Posts by petri33


log in
1) Message boards : Number crunching : linux vs windows (Message 1592081)
Posted 9 hours ago by Profile petri33
There may be an interesting case in the top hosts: http://setiathome.berkeley.edu/top_hosts.php

One running stock on Windows, one running lunatics on linux. Both have i7 and 4 x GTX780.

Only time will tell ...

But in the mean time You may peek at the daily scores from the links to the Free DC or Boinc stats.
2) Message boards : Number crunching : AP V7 (Message 1591807)
Posted 22 hours ago by Profile petri33
Scenario one Running 3 MB:
0.33 + 0.33 + 0-33 = 0-99. Now one MB quits -> only 0.33 + 0.33 = 0.66 running -> can start new AP. Now running 0.33+0.33+0.34=1.00 .

Scenario two Running one AP and two MB:
0.34 + 0.33 + 0.33 = 1.00- Now one MB finishes. Runnung 0.34 + 0.33 = 0.67. Can start a new MB or if an AP is scheduled to run next must wait for more free GPU.

Three: running 2 AP = 0.68. Must wait. After the other AP finishes can start AP or MB.

Scenarios one and three behave as expected. It's the "Scenario two" where BOINC is inconsistent. Most of the time, it will start a new MB whether or not an AP is the next task at the top of the queue (which is what I assume you mean by "scheduled"). That AP (and sometimes several APs) will remain in a "Ready to start" status until another AP finishes and frees up that .34 GPU. MBs that are lower in the queue will be started ahead of the APs if only .33 is available. Normally, the only time 2 APs run on a single GPU is when the 2 MBs running with the first AP happen to finish simultaneously (or nearly so).

But every once in awhile, BOINC lets an MB finish without starting another task of any kind. On rare occasions, it will actually do that on two different GPUs. Then, when the final MB finishes on one of the GPUs, it will go ahead and start a second AP on that GPU while simultaneously starting the next available MB on the second GPU. I've never been able to identify a pattern in any of this. :^)


Yes, I have seen that same and it does seem inconsistent to me too, but I guess that the 'next' in queue is not defined by deadline alone and may be not first in first out order either.

The average wait time for tasks is minimized if short tasks are processed first even though the throughput remains the same.
3) Message boards : Number crunching : AP V7 (Message 1591760)
Posted 1 day ago by Profile petri33
You didn't mention whether, when a single AP task is running on your gpu, an MB task is also running. Based on your app_config.xml, I'd expect that to be the case.

I've found that BOINC is sometimes unpredictable when a host is running mixed AP and MB, and the <gpu_usage> values are different for the two types. On my T7400, which currently has a GTX 780, a GTX 670, and a GTX 660, I've tried to set up the app_config.xml to run either 3 MB tasks on each GPU, or 1 AP and 2 MB tasks, which I've found makes about the most efficient use of those GPUs.

Theoretically, <gpu_usage> of .34 for the AP and .33 for the MB should do that, and it does.....most of the time. When an AP comes to the top of the queue, it will start up when an MB finishes, if there isn't already an AP running on that GPU. Once all 3 GPUs have a single AP running, BOINC will usually bypass APs and replace an MB that finishes with the next MB in line. However, sometimes it doesn't do that. In those cases, when an MB finishes on a GPU that already has an AP running, it won't start the next MB but will instead wait until the last MB finishes and then launch a second AP on that GPU, which makes that GPU noticeably underutilized.

So, sometimes I just set both AP and MB <gpu_usage> values to .33 and take my chances that I won't wind up with 3 APs running at the same time on one GPU. That's usually not a problem when APs are scattered in the queue, as is usually the case, but every once in a while the scheduler sends a whole block of APs at once, or on consecutive work fetches.

In any event, with your settings, you might not be able to get 2 APs running at the same time unless the 2nd and 3rd MBs finish at exactly the same time, or if BOINC decides to wait for all MBs to finish in order to launch the 2nd AP.

EDIT: Actually, the more I think about it, the more it seems the difficulty might be in launching the first AP, since when a single MB finishes, it only frees up .33 of a GPU, while the AP needs .5 GPU to start. BOINC would have to wait for a second MB to finish before it could launch the AP.


Scenario one Running 3 MB:
0.33 + 0.33 + 0-33 = 0-99. Now one MB quits -> only 0.33 + 0.33 = 0.66 running -> can start new AP. Now running 0.33+0.33+0.34=1.00 .

Scenario two Running one AP and two MB:
0.34 + 0.33 + 0.33 = 1.00- Now one MB finishes. Runnung 0.34 + 0.33 = 0.67. Can start a new MB or if an AP is scheduled to run next must wait for more free GPU.

Three: running 2 AP = 0.68. Must wait. After the other AP finishes can start AP or MB.
4) Message boards : Number crunching : Mismatch in time elapsed values? (Message 1583675)
Posted 17 days ago by Profile petri33

Also in my testing I found leaving HT enabled & running half the threads produced slightly higher output than HT disabled.
<-- I second to that.

My quess is that the processor is utilized to the max processes using real FPU's and then the additional fake CPU cores are used for whatever other (integer/logic) processing is needed.
5) Message boards : Number crunching : The ultimate build (Message 1582709)
Posted 19 days ago by Profile petri33
I'd ...

build somenhing with Seasonic X-1250 + Asus P9X79-E + el cheapo (used) 6/12 core intel LGA2011 processor + H20 cooling and then a bunch of extension cords and some 8 nvidia 7(5/9/x)0's + linux on SSD with the GPU's hanging in a circular free air iron rack (self made) over a 16" blower mounted below them. The RAM modules would have to be just to be sure Cas 6 at 2200+ MHz if I ever wanted to do any processing with the CPU.

... wish I had time to write to Santa.
6) Message boards : Number crunching : Max temp for getting the normal life out of a video card? (Message 1582695)
Posted 19 days ago by Profile petri33
My four gtx-780's are running at 73-95 C. The case is open on both sides and a 16" table top fan is blowing air directly attached to them. The card #3 has its second fan jammed for after a year of 24/7 computing. I have had to resort to (with all of my 780's) boring a hole near the axel and spraying in some rust remover oil - it seems to help getting the fans rolling again. The #3 needs some re-oiling though.

The room temp is at 24-28 C in the computer room and the Seti rig keeps the whole house warm since the heat is sucked out from the ceiling vent and blown out after it has warmed up the intake air in the X-section of the house ventilation system.

You should see my machine back at the top of the charts when it has its RAC back to normal.

--
petri33
7) Message boards : Number crunching : Step by Step, compile NVIDIA/MB CUDA app under Linux (Fedora 19) (Message 1527063)
Posted 11 Jun 2014 by Profile petri33
How about Maxwell being compute 50 ...

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

Just a guess.
8) Message boards : Number crunching : Compiling Applications for Linux (Message 1521301)
Posted 26 May 2014 by Profile petri33
There is a test suite available from either Lunatics or Crunchers Anonymous. I have not checked if You are a member of either of those, but it would be easiest to download from those places. I'm not sure if it is available without a login.

It is a bunch of testWu's and a shell script that runs a reference/standard app and multiple your own apps one after another and displays the execution times and the validity of the results.

I'll check where to find the needed files (unless someone with more knowledge can answer immediately.) EDIT: Found it, please check if You can open it.

Can you open this link http://www.arkayn.us/forum/index.php?board=9.0 ?
9) Message boards : Number crunching : Compiling Applications for Linux (Message 1519994)
Posted 22 May 2014 by Profile petri33
...

in addition you will probably want to examine the resulting Makefile in client/, locate the list of NVCC "-gencode" options, and add the newer compute capabilites that didn't exist before (if needed)
...



1) loacte sm_20 in some file
2) use your imagination and add an appropriate sm_35 if you have sm_35 capability.

in file 'configure' like this ..
... # AS_IF([echo $build_cpu | grep -q "_64"], if test x_$target_cpu = x_x86_64 ; then NVCCFLAGS+=" -m64 " else NVCCFLAGS+=" -m32 " fi # Add architecture to flags ## PCHIRP INCOMPATIBLE NVCCFLAGS+=" -gencode arch=compute_10,code=sm_10 -gencode arch=compute_11,code=sm_11 -gencode arch=compute_12,code=sm_12 -gencode arch=compute_13,code=sm_13" ## NVCCFLAGS+=" -gencode arch=compute_20,code=sm_20 -gencode arch=compute_20,code=sm_21 -gencode arch=compute_20,code=compute_20" NVCCFLAGS+=" -gencode arch=compute_35,code=sm_35 -gencode arch=compute_35,code=compute_35 --use_fast_math" ##NVCCFLAGS+=" -arch sm_50 --use_fast_math " ##NVCCFLAGS+=" -gencode arch=compute_35,code=sm_35 -gencode arch=compute_50,code=compute_50 --use_fast_math " ...
10) Message boards : Number crunching : Compiling Applications for Linux (Message 1519991)
Posted 22 May 2014 by Profile petri33


In what file do I comment out std_fixes.h and utils.h includes? I know how to do this, but I don't know what file they're in.


Find all files containing text std_fixes in your current directory:
"grep std_fixes *"

Someone might know the syntax for "find" command..
11) Message boards : Number crunching : Compiling Applications for Linux (Message 1519990)
Posted 22 May 2014 by Profile petri33
Thanks JAson!
12) Message boards : Number crunching : Compiling Applications for Linux (Message 1519194)
Posted 21 May 2014 by Profile petri33
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);
...
13) Message boards : Number crunching : Compiling Applications for Linux (Message 1519100)
Posted 21 May 2014 by Profile petri33
...

I successfully installed boinc and with the -DUSE_CUDA, I get the "can't find #include std_fixes.h" error. So successfully installing boinc didn't clear that error.

With out -DUSE_CUDA, I get the "extern int cudaAcc_SafeCall_No_Sync(...)" not in scope error.
...


So You need both -DUSE_CUDA and like Claggy said specify the boinc directory BOINCDIR=${HOME}/boinc for the Xbranch to compile.
14) Message boards : Number crunching : Compiling Applications for Linux (Message 1518813)
Posted 20 May 2014 by Profile petri33
Hmm, I guess I have had to specify BOINCDIR=/home/petri/boinc for that.
See an earlier message my configure scripts .

It was a try, edit and retry session a long time a go. I'm sorry I did not take notes what I did to get all to compile.
15) Message boards : Number crunching : Compiling Applications for Linux (Message 1518789)
Posted 20 May 2014 by Profile petri33
I'm not sure, but You might need -DUSE_CUDA when configuring the Xbranch ...
16) Message boards : Number crunching : Compiling Applications for Linux (Message 1517879)
Posted 18 May 2014 by Profile petri33
http://boinc.berkeley.edu/trac/wiki/SourceCodeGit


Any idea where I can get that directory with all header files?
17) Message boards : Number crunching : Did the work unit limit change? (Message 1517806)
Posted 18 May 2014 by Profile petri33
I could.
I will not.
I'll take the mixture.
(going childish)
Would You be asking if I had 240k.
--

Br
Tim
[/quote]

Thanks Tim! (looked just at your tasklist -- 800 MB ;) )

Br
Petri
18) Message boards : Number crunching : Who uses PCIe extenders; do they work? (Message 1517590)
Posted 17 May 2014 by Profile petri33
I have 1 inch raisers for 2 GPU's. No cords.
19) Message boards : Number crunching : Did the work unit limit change? (Message 1517586)
Posted 17 May 2014 by Profile petri33
I could.
I will not.
I'll take the mixture.
(going childish)
Would You be asking if I had 240k.
--

Now I'm a bit down.

Once upon a time there was a Setiland. All people ware treated equally and there was a feeling of harmony around.

One day TheLimit was raised!

Most people noticed but nothing... But the evil was lurking around. An AP hoarder amidst us Seitzens. The universe kept on expanding as time passed by.

But one day - The AP stopped to appear. That was not evident immediately. Everyones cache was filled with a mixture of MB and AP - so nobody realised that until it was too late. No one got no AP anymore. Their caches were run out of AP.

Except for one - there was this evil hoarder. He had set a spell to the magic internet page and made sure that hed'd get no MB. This way he got all the AP in the wolrd!

Oh the groaning and moaning every morning, throughout the day and every evening. No greater evil could have hit us - people said!

Some went to the extent that they switched to another project - some waited their mouths shut.

Time passed by...

And there was this one morning when all remaining GPU coolers whirred so beautifully and all the worlds results loked a bit more shiny. Something must have happened said everyone - no one knew what.

But the a knight in an Arctic Silver IV shining body came and told every person still in the project that the CPU cooler of the hoarder had caught a fire and caused a meltdown of those horrible AP hoarding GPU cores and that the King of the Setiland had issued a permanent order: NO ONE (Tim) WITH A SETTING THAT SAYS AP ONLY AND NOT ALLOW ANY OTHER TYPE OF WORK SHALL HAVE A LIMIT OF 1 TASK A DAY INSTEAD OF 100 PER GPU AND CPU.

The people hoorrared for a total of three days and .....

--

Well - the children are now asleep. Mee too in a an hour or so.

I just looked at at the tasks of another top 10 cruncher and was kind of disappointed. When running AP only I could get 240 000 a day - but I do not want to. I have AP as a preference, but atleast I have this "send work from other projects when available" on.

How would you explain if You had N hunderd AP only? (N is a big number)


Am I cheating somehow?

I don’t think so.

Is there somewhere written that I will take MB wu’s or AP wu’s?

I don’t think so.

I can do whatever I want to my preferences, asking no one what to do because ARE MINE.
No one will tell me if I will download 800 AP’s or 800 MB WU’s
Yes I can take 800 Ap tasks and I will do it again because the server allow, and it is legal.
We all wanted 100 wu’s per Gpu as I remember, and now we are complaining?

By the way…
This machine is going to retire at about one month or two.
We have here at the office a new build with dual xeon, and 8 Gpus.I will install seti and run 1000 AP tasks.

And a question to petri33…
If you were at 2nd position, and I was at 3rd, will you have those questions?

Br
Tim
20) Message boards : Number crunching : Compiling Applications for Linux (Message 1516869)
Posted 16 May 2014 by Profile petri33
More suitable = better, shorter, with no experimental stuff, no hard coded paths, ...


Next 20

Copyright © 2014 University of California