Wow, that is exactly what I wanted, it is "equivalent" with +/ . *
applied the unpartitioned matrices A and B.
Can you explain why you've added /&, ?
Thanks for the help!

On Mon, Nov 7, 2011 at 9:50 PM, Ric Sherlock <[email protected]> wrote:
> or did you want
>   tA +/ .*&.>/&, tB
> ┌─────────┬─────────┬─────────┐
> │330 345  │360  375 │ 390  405│
> │870 921  │972 1023 │1074 1125│
> ├─────────┼─────────┼─────────┤
> │1410 1497│1584 1671│1758 1845│
> │1950 2073│2196 2319│2442 2565│
> ├─────────┼─────────┼─────────┤
> │2490 2649│2808 2967│3126 3285│
> │3030 3225│3420 3615│3810 4005│
> └─────────┴─────────┴─────────┘
>
>
> On Tue, Nov 8, 2011 at 9:40 AM, Ric Sherlock <[email protected]> wrote:
>> (,tA) +/ .*&.> (,tB)
>>
>> On Tue, Nov 8, 2011 at 9:09 AM, Jan Jacobs <[email protected]> wrote:
>>> ls,
>>> for educational purposes I want to breakdown a CUDA matrix
>>> multiplication in smaller parts.
>>> What I do not understand is that for B=:A =: i.6 6
>>>  tA =: (2 2,:2 2)<;._3 A
>>>  tB =: (2 2,:2 2)<;._3 B
>>>  tA (+/ . *)&.> tB
>>> works fine, but
>>>  tA =: (2 6,:2 6)<;._3 A
>>>  tB =: (6 2,:6 2)<;._3 B
>>>  tA (+/ . *)&.> tB
>>> does not:
>>> |length error
>>> |   tA    (+/ .*)"1&.>tB
>>>
>>> The shapes of the various boxes are as expected.
>>> What is going wrong?
>>> Thanks in advance for the help.
>>> Jan.
>>>
>>>
>>> --
>>> Jan Jacobs
>>> Esdoornstraat 33
>>> 5995AN Kessel
>>> W: www.sommaps.com
>>> T: +31 77 462 1887
>>> M: +31 6 23 82 55 21
>>> E: [email protected]
>>> ----------------------------------------------------------------------
>>> For information about J forums see http://www.jsoftware.com/forums.htm
>>>
>>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm



-- 
Jan Jacobs
Esdoornstraat 33
5995AN Kessel
W: www.sommaps.com
T: +31 77 462 1887
M: +31 6 23 82 55 21
E: [email protected]
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to