/2014 12:48 PM 1,742 cerrno
01/22/2014 12:48 PM 2,051 cfenv
01/22/2014 12:48 PM 1,861 cfloat
01/22/2014 12:48 PM 23,596 chrono
01/22/2014 12:48 PM 2,157 cinttypes
01/22/2014 12:48 PM 1,490 ciso646
01/22/2014 12:48 PM 1,885 climits
01/22/2014 12:48 PM 1,905 clocale
01/22/2014 12:48 PM 42,445 cmath
01/22/2014 12:48 PM 52,517 complex
01/22/2014 12:48 PM 1,461 complex.h
01/22/2014 12:48 PM 7,992 condition_variable
01/22/2014 12:48 PM 1,949 csetjmp
01/22/2014 12:48 PM 1,855 csignal
01/22/2014 12:48 PM 1,407 cstdalign
01/22/2014 12:48 PM 1,844 cstdarg
01/22/2014 12:48 PM 1,401 cstdbool
01/22/2014 12:48 PM 1,603 cstddef
01/22/2014 12:48 PM 2,167 cstdint
01/22/2014 12:48 PM 4,262 cstdio
01/22/2014 12:48 PM 6,157 cstdlib

Maybe someone familiar with GNU toolset could suggest where issue could hide?
7) Message boards : Number crunching : Electricity Usage (Message 1531183)
Posted 35 days ago by Profile Raistmer
Here we have sealed measuring devices in flat or immediately near it, observable by inhabitants. They send device data to electric company each month (advanced devices allow 3 different rates for different time of a day, night goes with discounts). Also each few months representative of company check current measuring device state to see if sent values are right.
That way no easy cheating from both of sides involved. Just to get bill and pay for it is a direct way to corruption from power companies side.
8) Message boards : Number crunching : Massive failure rate for GPU taks on Nvidia (Message 1528615)
Posted 41 days ago by Profile Raistmer
Текст протокола

<core_client_version>7.2.42</core_client_version>

http://setiathome.berkeley.edu/result.php?resultid=3586557052

Too little in log to blame GPU drivers.... No stderr at all and one could expect at least something in case of GPU failure from initial CPU part of app...

EDIT: I would check if app binary exists at all and not deleted by some too carefull antivirus...
9) Message boards : Number crunching : ati radeon 290 (Message 1528265)
Posted 42 days ago by Profile Raistmer
Does it mean fix work or just that plan used? AFAIK one host can get apps from few conforming plan classes. So just to get some usage doesn't mean that plan used where no another worked before.

Some more direct confirmation like host with recent ATi GPU w/o CAL driver working with SETI stock app would be good.
10) Message boards : Number crunching : ati radeon 290 (Message 1527892)
Posted 44 days ago by Profile Raistmer
Eric added new plan class. Please check if the issue is fixed.
11) Message boards : Number crunching : Exotic Hardware (Message 1525862)
Posted 49 days ago by Profile Raistmer

According to http://community.arm.com/message/18342#18342 I had to learn that the Mali-400 GPU does not support OpenCL.

Cheers,
Juergen


Well, then this GPU not suitable for SETI in the current state. Though there were attempts to implement FFT via non-compute shaders on the dawn of GPGPU I'm not sure it's practical to try to adapt them to this GPU. New GPU with OpenCL support will be widely available well before such porting even will start.
12) Message boards : Number crunching : Exotic Hardware (Message 1525838)
Posted 50 days ago by Profile Raistmer
Current SETI apps are OpenCL 1.0 conformant so can be used on any OpenCL 1.0 and upper device if its floating point precision is good enough.

All that needed is volunteer with corresponding SDK knowledge for porting/rebulding.
13) Message boards : Number crunching : Exotic Hardware (Message 1525811)
Posted 50 days ago by Profile Raistmer
From that article (about FFT) one could conclude that that GPU supports OpenCL. This fact makes it not SO exotic hardware and eventually it can be used by SETI as a matter of SDK installation and OpenCL apps rebuild.
14) Message boards : Number crunching : Some questions about BOINC for Android... (Message 1524444)
Posted 53 days ago by Profile Raistmer
A few follow up thoughts ... The HP 7 Plus 1301 has a micro SD slot. Would adding external memory in this form help with the crunching?

Have not gotten an answer yet to the 're-naming' question but have found a useful site for HP 7 Plus --
http://h30434.www3.hp.com/t5/Android-Tablets-e-g-HP-Slate-7/bd-p/android. If/when I do get a renaming answer - will post it here. I can't be the only one with a 1301 - can I?

Moving the app to SD is not supported.
Changing the machine identity is not currently supported. However NativeBOINC supports this feature.

NativeBOINC supports both those features.
After ~year of development of stock version NativeBOINC remains superior one.
15) Message boards : Number crunching : Volunteer for SETI@home porting to x86 Android platform required (Message 1521574)
Posted 61 days ago by Profile Raistmer
If you want to help SETI project by building MultiBeam or AstroPulse applications for Android x86 platform (new devices on Intel Atom and other gadgets) please let us know. SETI code base is open for Anonymous access and available here:
https://setisvn.ssl.berkeley.edu/svn/seti_boinc/
16) Message boards : Number crunching : Intel HD 4600 (Message 1520187)
Posted 66 days ago by Profile Raistmer
try to extend desktop onto that gpu
17) Message boards : Number crunching : setup for GPU type AP on ATI 5770 and AMD A10-5800k (Message 1519463)
Posted 67 days ago by Profile Raistmer
Another option is exclude the A10-5800K from doing AP, and make it do Seti v7 exclusively, and exclude the HD5770 from doing Seti v7 and make it do AP exclusively,
In the cc_config.xml:
Claggy


I would call this suboptimal cause both GPUs are ATi.
18) Message boards : Number crunching : setup for GPU type AP on ATI 5770 and AMD A10-5800k (Message 1519173)
Posted 67 days ago by Profile Raistmer
BOINC does't device-centric so "not in easy way".
You would need 2 separate BOINC installations with one enabled and another deisblad GPUs in them.
19) Message boards : Politics : Crimea 2 (Message 1514954)
Posted 77 days ago by Profile Raistmer
Are we meant to believe the results of this latest referendum will be any different from that in the Crimea.

Putin's 'Human Rights Council' Accidentally Posts Real Crimean Election Results

As you may recall, the official Crimean election results, as reported widely in the Western press, showed a 97 percent vote in favor of annexation with a turnout of 83 percent. No international observers were allowed. The pro-Russia election pressure would have raised the already weak vote in favor of annexation, of course.

Yesterday, however, according to a major Ukrainian news site, TSN.ua, the website of the President of Russia’s Council on Civil Society and Human Rights (shortened to President’s Human Rights Council) posted a report that was quickly taken down as if it were toxic radioactive waste. According to this purported report about the March referendum to annex Crimea, the turnout of Crimean voters was only 30 percent. And of these, only half voted for the referendum–meaning only 15 percent of Crimean citizens voted for annexation.

The TSN report does not link to a copy of the cited report. However, there is a report of the Human Rights Council, entitled “Problems of Crimean Residents,” still up on the president-sovet.ru website, which discusses the Council’s estimates of the results of the March 16 referendum. Quoting from that report: “In Crimea, according to various indicators, 50-60% voted for unification with Russia with a voter turnout (yavka) of 30-50%.” This leads to a range of between 15 percent (50% x 30%) and 30 percent (60% x 50%) voting for annexation. The turnout in the Crimean district of Sevastopol, according to the Council, was higher: 50-80%.

The original version, in Russian, and a version clumsily translated by Google, are below:

Crimea vote


To make sure no one misses this:

Official Kremlin results: 97 percent of polled voters for annexation, turnout 83 percent, and 82 percent of total Crimean population voting in favor.

President’s Human Rights Council mid-point estimate: 55 percent of polled voters for annexation, turnout 40 percent, 22.5 percent of total Crimean population voting in favor.


Проблемы жителей Крыма
В связи с многочисленными ссылками СМИ на обзор "Проблемы жителей Крыма" как на официальный документ Совета при Президенте РФ по развитию гражданского общества и правам человека, выражающий оценку Советом крымского референдума, разъясняем, что он таковым не является.

Один из его авторов, член Совета Евгений Бобров, с 16 по 18 апреля 2014 года совершил частную поездку в Крым, итогом которой стал обзор о проблемах местных жителей в разных сферах. Текст обзора был 22 апреля 2014 года опубликован Бобровым в материалах постоянной комиссии по миграционной политике и защите прав человека в сфере межнациональных отношений, председателем которой он является, и в его персональном блоге на сайте Совета.

В обзоре тезисно изложены некоторые аспекты проблем, с которыми столкнулись жители Крыма в процессе интеграции новых субъектов Российской Федерации в экономическую, финансовую, кредитную и правовую системы страны. Информация разделена по темам "гражданство", "регистрация (прописка)", "убежище", "регистрация прав на недвижимость", "продовольствие и продукты", "образовательные учреждения", "межнациональные отношения", "религия и церковь", "резонансные убийства, задержания и похищения граждан", банковская система. пенсии и социальные пособия", "судебная система", "средства массовой информации", "бизнес", земля и положение сельских жителей", "туризм", "референдум". При этом обзор не содержит политических оценок, результатов официальных опросов, обследований или экспертиз. В нем представлены лишь личные наблюдения и мнения авторов. В частности, говоря о крымском референдуме, авторы обзора передают исключительно оценочные суждения своих собеседников, никак не оценивая их объективность и точность.

Совет просит представителей СМИ не извращать содержание данного обзора.

Count votes further while peoples burnt alive, hypocrites...
20) Message boards : Politics : Crimea 2 (Message 1514949)
Posted 77 days ago by Profile Raistmer
Actually, after Odessa's events I have nothing more to say to all these damned hypocrites.
Continue to discuss how much % voted, continue to support atrocities that happened and will be happened... it's on hands of such supporters...


Next 20

Copyright © 2014 University of California