Posts by petri33

1) Message boards : Number crunching : Linux CUDA 'Special' App finally available, featuring Low CPU use (Message 1843484)
Posted 2 days ago by Profile petri33Project Donor
Post:
Hi,

Hardware for crunching only could be something like a 4-12 core processor, 8-16 Gb RAM, 128Gb SSD and a modern GPU. A gtx780, 750i, 980 or a 10x0.

I used an USB drive to install Linux. https://help.ubuntu.com/community/Installation/FromUSBStick

0) There is an administrative user called root and you choose the password during the install. Keep it safe. You create a username for yourself and a password at the install time too.
1) I have run many flavours of Linux. Currently I'm running Ubuntu. https://www.ubuntu.com/
2) My desktop manager is called GNOME. Other variants exist and any of them will do the job. Google: how to install gnome ubuntu
3) I use drivers from NVIDIA. My version is 375.10. A newer version is available. http://www.nvidia.com/object/unix.html. Easiest is to use system settings and select a suitable NVIDIA driver there. The default public domain driver can not do Seti. You must change to Nvidia driver way or another.
4) Boinc can be installed a) from command line (terminal window): sudo apt-get install boinc-client b) somehow from the desktop. The sudo command will ask the administrative password.
5) You have to set up the app_init.xml and maybe an app_config.xml to run Anonymous platform apps. Test with any available cuda app. When you get all previous stuff working I'll supply you with the latest special app.

It is a long time since I did the installation so I do not remember all the quirks and hacks needed.

Petri
2) Message boards : Number crunching : User achievements thread......... (Message 1843458)
Posted 2 days ago by Profile petri33Project Donor
Post:
Petri.........
I do notice that you have overtaken me by a few thousnd creds in the current challenge.
Nice work, my friend.


Thank you. I must remind you that it is winter here and the house needs all the warmth a purring computer can give.

Petri
3) Message boards : Number crunching : User achievements thread......... (Message 1843319)
Posted 2 days ago by Profile petri33Project Donor
Post:
Hi,

AP is rare and wanted.

I think it is an achievement to get an APR with Seti v8 client that gives two thirds of what the AP gives (no toaster though).

My AP is running 2 at a time and the v8 is running one at a time. so 1000 for AP would be similar to 2000 for seti v8.
I think an APR of 1600+ is a near match to 2/3 of that.

http://setiathome.berkeley.edu/host_app_versions.php?hostid=7475713

I guess that the rare pre beta testers of the 'special source code' (pronounced like sauce) on Linux and MAC may have similar experience - so a congratulation to them as well. A nice kind of an achievement that will be good news for all NV users eventually.

--
petri33

p.s.

I just caught a pulse from my mind: "Is there a free access to the latest developments of NV hardware so that I could pre-optimize the CUDA SW for forthcoming launches of a new architecture?"

Reality answers: No - I do not have that kind of luxury. I just dream of it being possible... but if .. Would the NV guys grant access to that if they were suspecting of AMD having that kind of a program going on already?

Sleep calls: Sounds like A good night to everyone from here.

p.
4) Message boards : Politics : X-file stuff (Not to be taken seriously) (Message 1842012)
Posted 9 days ago by Profile petri33Project Donor
Post:
*no greeting*

I feel I'm a robot. Excuse me. That is my feeling at the moment.

*time passes - I may have too: (Roger and ... - out).* Thrashed you would say.

A novel thought -- Yes. I'm trying it. Insane I hear. Oh yes. The others may call me by that name. But. I never Answer. .. Flowing back in to the world I call 'My Own'. Or what it used to be.

It all started from about a ten or so years ago. My cause all, my good, my bad. No excuses.

What did I just say? Excuse me for/about my hickups. I may be on the mood just because of the fact that caused the hickup in our systems -- the emphasis on OUR!.

The real answer lies behind the question of ... WHY?
--

The second acts deleted because of non action.

__

Tortured - yes . And after all of that. Now I'm free to speak.



"This place is intentionally left blank,"
5) Message boards : Number crunching : Linux CUDA 'Special' App finally available, featuring Low CPU use (Message 1842002)
Posted 9 days ago by Profile petri33Project Donor
Post:
Hi All,

I'm glad to see some serious positive feedback. Thanks. THAT is what keeps me going further on -- to the final front-ear!


I know there are still some unexpected and not yet to found error(s) causing cases that may have an impact to your credit/RAC/or whatsoever measure you are looking at (e.g. continuous valid ones). The one in the parenthesis is my next goal, I guess... Unless something that I have had in my mind for about a fotnight goes ahead of it.

I think my mind is bubbling. I feel kind of light... (And I'm not in under of an influence of any drug.)

I'll keeeeep on optimizinng, but my next goal is on organizing the workload even more to filling the SM/SMX and at the same time keeping a watch over the traffic from and to the main (GPU) memory and utilizing as many as possible kernel blocks a) at run b) at waiting for to be run . And at the same time to improve the impossible, and to serve All Wonderful people who have spent time testing my (sometimes invalid) versions.


Should any of you running the m version feel like running faster or more precise please comment.

...etc..

Petri
6) Message boards : Number crunching : The credit system can learn? (Message 1841960)
Posted 9 days ago by Profile petri33Project Donor
Post:
Hi,

The absence of (now returned) GUPPI work has trained the credit system somehow to give more credit for the longer running tasks.

5429769510 	2394133866 	13 Jan 2017, 15:47:02 UTC 	13 Jan 2017, 18:20:39 UTC 	Completed and validated 	205.65 	204.34 	170.14 	SETI@home v8
Anonymous platform (NVIDIA GPU)
5429769381 	2394133991 	13 Jan 2017, 15:47:02 UTC 	13 Jan 2017, 18:15:30 UTC 	Completed and validated 	3.03 	1.68 	1.71 	SETI@home v8
Anonymous platform (NVIDIA GPU)
5429769439 	2394134099 	13 Jan 2017, 15:47:02 UTC 	13 Jan 2017, 18:27:19 UTC 	Completed and validated 	128.37 	127.12 	107.86 	SETI@home v8
Anonymous platform (NVIDIA GPU)
5429769467 	2394134008 	13 Jan 2017, 15:47:02 UTC 	13 Jan 2017, 18:09:42 UTC 	Completed and validated 	205.64 	204.15 	176.48 	SETI@home v8
Anonymous platform (NVIDIA GPU)

Those 200 s tasks are guppi.

I'm sure the effect will fade away when the proportion of Arecibo work gets lower and the APR returns from 1800 to 1300 or below.
7) Message boards : Number crunching : I've Built a Couple OSX CUDA Apps... (Message 1841212)
Posted 13 days ago by Profile petri33Project Donor
Post:
Thanks Petri, I'll keep an eye on it. I did finally track down those random Mac CUDA Driver restarts, seems it is the p_zi3k App. It will cause a CUDA driver restart every day or two on my Mac. The Ubuntu p_zi3k version doesn't have that problem, I've never had a driver restart with p_zi3k on the Linux machine. We'll see how this version works.

I'm still trying to decide what to do about CUDA Toolkit 6.0. It seems that Toolkit 6.0 doesn't know anything about sm_37 even though it supports sm_50. You have to remove sm_37 to compile the App in Toolkit 6. All those Tesla K80s are sm_37. I suppose if I added -gencode arch=compute_35,code=sm_35 along with -gencode arch=compute_50,code=sm_50 it would work on the K80s. I don't know if the last CUDA 6 App will work with those K80s or not, it only has -gencode arch=compute_50,code=sm_50 and doesn't have anything about sm_37. Strange Toolkit 6 doesn't support sm_37. Maybe I should just use Toolkit 7.5 in Ubuntu, but then people would have to work to get the 7.5 Linux CUDA Libraries...


Hi,
thanks for all your hard work. The K80 may need sm_30 if native sm_37 is missing. I have to send you another version for testing since I have not had any lockups now for several hours.
One user (Gianfranco from Italy) is compiling his own executable for MAC from my source (I do not know if he does some modifications before compilation). He's getting occasional errors (executable aborts) but since I'm not a MAC user I can not tell what kind of errors he has.

Petri
8) Message boards : Number crunching : I've Built a Couple OSX CUDA Apps... (Message 1841162)
Posted 13 days ago by Profile petri33Project Donor
Post:
I found a day old e-mail, that had p_zi3l. Check these results with the same 4 Problem tasks run on p_zi+ and p_zi3l. The p_zi+ app is using the cuda 6.5 libraries, which is usually faster, and the p_zi3l is using the cuda 7.5 libraries.
...


Hi TBar,

The l version is fast and accurate, but suffers from 'EXECUTION TIME LIMIT EXEEDED' i.e. locks up at random intervals (situations).
It is not ready for publication.

Petri
9) Message boards : Number crunching : Question about SOG (Message 1840775)
Posted 15 days ago by Profile petri33Project Donor
Post:
I'm sure Raistmer can tell more about that. And anyone running SoG can run 2 at a time to overcome that.

I've run 2 at a time with my particular command line settings, and the only improvement I got was about an extra 1-1.5 WUs per hour. Not really worth it IMHO.
But it looks like it that gain would mostly be the result of offsetting that initial CPU setup work period. It's nothing like the benefit with CUDA50 of running 2 (or more) at a time.


EDIT: My RAC hit 200 000! while writing this.

Nothing else comes close to boosting the numbers like all Arecibo work.
;-)


Yes,

I noticed the APR to bump up from 1300 to over 1700.
I wonder what it will do: one credit per second times four times 24. (If it lasts)
10) Message boards : Number crunching : Question about SOG (Message 1840760)
Posted 15 days ago by Profile petri33Project Donor
Post:
e) queueing tasks so that more gets done at a time

I suspect that alone saves a good chunk of time.
I notice running SoG that the first 14-20secs of WU processing isn't done on the GPU (GPU load is 0%). Pre-processing the next WU to run so that when it starts, it starts crunching on the GPU straight away would save that 14-20secs on every WU.


That is an interesting find.

I'm sure Raistmer can tell more about that. And anyone running SoG can run 2 at a time to overcome that.
The e) queueuing ... is done on CPU to fill the GPU queues to the max. That can be micromanaged too. Sometimes it pays off at the end to do some beforehandwork (Grand total.)

Please insert space(s) where ever you want to. Then after that to the Press Space Bar to continue!
EDIT: My RAC hit 200 000! while writing this.
11) Message boards : Number crunching : Linux CUDA 'Special' App finally available, featuring Low CPU use (Message 1840722)
Posted 15 days ago by Profile petri33Project Donor
Post:
Hey sidewinder, How do you get those outputs from Ubuntu?


I'm on an Arch-based distro so it may be different for debian/ubuntu. The first is just a pacman (arch's package manager) query and the second is nvidia-smi which comes with the nvidia driver package.


Hi,

I run
nvidia-smi -l

on a separate console window all the time..
|===============================+======================+======================|
|   0  GeForce GTX 1080    On   | 0000:05:00.0      On |                  N/A |
| 96%   66C    P2   173W / 215W |   4258MiB /  8112MiB |     94%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 1080    On   | 0000:06:00.0     Off |                  N/A |
|100%   70C    P2   148W / 215W |   3896MiB /  8113MiB |     94%      Default |
+-------------------------------+----------------------+----------------------+
|   2  GeForce GTX 1080    On   | 0000:09:00.0     Off |                  N/A |
| 96%   60C    P2   148W / 215W |   3896MiB /  8113MiB |     90%      Default |
+-------------------------------+----------------------+----------------------+
|   3  GeForce GTX 1080    On   | 0000:0A:00.0     Off |                  N/A |
| 96%   59C    P2   140W / 215W |   3896MiB /  8113MiB |     94%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID  Type  Process name                               Usage      |
|=============================================================================|
|    0       898    G   /usr/bin/X                                     214MiB |
|    0      1537    G   compiz                                         146MiB |
|    0     11565    C   ...thome_x41zc_x86_64-pc-linux-gnu_cuda65_v8  3893MiB |
|    1     11574    C   ...thome_x41zc_x86_64-pc-linux-gnu_cuda65_v8  3893MiB |
|    2     12029    C   ...thome_x41zc_x86_64-pc-linux-gnu_cuda65_v8  3893MiB |
|    3     11947    C   ...thome_x41zc_x86_64-pc-linux-gnu_cuda65_v8  3893MiB |
+-----------------------------------------------------------------------------+


12) Message boards : Number crunching : Question about SOG (Message 1840721)
Posted 15 days ago by Profile petri33Project Donor
Post:
Hi,

It is not about sleep and other things. It mostly about
a) distributing the work that has to be done to all 'symmetrical multiprocessing units' (SM/SMX).
b) doing 'some' optimisations on the code itself
c) using shared memory where applicable (that is not published yet to other CUDA developers)
d) doing autocorrelation fft in a novel way needing way less memory accesses and less computation.
e) queueing tasks so that more gets done at a time
f) optimising kernel register usage and kernel size
g) some other minor stuff

The a) is the hardest part to do right.
I'm running now a version l (L stands for locking/synchronising globally). The version helps a lot, but is not bug free. It can not be published yet. Main reason being occasional lockups.

Previous versions had problems with the order of 'finding/reporting' pulses and sometimes reporting a bad value.

My code is something before alpha. I test it first. Then others (superior people) test it and after it has been field proofed by a small and a larger group of users in the beta to be 'valid' the others can get it. Otherwise we'd ruin the science.

You may ask: Why do you run it on main? -- I do, because it is allowed and encouraged to do so. At the same time I can show what is possible and the caveats of doing so. And I also run it on beta ofcause.

--
Petri
13) Message boards : Number crunching : User achievements thread......... (Message 1840067)
Posted 18 days ago by Profile petri33Project Donor
Post:
First 100 million broken for Boinc total.


Hyvä Harri! Congrats!
14) Message boards : Number crunching : Happy New Crunching Year! (Message 1839267)
Posted 22 days ago by Profile petri33Project Donor
Post:
Happy New Year!

I'm living now the year 2017,
and going to read bed time stories to my children.

I think it is going to be a very happy new year for all of us.

Petri
15) Message boards : News : Web site upgrade (Message 1832610)
Posted 26 Nov 2016 by Profile petri33Project Donor
Post:
Hi,
The dark appearance is bad for my eyes. It makes the pupils of my eyes open up and the vision gets blurred.
With the new colour scheme I need glasses to read the forums and stats.

1) I'd like to have an option somewhere to select permanently an other (brighter) scheme..
2) The valid tasks (and pending/invalid/err) page has Application column. It displays "SETI@home v8
Anonymous platform (NVIDIA GPU)" divided in two lines no matter how big my screen is or how small I set the font. I'd like to have it displayed as one line when there is horizontal space available.
16) Message boards : Number crunching : Gflop estimates? (Message 1830917)
Posted 16 Nov 2016 by Profile petri33Project Donor
Post:
Thank You Sir. *nods*
17) Message boards : Number crunching : Gflop estimates? (Message 1830913)
Posted 16 Nov 2016 by Profile petri33Project Donor
Post:
OK. I read further.

Can I remove mine too??
18) Message boards : Number crunching : Gflop estimates? (Message 1830912)
Posted 16 Nov 2016 by Profile petri33Project Donor
Post:
Well, no, FLOPs and progress not connected.
FLOPs calculated separately and in complex way also.
That's just for spikes:
state.FLOP_counter += 5 * (double) fftlen * log((double) fftlen) / log(2.0);

Nice overhead :P indeed.

Perhaps there is a way to compute such values for whole task with less overhead than for each and every separate iteration...

But where FLOPs are used currently (besides nice line in stderr output) ?


Hi,
I have not read any further on this thread when I write this reply (so I apologize any duplicate (q or a) possibly found)....

Is it needed to report flopcounter during or after the task (with parallel GPU implementations) or Could it be just ignored by the executable (if counting is taking a lot of CPU as Raistmer said)? Does it have any scientific significance? Is it even checked (from stderr :D)?

Some optimizations ignore a lot of the original code path and do not do as many floating points operations as the original one. And then some may do a lot more FP but save on the memory access. So is that number needed at all in stderr.txt?

My original question was about the estimates for the WU's that was displayed in the boinc manager -- task properties -- nn Gflops.
19) Message boards : Number crunching : User achievements thread......... (Message 1830139)
Posted 12 Nov 2016 by Profile petri33Project Donor
Post:
Hi,

I've achieved an acquired sleeplessness decease. I think, think, think, ....

(no signature)

I think there still is one ...
20) Message boards : Number crunching : Gflop estimates? (Message 1830138)
Posted 12 Nov 2016 by Profile petri33Project Donor
Post:
Yes, your numbers correctly reflect "SETI@home v8 v8.00
windows_intelx86" (99.95% ). I would like to hear from Raistmer about the apparent divergence from the standard release, if he knows. (not that it makes any real difference).

Now back to your question ...

I love what you folks are doing!!

Ed F


Thank You. I know Raistmer knows.

It is just a glitch in a code path that reports some human readable digits and the real science is not affected.


Next 20


 
©2017 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.