[Campaign-news] New k-centers

whsu at sfsu.edu whsu at sfsu.edu
Fri Feb 12 10:54:55 PST 2010


Hi Kai,

Ah, I just got the profiler to work from the command line. Thanks!

Bill

Quoting Kai Kohlhoff <kohlhoff at stanford.edu>:

> Hi Bill (and Marc),
>
> You are right, we should probably agree on some default directory   
> for the includes (I think neither mine nor Marc's is) and provide   
> the Makefile correspondingly.
>
> The file should of course not be hardcoded in the final release, in   
> fact I don't think that the main() should be provided.    
> kcenterGPU.cu can be included like a header file in users' projects,  
>  which of course makes sense for a library.  I think we should later  
>  move the main into a separate file as one example on how to use the  
>  library.  If you prefer providing a file name on the command line,   
> please go ahead and make any changes you like.
>
> I will look into that error; it seems that I need to do some more   
> testing for higher dimensional data.
>
> Profiling works fine for me.  I am developing on our Tesla cluster,   
> but have just tested the profiler on my MacBook Pro.  Did you start   
> the profiler from within a Terminal window?  It needs to have the   
> correct PATH variables loaded, so just double clicking the program   
> icon does not work.
>
> - Kai
>
>
>
>
> On Feb 11, 2010, at 12:34 PM, whsu at sfsu.edu wrote:
>
>> Hi Kai (and Marc),
>>
>> Played with the new commits a little bit... a couple little comments...
>>
>> In the kcentersGPU Makefile, the uncommented CXXFLAGS line point to  
>>  Kai's home directory. It compiles fine once I switch back to the   
>> line that was commented out.
>>
>> In kcentersGPU.cu, in main(), the filename argument for the   
>> readData() call is hardcoded. The code works for N200K10D5.dat, but  
>>  complains for the other data file:
>>
>> ****
>>
>> Error in dataIO::readFile(): can only read floating point numbers
>>                             Please check parameters and file format
>> Data clusters:
>> Error in printClusters(). Dimension < 1 not permitted in this universe
>>
>> ****
>>
>> I can hack it to allow the user to specify a command line argument   
>> for the filename. Should I?
>>
>> Finally, it still mysteriously refuses to be profiled. It seems to   
>> run fine, over several runs (as it's supposed to), then I get this   
>> message:
>>
>>    Error in reading profiler output
>>
>> I'll poke around the documentation some more later today.
>>
>> Bill
>>
>> Quoting Kai Kohlhoff <kohlhoff at stanford.edu>:
>>
>>> Hi Bill and Marc,
>>>
>>> I just committed the new version of n-dimensional k-centers.    
>>> You'll  find it in a new directory 'kcentersGPU' (We can delete   
>>> the current  1-D version).  It is much nicer to work with after I   
>>> cleaned up the  code and followed Marc's examples for proper   
>>> doxygen documentation.
>>>
>>> See you on Friday,
>>> Kai
>>>
>>>
>>> On Feb 9, 2010, at 12:52 PM, whsu at sfsu.edu wrote:
>>>
>>>> Normally I wouldn't mind meeting that late, but I have an East   
>>>> bay  event that evening. How about 6-7?
>>>>
>>>> That's getting a bit late to meet on the SFSU campus. How about   
>>>> we  meet at my place? It's 43 Belcher Street, near the Church and  
>>>>   Market intersection.
>>>>
>>>> Bill
>>>>
>>>> Quoting Marc Sosnick <marcsosnick at mac.com>:
>>>>
>>>>> Kai/Bill:
>>>>>
>>>>> I can meet as late as 7 PM.
>>>>>
>>>>> Marc
>>>>>
>>>>>
>>>>> On Feb 9, 2010, at 11:45 AM, Kai Kohlhoff wrote:
>>>>>
>>>>>> Hi Bill and Marc,
>>>>>>
>>>>>> Sorry for the slow response time...
>>>>>>
>>>>>> To answer your question with another question:  What would be   
>>>>>> the    latest time that the two of you could comfortably do   
>>>>>> without    spoiling your weekend?  Anything works for me; but   
>>>>>> since I would    meet friends in the city for the evening   
>>>>>> afterwards it would be    great if I don't have to bridge too   
>>>>>> much time.
>>>>>>
>>>>>> Thanks,
>>>>>> Kai
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Feb 4, 2010, at 4:03 PM, whsu at sfsu.edu wrote:
>>>>>>
>>>>>>> Ok, let's do the 12th in the afternoon at SFSU. What's an easy  
>>>>>>>    time  for you, Kai?
>>>>>>>
>>>>>>> Bill
>>>>>>>
>>>>>>> Quoting Kai Kohlhoff <kohlhoff at stanford.edu>:
>>>>>>>
>>>>>>>> Hi Bill and Marc,
>>>>>>>>
>>>>>>>> The 12th is fine with me.  (Late) afternoon at  SFSU this time?
>>>>>>>>
>>>>>>>> I suggest that we use the meeting to agree on a release date   
>>>>>>>>  for    campaign 0.5 and draw up a timeline with things that   
>>>>>>>>  have to   happen by then.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Kai
>>>>>>>>
>>>>>>>>
>>>>>>>> On Feb 4, 2010, at 9:19 AM, whsu at sfsu.edu wrote:
>>>>>>>>
>>>>>>>>> Hi Kai and Marc,
>>>>>>>>>
>>>>>>>>> My meeting for the 12th just got rescheduled (happens a lot   
>>>>>>>>>  for    these people,
>>>>>>>>> sigh). So if the 12th will work better, I'm fine with that.
>>>>>>>>>
>>>>>>>>> I'll try to check out the new commits with the profiler in   
>>>>>>>>> the    next  few days...
>>>>>>>>>
>>>>>>>>> Bill
>>>>>>>>>
>>>>>>>>> Quoting Kai Kohlhoff <kohlhoff at stanford.edu>:
>>>>>>>>>
>>>>>>>>>> Hi Bill and Marc,
>>>>>>>>>>
>>>>>>>>>> Welcome back from Hong Kong.
>>>>>>>>>>
>>>>>>>>>> Wed 10th sounds good to me for a meeting, although I have a  
>>>>>>>>>>      Seminar  from 2.15pm - 3.30pm, so it would be great if  
>>>>>>>>>>  we  could    meet here at  Stanford some time after that.
>>>>>>>>>>
>>>>>>>>>> I will commit a new version of the kcenters code tomorrow    
>>>>>>>>>> that    might  not have the problem that you are    
>>>>>>>>>> experiencing.   Perhaps   you could  wait and try that new   
>>>>>>>>>>  version.
>>>>>>>>>>
>>>>>>>>>> Thanks,
>>>>>>>>>> Kai
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>> On Feb 1, 2010, at 7:00 PM, whsu at sfsu.edu wrote:
>>>>>>>>>>
>>>>>>>>>>> Hi Kai (and Marc),
>>>>>>>>>>>
>>>>>>>>>>> Sorry about spacing out on getting back to you. I'm busy    
>>>>>>>>>>> with a  gig on
>>>>>>>>>>> the 19th; we should probably meet before the big meeting   
>>>>>>>>>>> on  the  22nd.
>>>>>>>>>>> We've been meeting on Fridays, but I have another meeting   
>>>>>>>>>>> on   the 12th
>>>>>>>>>>> in the afternoon as well. Will say Wed the 10th work?
>>>>>>>>>>>
>>>>>>>>>>> I'm getting stuck in the profiling; keep getting this    
>>>>>>>>>>> cryptic    error message
>>>>>>>>>>> with the kcenters code, while the profiler works with some  
>>>>>>>>>>>   of   the other
>>>>>>>>>>> examples that I've tried. Perhaps it's because of the   
>>>>>>>>>>> input    redirection;
>>>>>>>>>>> I'll try tweaking the code a bit.
>>>>>>>>>>>
>>>>>>>>>>> Bill
>>>>>>>>>>
>>>>>>>>>> -----------------------------------------------------
>>>>>>>>>> Kai Kohlhoff, PhD
>>>>>>>>>> Stanford University
>>>>>>>>>> School of Medicine, Bioengineering
>>>>>>>>>> Stanford, CA 94305-5448, USA
>>>>>>>>>> T: ++1 (650) 724 1575
>>>>>>>>>> E: kohlhoff at stanford.edu
>>>>>>>>>>
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> _______________________________________________
>>>>>>>>> Campaign-news mailing list
>>>>>>>>> Campaign-news at simtk.org
>>>>>>>>> https://simtk.org/mailman/listinfo/campaign-news
>>>>>>>>
>>>>>>>> -----------------------------------------------------
>>>>>>>> Kai Kohlhoff, PhD
>>>>>>>> Stanford University
>>>>>>>> School of Medicine, Bioengineering
>>>>>>>> Stanford, CA 94305-5448, USA
>>>>>>>> T: ++1 (650) 724 1575
>>>>>>>> E: kohlhoff at stanford.edu
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> Campaign-news mailing list
>>>>>>> Campaign-news at simtk.org
>>>>>>> https://simtk.org/mailman/listinfo/campaign-news
>>>>>>
>>>>>> -----------------------------------------------------
>>>>>> Kai Kohlhoff, PhD
>>>>>> Stanford University
>>>>>> School of Medicine, Bioengineering
>>>>>> Stanford, CA 94305-5448, USA
>>>>>> T: ++1 (650) 724 1575
>>>>>> E: kohlhoff at stanford.edu
>>>>>>
>>>>>> _______________________________________________
>>>>>> Campaign-news mailing list
>>>>>> Campaign-news at simtk.org
>>>>>> https://simtk.org/mailman/listinfo/campaign-news
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> Campaign-news mailing list
>>>> Campaign-news at simtk.org
>>>> https://simtk.org/mailman/listinfo/campaign-news
>>>
>>> -----------------------------------------------------
>>> Kai Kohlhoff, PhD
>>> Stanford University
>>> School of Medicine, Bioengineering
>>> Stanford, CA 94305-5448, USA
>>> T: ++1 (650) 724 1575
>>> E: kohlhoff at stanford.edu
>>>
>>>
>>
>>
>>
>> _______________________________________________
>> Campaign-news mailing list
>> Campaign-news at simtk.org
>> https://simtk.org/mailman/listinfo/campaign-news
>
> -----------------------------------------------------
> Kai Kohlhoff, PhD
> Stanford University
> School of Medicine, Bioengineering
> Stanford, CA 94305-5448, USA
> T: ++1 (650) 724 1575
> E: kohlhoff at stanford.edu
>
>





More information about the Campaign-news mailing list