Hi

On Tuesday, October 21, 2014 10:20:25 PM UTC+2, Stefan Karpinski wrote:
>
> On Tue, Oct 21, 2014 at 4:13 PM, David van Leeuwen <[email protected] 
> <javascript:>> wrote:
>
>> Does anybody know if `a' * b` is translated to a single BLAS call with 
>> the correct transposition options set, or that this is translated to a 
>> transposition followed by a BLAS call?
>
>
> Yes, `a' * b` is parsed specially and calls the wonderfully named Ac_mul_B 
> function, which calls BLAS without doing an explicit conjugate transpose. 
> We would like to change this by making (conjugate) transpose lazy so that 
> both version use the same BLAS call.
>

That is very cool---it makes the code better readable. 

I also noticed that it helped me to write out some broadcast() statements, 
again both in speed and memory allocation---this is of course along the 
philosophy of Julia.  However, I really needed to assert the type in all 
involved arrays before using them in the for-loops, or I would loose a 
factor 10 or thereabouts in speed.  

---david

Reply via email to