[Haskell-cafe] Blitting one IOUArray into another

[Bulat Ziganshin]

Hello Denis,

Wednesday, January 7, 2009, 6:56:36 PM, you wrote:

memory allocated for i :)))

each new copy of i needs one word. the situation was much worse with
Int64, of course :)


> Hi all,

> I'm seeing a lot of unexpected memory allocation with some simple code that
> copies the contents of one vector (IOUArray Int Int64) into another of the
> same type and dimensions.  In particular, profiling reveals that `copyInto'
> is allocating oodles and oodles of memory.

> My small test case creates two 50000-element arrays and performs 10000
> copies from one into the other.  Since the elements are Int64s and the
> arrays are unboxed, each array should be

>     50000 elements * 8 bytes per element = 400,000 bytes

> and so the arrays should only take 800,000 bytes total.  I understand
> there's some overhead for thunks and whatnot, but the profiler reported
> allocation is around 40 billion bytes.  And almost all of that allocation is
> in my copying function, not in main (main allocates the arrays).

> I've attached two versions of the code, just for comparison.  The only
> difference is the way the copying is done.  One calls `getBounds' to figure
> out the bounds, and one is given the bounds for copying.  They're both about
> the same speed and allocation (originally I was using IOUArray Int64 Int64,
> and there was a much greater allocation difference between the two versions;
> but that went away.  Oh well).

> So, does anyone know why copying takes so much allocation?  I expect there
> is _some_ way to just move memory from one array to another, like a memcpy
> -- how can I do that?
>                               Denis




-- 
Best regards,
 Bulat                            mailto:Bulat.Ziganshin at gmail.com