Bill Allombert on Tue, 20 Jan 2015 23:55:31 +0100


[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]

Re: Stack size bugs and parisizemax


On Tue, Jan 20, 2015 at 10:54:08PM +0100, Jeroen Demeyer wrote:
> On 2015-01-20 21:32, Bill Allombert wrote:
> >The purpose of parisizemax was to allow to have a nearly unlimited stack
> >(limited only by the available hardware) without increasing the actual
> >memory usage (in the sense that the final stack size will be close to the
> >minimum stack size for which the computation would have succeed, up to a factor
> >of two).
> I don't think that parisizemax should behave like that. If an
> algorithm can work significantly faster by using a larger stack size
> than the minimum, it should do that.

I think we all agree on that, but this is no contradictory with the above:
performing more GC should still be OK if it does not slow down the computation
much.

We also agree that garbage collection in second-level loops is problematic
when doing linear algebra (because garbage collection is very slow then).

Do you think supporting what you call 'desesparation mode' to be valuable ?
So far Karim and I have been less than convinced.

Cheers,
Bill.