Re: [Starpu-devel] [PATCH 0/7][Review needed] A first step towards the factorization of heft and dm*.

[Cyril Roelandt <cyril.roelandt@inria.fr>]

On 10/19/2012 05:09 PM, Samuel Thibault wrote:
> Cyril Roelandt, le Tue 16 Oct 2012 23:45:20 +0200, a écrit :
> > These seven patches look quite good to me (the test suite is ok on Hannibal),
> > but I'd like to know what you think before pushing them.
>
> You could check at least with cholesky that there are no performance
> regressions.  We'll have to check with magma anyway.

Following our discussion on #starpu, I'm sending a second version of 
this patch series.

Patch 1: no changes.
Patch 2: I used _starpu_fifo_push_sorted_task(), that I made available 
to all schedulers.
Patch 3: I added a few fixes in the pre_exec_hook of heft.
Patch 4: unchanged.
Patch 5: I used exp_end[worker][nimpl] rather than fifo->exp_end in 
compute_all_performance_predicitons().
Patch 6: unchanged.
Patch 7: dropped. We'll take care of task bundles in dm* later.



Cyril Roelandt.