Fwd: [RFC][gomp4] Offloading patches (2/3): Add tables generation

Ilya Verbin iverbin@gmail.com
Wed Jan 29 16:06:00 GMT 2014


2014/1/29 Bernd Schmidt <bernds@codesourcery.com>:
> I was worried the answer was going to be something like this. These are
> produced by two different compilers/linkers, aren't they? This seems really
> fragile to me and a recipe for hard-to-debug silent failures.

Yes.

> First we let the driver compile and load all the ptx code, then we'd call a
> cuda library function with the name of the function passed as a string. So I
> guess the host table would contain pointers to the host version of the
> functions, and the target table would contain a pointer to the name.

We also wanted to use function names at the beginning of our work [1].
Then Jakub noticed that the names aren't necessarily unique: [2], [3], [4], etc.
This led to the current approach of host/target address tables.

>
> Bernd
>

[1] http://gcc.gnu.org/ml/gcc/2013-08/msg00251.html
[2] http://gcc.gnu.org/ml/gcc-patches/2013-09/msg00250.html
[3] http://gcc.gnu.org/ml/gcc/2013-09/msg00146.html
[4] http://gcc.gnu.org/ml/gcc/2013-10/msg00184.html

  -- Ilya



More information about the Gcc-patches mailing list