Sumant,

I think this has something to do with the initial contents
(time (type-of (make-foreign-array 'double-float :dimensions '(1024 1024))))
Evaluation took:
  0.001 seconds of real time
  0.000000 seconds of total run time (0.000000 user, 0.000000 system)
  0.00% CPU
  341,884 processor cycles
  0 bytes consed
 
MATRIX-DOUBLE-FLOAT
Seems pretty snappy.  Generally, I'm not too happy with the idea of setting values from a list, at least for large lists.  I tried to find an alternate way to do this, but I also wanted to mimic the arguments to make-array, so that's why it's in there.   I presume that your test is contrived and that you wouldn't really make a list of 1 million elements and then load it into a foreign array.  Can you try a problem where you're not using :initial-contents?  If you don't like to make an array with indeterminate contents, try this:

(time (type-of (map-grid :source '* :source-dims '(1024 1024) :destination-specification '((foreign-array 1024 1024) double-float))))
Evaluation took:
  2.487 seconds of real time
  2.490000 seconds of total run time (2.490000 user, 0.000000 system)
  [ Run times consist of 0.220 seconds GC time, and 2.270 seconds non-GC time. ]
  100.12% CPU
  5,292,435,364 processor cycles
  268,462,848 bytes consed
 
MATRIX-DOUBLE-FLOAT

(By the way, all make-foreign-array does is call make-grid,  i.e., it's less "direct" than make-grid.)

Liam

On Wed, Apr 27, 2011 at 9:31 AM, Sumant Oemrawsingh <soemraws@xs4all.nl> wrote:
Hi,

I ran into a problem making foreign arrays. I have a two lists of
lists of floats called *data-50* and *data-1024*, which are to be used
as the initial contents of a matrix (first is 50x50, second
1024x1024). I do the following:

CL-USER> (time (type-of (make-grid '((array) single-float) :initial-contents *data-50*)))
Evaluation took:
 0.000 seconds of real time
 0.000000 seconds of total run time (0.000000 user, 0.000000 system)
 100.00% CPU
 228,096 processor cycles
 27,744 bytes consed

(SIMPLE-ARRAY SINGLE-FLOAT (50 50))
CL-USER> (time (type-of (make-grid '((foreign-array) single-float) :initial-contents *data-50*)))
Evaluation took:
 0.036 seconds of real time
 0.026995 seconds of total run time (0.026995 user, 0.000000 system)
 75.00% CPU
 79,104,003 processor cycles
 294,688 bytes consed

MATRIX-SINGLE-FLOAT


With a 50x50 array, everything works out fine. However, when moving to
the 1024x1024 data:


CL-USER> (time (type-of (make-grid '((array) single-float) :initial-contents *data-1024*)))
Evaluation took:
 0.036 seconds of real time
 0.035995 seconds of total run time (0.034995 user, 0.001000 system)
 100.00% CPU
 80,301,650 processor cycles
 4,194,320 bytes consed

(SIMPLE-ARRAY SINGLE-FLOAT (1024 1024))
CL-USER> (time (type-of (make-grid '((foreign-array) single-float) :initial-contents *data-1024*)))
Evaluation took:
 686.167 seconds of real time
 683.878034 seconds of total run time (683.608075 user, 0.269959 system)
 [ Run times consist of 0.196 seconds GC time, and 683.683 seconds non-GC time. ]
 99.67% CPU
 3 forms interpreted
 1,502,208,488,440 processor cycles
 59,838,352 bytes consed

 before it was aborted by a non-local transfer of control.

; Evaluation aborted on NIL.


After over 10 minutes of intensive work, I gave up and aborted. When
using make-foreign-array directly, instead of make-grid, I get the
same problem. I would understand that foreign arrays take a bit more
time to make; in the *data-50* example, about an order of magnitude
more bytes were consed to make the foreign array (and needing more
processor cycles).

While this is not very thorough, I would expect there to be some
linear relation between the array size and the time it
takes. E.g. 50x50 foreign array takes less than a 0.1 second, so I
would expect 1024x1024 to take less than 42.0 seconds. While that is
still a long time for a normal-sized array, clearly, this is not the
case. I can run some more tests later to find out where the problem
lies. Or is this a known bug or limitation for foreign arrays and
sbcl? Or am I just doing something wrong here?

Thanks,
Sumant

--
Sumant Oemrawsingh

_______________________________________________
GSLL-devel mailing list
GSLL-devel@common-lisp.net
http://common-lisp.net/cgi-bin/mailman/listinfo/gsll-devel