On Jul 5, 2008, at 8:22 PM, Slava Pestov wrote:

>> The >double-blas-matrix step takes a while because it's implemented a
>> bit naively, so the second test still takes a couple seconds overall.
>
> Any hope of speeding that up as well?

Most of the time is spent in >c-double-array, which ultimately winds  
up calling set-float-nth 4000^2 times. If the argument to >double-blas- 
matrix was a sequence of float-arrays, I could special-case it so that  
it memcpy'd row-by-row to the destination buffer. But then populating  
4,000 4,000-element float-arrays isn't terribly fast either.

If you're lucky enough to have a matrix or array of unboxed floats  
already in memory, you can use <double-blas-matrix>, <double-blas- 
vector> etc. to wrap the memory directly.

-Joe

-------------------------------------------------------------------------
Sponsored by: SourceForge.net Community Choice Awards: VOTE NOW!
Studies have shown that voting for your favorite open source project,
along with a healthy diet, reduces your potential for chronic lameness
and boredom. Vote Now at http://www.sourceforge.net/community/cca08
_______________________________________________
Factor-talk mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/factor-talk

Reply via email to