Posts by ivan


log in
1) Message boards : Number crunching : Linux 64, Mint 17, and Nvidia 340.24 driver (Message 1546268)
Posted 9 days ago by Profile ivan
I never managed to get BOINC to recognise NV GPUs with anything other than the standard Ubuntu nvidia drivers which at this point stands at 331 for Trusty. I don't know how much extra customising Mint does to the nvidia drivers.

Mint doesn't provide Nvidia drivers out-of-the-box, they give you xorg's nouveau driver. You can use Driver Manager to install proprietary drivers, currently mine is giving me the option of 331.38 or 304.117 -- or nouveau. I always download and install from the Nvidia site, however
2) Message boards : Number crunching : Linux 64, Mint 17, and Nvidia 340.24 driver (Message 1545733)
Posted 10 days ago by Profile ivan
I upgraded my home cruncher to Linux Mint 17 at the weekend because there were no more upgrades for Mint 15. Unfortunately you can't really upgrade, you have to do a new install, but I learnt long ago to keep /home on a separate partition so you don't have to recreate all your data and many of your personalisations.
I had to do a lot of fighting to get boimcmgr recompiled (boinc and boinccmd were OK) -- long story short, don't upgrade to wx3.0, stay with wx2.8.
I'm still using the stock applications, despite Petri having long ago sent me the recipes for updating Nvidia versions of MB and AP applications. So, I can only run AP on my video cards (1x GT 660 Ti + 1x GT 640), but I have been using libsleep.so to get around the busy-wait bug in the Nvidia drivers.
However, yesterday I got notification of several security updates, including the kernel, so as is my wont I downloaded the latest Nvidia driver too -- 340.24 (previously using 337.19). I'd had, as usual, the devil's own time banning the stock nouveau driver in favour of the proprietary Nvidia one when I upgraded but the step up to 340.24 went relatively easily -- one tip I learnt over the weekend, to kill X so you can go to runlevel 2 to install a new driver, use
sudo service stop mdm

Later, I realised that the gkrellm display was showing solid green CPU usage instead of the spiky look I was used to since starting using libsleep. Fired up top and sure enough, all four GPU AP jobs were using 100% CPU time and the processors were showing no idle time at all. I let it run overnight and this morning found that all recent AP results had CPU usage the same as overall time whereas heretofore it was in the ~10-30% range. So, I reinstalled the 337.19 driver and rebooted, and now the GPU AP jobs are back to 5-20% CPU and idle time is around 40% of a CPU.
So, beware if upgrading beyond 337.19 -- keep an eye on your CPU utilisation.
3) Message boards : Number crunching : Panic Mode On (88) Server Problems? (Message 1542532)
Posted 16 days ago by Profile ivan
Here's one of his machines, 7309756, that got into my database before he hid them. Looks like he ran S@H on it for about 4 weeks, then stopped cold on July 6. A lot of WUs successfully processed, which is terrific, but he might have left 100 in limbo if that machine doesn't connect again. Hope he doesn't do it that way for his whole data center.

Hmm, I just had a dual-node machine with those processors (tho' @ 2.5 GHz) ordered for me. I'll probably not run hyperthreading though, so 2x 20-core machines. :-)
4) Message boards : Number crunching : ASIC computers (Message 1540199)
Posted 20 days ago by Profile ivan
I'm not sure the ASICs could be repurposed in that way. See ASIC for seti.

Given the meaning -- Application-Specific Integrated Circuit -- I'm sure they can't be re-purposed. FPGAs (Field Programmable Gate Arrays) maybe, but they'd be devilishly hard to programme for SETI.
5) Message boards : Number crunching : Some questions about BOINC for Android... (Message 1539682)
Posted 21 days ago by Profile ivan
Twenty years from now, when your phone has more power than your desktop does now, maybe it'll be worthwhile, but not for a while.

See my recent post in the Milestones thread...
6) Message boards : Number crunching : SETI/BOINC Milestones [ v2.0 ] - XXVII (Message 1539660)
Posted 21 days ago by Profile ivan

And I think no one here sees things differently :)
I see it the same as you do. What's funny is that it really will be someone running SETI on an old cell phone that crunches the work unit with the alien contact in it.

You might be surprised. I recently found an app to compile gcc programmes, including Fortran, on my (original, 2012) Nexus 7 Android tablet (CCTools, if you're interested). For laffs I ran a 1975 Monte-Carlo programme that I used for my thesis. IIRC I got the run-time down to 5 hours on the University mainframe back then; on one core of the four in the Nexus 7 it took 78 seconds... (Admittedly, a 2 GHz Xeon with the Intel compiler takes 1.47 secs.)

100k is 100k more.

Indeed.
7) Message boards : Number crunching : Your thoughts on the upcoming Haswell E CPUs (Message 1538999)
Posted 22 days ago by Profile ivan
This question is slightly off topic but still to do with CPUs. Has anyone got a Devil's Canyon 4790K? These are stock clock at 4.0 GHz. Am just interested to see what performance is like?

http://www.theregister.co.uk/2014/06/30/review_intel_devils_canyon_cpu/
8) Message boards : Number crunching : Volunteer for SETI@home porting to x86 Android platform required (Message 1534744)
Posted 30 days ago by Profile ivan
-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/.
9) Message boards : Number crunching : How number crunching affects a computer (Message 1533376)
Posted 34 days ago by Profile ivan
If you run a laptop then you take your chances. These "Bic Lighters" I call them because they seem to be disposable will burn fairly quick. The newer ones with medium to high end GPU's in them are not going to last long if you do not have a cooling system on them.
A Laptop fan under them is a must.

I don't have much laptop experience other than Dells. Them I find usually acceptable until the ambient gets to 23-23 C, then they start to get much hotter. My original "big" Dell, their first Pentium, was way under-engineered thermally. By the time I retired it, because the power socket had died, it had had four motherboards, two disk drives, two DVD drives and two power supplies, all replaced under warranty. Then I inherited this one:

https://lh4.googleusercontent.com/-mklyxhhJQr8/T9ee4dM5ogI/AAAAAAAAAUk/du93tkvbHh4/w883-h662-no/S7001074.JPG
https://lh3.googleusercontent.com/-sXMS2BgAXOY/T9ee4zLWNrI/AAAAAAAAAUk/A6oEQaxKrx0/w883-h662-no/S7001075.JPG
https://lh3.googleusercontent.com/-VCGtytnfY3o/T9ee4j4_GFI/AAAAAAAAAUk/R0BmqI7ehCM/w883-h662-no/S7001076.JPG

I do wish laptop manufacturers would give us removable, cleanable, dust filters rather than our having to do a complete tear-down to remove the lint build-up!
10) Message boards : Number crunching : Panic Mode On (88) Server Problems? (Message 1530041)
Posted 43 days ago by Profile ivan
I'm currently seeing painfully slow responses (here in the UK) from Berkeley web-pages, including the Cricket graphs. Other US sites such as BoincStats and CNN are responding quickly, so it must be a localised problem.
11) Message boards : Number crunching : Compiling Applications for Linux (Message 1528105)
Posted 48 days ago by Profile ivan
make clean
./configure --enable-float --enable-type-prefix
make
sudo make install

Trying to build fftw 3.3.4 libs, but don't understand what the --enable-type-prefix is supposed to do.

Claggy

--enable-type-prefix Adds a `d' or `s' prefix to all installed libraries and header files to indicate the floating-point precision.

http://www.fftw.org/fftw2_doc/fftw_6.html, or perhaps
./configure --help
12) Message boards : Number crunching : Step by Step, compile NVIDIA/MB CUDA app under Linux (Fedora 19) (Message 1527107)
Posted 51 days ago by Profile ivan
How about Maxwell being compute 50 ...

-gencode=arch=compute_50,code=sm_50
-gencode=arch=compute_50,code=compute_50

Just a guess.

Now that's a puzzle. Yes, my results are saying "computeCap 5.0". I'll have to check tomorrow whether the URL I quoted matches the documentation I have on my Windows systems.

[Edit]
Actually, due to the wonders of broadband, ssh, and cygwin:
admD405@W7-SE-D304-01 /cygdrive/d/ProgramData/HoloMov/HoloMovie
$ grep sm HoloMovie_vs2010.vcxproj
<CodeGeneration>compute_10,sm_10;compute_20,sm_20</CodeGeneration>
<CodeGeneration>compute_10,sm_10;compute_20,sm_20</CodeGeneration>
<CodeGeneration>compute_10,sm_10;compute_20,sm_20</CodeGeneration>
<CodeGeneration>compute_20,sm_20;compute_30,sm_30;compute_35,sm_35;compute_50,sm_50;compute_50,compute_50</CodeGeneration>
so yes, my latest docs did specify sm_50 -- but I still need to check that I actually applied it to x64 Release version...
[/Edit]
13) Message boards : Number crunching : Step by Step, compile NVIDIA/MB CUDA app under Linux (Fedora 19) (Message 1527048)
Posted 51 days ago by Profile ivan
**And again, I get "Compiling entry function...sm_10, sm_11, sm_12, sm_13, sm_21, and sm_20", but nothing for sm_30 or sm_35

Maxwell Compatability Guide says to use
/usr/local/cuda/bin/nvcc
-gencode=arch=compute_20,code=sm_20
-gencode=arch=compute_30,code=sm_30
-gencode=arch=compute_35,code=sm_35
-gencode=arch=compute_35,code=compute_35
-O2 -o mykernel.o -c mykernel.cu

Note the omission of the 1xs -- on compiling my hologram application with CUDA 6 on Windows (haven't upgraded Linux yet) I got warnings that they are deprecated and soon to be removed. OTOH, after putting all the recommended options into the Windows compile my GTX 750 Ti still ran my reconstructions at exactly the same speed.
14) Message boards : Number crunching : SETI/BOINC Milestones [ v2.0 ] - XXVII (Message 1525929)
Posted 54 days ago by Profile ivan
Another milestone:
Me: Total credit 125,751,491
Big Reg: Total credit 125,751,136

...which puts me back into second place in the UK, and 56th overall (which won't last as finrod is steaming up behind me at a RAC over 300k, so I'll slip down again in a few days). If all holds together, I'll take over the UK lead in about ten months...

Not sure what's happened to Big Reg, hope it's not bad news. He stopped crunching a while ago and changed his username to "Gone".
15) Message boards : Number crunching : Compiling Applications for Linux (Message 1520969)
Posted 68 days ago by Profile ivan
with the sleep thing, according to http://linux.die.net/man/3/sleep , sleep() is defined in unistd.h , so

#include <unistd.h> somewhere near the top of the file it's complaining about, ( analyzefuncs.cpp )

Yeah, that's the change I had to make to compile on this Mint system.
Linux HomePC01 3.8.0-35-generic #50-Ubuntu SMP Tue Dec 3 01:24:59 UTC 2013 x86_64 x86_64 x86_64 GNU/Linux
16) Message boards : Number crunching : Compiling Applications for Linux (Message 1519311)
Posted 72 days ago by Profile ivan
My first lines of cuda/cudaAcc_utilities.h is different to that in the repository ... (One extra #include and 2 declarations)

#ifndef _CUDA_ACC_UTILITIES_H
#define _CUDA_ACC_UTILITIES_H

#include<driver_types.h>

void cudaAcc_setFuncDeviceCacheConfigShared(void);
void cudaAcc_setFuncDeviceCacheConfigL1(void);

extern int cudaAcc_SafeCall_No_Sync(cudaError_t err, char* errMessage, char* file, int line);
extern int cudaAcc_SafeCall_Sync(cudaError_t err, char* errMessage, char* file, int line);
extern int cudaAcc_SafeCall_No_Sync_wExit(cudaError_t err, char* errMessage, char* file, int line);
...


Well, cudaError_t is typedefed in driver_types.h:

/usr/local/cuda/include/driver_types.h:typedef __device_builtin__ enum cudaError cudaError_t;

and while I can find references to that file in some .o and .Po files in my s@h build area, I can't find any includes for it in s@h or BOINC source files. It is included in several other cuda header files though:

[HomePC01:~] > grep driver_types.h /usr/local/cuda/include/*.h
/usr/local/cuda/include/builtin_types.h:#include "driver_types.h"
/usr/local/cuda/include/channel_descriptor.h:#include "driver_types.h"
/usr/local/cuda/include/cublas_api.h:#include "driver_types.h"
/usr/local/cuda/include/cuda_device_runtime_api.h:#include "driver_types.h"
/usr/local/cuda/include/cuda_profiler_api.h:#include "driver_types.h"
/usr/local/cuda/include/cuda_surface_types.h:#include "driver_types.h"
/usr/local/cuda/include/cuda_texture_types.h:#include "driver_types.h"
/usr/local/cuda/include/cufft.h:#include "driver_types.h"
/usr/local/cuda/include/cufftw.h:#include "driver_types.h"
/usr/local/cuda/include/cusparse.h:#include "driver_types.h"
/usr/local/cuda/include/cusparse_v2.h:#include "driver_types.h"
/usr/local/cuda/include/driver_functions.h:#include "driver_types.h"
/usr/local/cuda/include/surface_types.h:#include "driver_types.h"
/usr/local/cuda/include/texture_types.h:#include "driver_types.h"

I have vague memories of having to include an extra header file once, but I don't remember exactly when or where.
17) Message boards : Number crunching : SETI/BOINC Milestones [ v2.0 ] - XXVII (Message 1518258)
Posted 74 days ago by Profile ivan
A new milestone:
Total credit 123,456,815

Yep, passed 123,456,789!
18) Message boards : Number crunching : Panic Mode On (88) Server Problems? (Message 1515688)
Posted 80 days ago by Profile ivan
Upload problems at present? Everything looks OK but may be a temporary problem?

Might be regional, I don't see anything obvious on the cricket graphs but my last contact was 20:59:47 UTC.
[Edit] Tja! Things started uploading 5 seconds after my original post! [/Edit]
19) Message boards : Number crunching : Could it be ? Tripple-Buffers in Nvidia drivers shorten WUs to crunsh ? (Message 1513669)
Posted 85 days ago by Profile ivan
I've no idea what tripple-buffers are :D

Are you trying to say that having triple buffering in the driver might make the WUs crunch faster?

Could you elaborate? What is it you see exactly under what conditions?

Triple-buffering is a technique for displaying video output. I can only see it in the Nvidia Control Panel under "3D settings". I'm sceptical that it could have any effect on number-crunching.
20) Message boards : Number crunching : Нужен совет по Intel GPU (Message 1511001)
Posted 92 days ago by Profile ivan
This is probably as good a place to ask as any: after I installed my new GTX 750 Ti in my work desktop, I noticed these two lines in the BOINC startup:
01-May-2014 15:21:52 [---] OpenCL: NVIDIA GPU 0: GeForce GTX 750 Ti (driver version 337.50, device version OpenCL 1.1 CUDA, 2048MB, 1867MB available, 2409 GFLOPS peak)
01-May-2014 15:21:52 [---] OpenCL CPU: Intel(R) Core(TM) i7-2600 CPU @ 3.40GHz (OpenCL driver vendor: Advanced Micro Devices, Inc., driver version 2.0 (sse2,avx), device version OpenCL 1.2 AMD-APP (938.2))

I'd not noticed before that I have an Intel CPU apparently using an AMD driver!
Anyway, is it possible to use the Intel GPU as well as the Nvidia, and how? I'm currently using a Lunatics install on a Win7 PC.


Next 20

Copyright © 2014 University of California