Overclock.net banner

XANSONS for COD: Project Resume Next Week

2K views 43 replies 10 participants last post by  mmonnin 
#1 ·
http://xansons4cod.com/xansons4cod/forum_thread.php?id=29
Quote:
The maintenance is almost over. The new native BOINC apps are being tested in the local network now. We expect the project to be restarted at the end of next week. This will be preceded by server and storage upgrade. The work on the database is still in progress and will continue in parallel with the computations.

Also, the paper "XANSONS for COD: a new small BOINC project in crystallography" was accepted for presentation at the BOINC:FAST'2017 conference which is to be held in Petrozavodsk, Russia in August.
I got a whopping 3 GPU tasks before it went down for maintenance quite awhile ago. Hopefully they can keep enough for another GPU project.
 
#3 ·
Quote:
Originally Posted by C4pt41n M0 R0n View Post

It was just for returning contributors last I checked, I imagine they'll open it up to others soon.
Where does it sat that? The last time there were tasks available I signed up and immediately got 3 tasks. Then the project shut down for the mentioned maintenance.
 
#6 ·
#7 ·
#8 ·
The tasks are too damn quick so I made an app_config to only use 1 CPU core. The server can't keep it fed so other projects were constantly pausing and resuming:

Code:

Code:
<app_config>
        <app_version>
                <app_name>xansons_cpu</app_name>
                <plan_class>mt_linux</plan_class>
                <avg_ncpus>1</avg_ncpus>
        </app_version>
</app_config>
Running 3x GPU tasks at once:

Code:

Code:
<app_config>
  <app>
    <name>xansons_gpu</name>
      <gpu_versions>
      <gpu_usage>0.33</gpu_usage>
      <cpu_usage>0.1</cpu_usage>
     </gpu_versions>
  </app>
</app_config>
 
#14 ·
#15 ·
I have this running on my 480 in Linux, and the initial results look very good. The sample size is small, but here are a couple of tasks for comparison vs my wingmen:


Edit Just noticed that computer 788 is Egilman and one of his 7990s.
:D
 
#16 ·
Quote:
Originally Posted by tictoc View Post

I have this running on my 480 in Linux, and the initial results look very good. The sample size is small, but here are a couple of tasks for comparison vs my wingmen:

Edit Just noticed that computer 788 is Egilman and one of his 7990s.
biggrin.gif
Now you just need to join Team OCN.
wink.gif


Well Found you and shows you in OCN but team page does not. ;( Guess it doesn't update members right away.
http://xansons4cod.com/xansons4cod/team_members.php?teamid=1027&offset=0&sort_by=expavg_credit
 
#18 ·
Quote:
Originally Posted by tictoc View Post

I have this running on my 480 in Linux, and the initial results look very good. The sample size is small, but here are a couple of tasks for comparison vs my wingmen:

Edit Just noticed that computer 788 is Egilman and one of his 7990s.
biggrin.gif
Just remember that 7990 is running 4x..... (I know it's an excuse, but then again I have the fastest Tahitis on the project, probably cause I have the only independent Tahitis on the project)

Still this is the first project I've seen a 480 faster than a tahiti of mine....... Opps wait a sec........ that is in Linus isn't it?
biggrin.gif


Well, that changes everything, try and get that time in windows..... {chuckle}

Decidedly an Nvidea/Linux favorable project. probably something I won't be running run long term..... Currently I've got 8 on it and it pencils out to 1.3 million a day.

Besides I have a strong dislike for projects that update once a day, makes it difficult to track. (and very boring)
 
#21 ·
Quote:
Originally Posted by Egilman View Post

Quote:
Originally Posted by tictoc View Post

I have this running on my 480 in Linux, and the initial results look very good. The sample size is small, but here are a couple of tasks for comparison vs my wingmen:

Edit Just noticed that computer 788 is Egilman and one of his 7990s.
biggrin.gif
Just remember that 7990 is running 4x..... (I know it's an excuse, but then again I have the fastest Tahitis on the project, probably cause I have the only independent Tahitis on the project)

Still this is the first project I've seen a 480 faster than a tahiti of mine....... Opps wait a sec........ that is in Linus isn't it?
biggrin.gif


Well, that changes everything, try and get that time in windows..... {chuckle}

Decidedly an Nvidea/Linux favorable project. probably something I won't be running run long term..... Currently I've got 8 on it and it pencils out to 1.3 million a day.

Besides I have a strong dislike for projects that update once a day, makes it difficult to track. (and very boring)
Just for clarification when you say 4x is that concurrently per GPU (8 tasks per video card, 16 tasks for the machine) or do you mean 4 tasks total for the machine?
 
#22 ·
Quote:
Originally Posted by tictoc View Post

Just for clarification when you say 4x is that concurrently per GPU (8 tasks per video card, 16 tasks for the machine) or do you mean 4 tasks total for the machine?
16 for the machine, 4 per GPU......

My app_config....

Code:

Code:
<app_config>
  <app>
    <name>xansons_gpu</name>
      <gpu_versions>
      <gpu_usage>0.25</gpu_usage>
      <cpu_usage>0.45</cpu_usage>
     </gpu_versions>
  </app>
</app_config>
Runs very smooth and cool, and I was giving it 8 cores in process lasso. At Collatz settings, (or a bit below)

One of the problems with tracking the WU's is that they are all different lengths ranging from 9 seconds to 22 minutes. (CPU's turned off in preferences) So we have to resort to wingman searches to find comparables.

Not an easy task.....

So, for those specific WU, if I was running just those WU I would be getting 4 per GPU in 10:27 or 2:37 per WU, for the whole machine the output would be 39-40 seconds per if it was running just that specific type of WU. But it gives out all the different WU sizes all at the same time so it is very difficult to track comparisons like this...
 
#24 ·
Quote:
Originally Posted by tictoc View Post

I'll bump my 480 up to 4x and then we can at least compare our two wildly different setups.
biggrin.gif
Would be interesting to see how it does at the higher performance settings, just remember that your in Linux with that card and it seems to run 45-50% faster in linux than windows...

And I cannot run a 7990 in linux...... (no functional drivers)
 
#25 ·
The Linux-Windows comparison is the most interesting thing to me. No other GPU project has been substantially better in Linux. For the most part an AMD GPU in Linux has been equal to or worse than running in Windows.

For NVIDIA that is generally not the case. Prior to the development of the new AMD Linux driver stack, AMD performance in Linux for most tasks was generally poor compared to NVIDIA.
 
#26 ·
Well, if the initial comparison holds and your 480 can beat my Tahitis running in linux, then I think they may have fixed the problem I think.....

The thing that irks me is these projects, if optimised correctly, run equally well on both types of hardware, Collatz proved that for good with it's nVideaOCL app.

Just get the project devs to implement it.....

(good luck with that)
 
This is an older thread, you may not receive a response, and could be reviving an old thread. Please consider creating a new thread.
Top