https://gcc.gnu.org/bugzilla/show_bug.cgi?id=115795
--- Comment #8 from Jordi Sala <jordi.sala at semidynamics dot com> ---
(In reply to JuzheZhong from comment #7)
> (In reply to Jordi Sala from comment #6)
> > Perfect, that's what I was looking for. I'm thinking of adding a way to tell
> > GCC to minimize, maximize or preserve SEW on vsetvl expand. Like
> > -mrvv-vsetvl-sew={maximize,minimize,preserve}
>
> No. I don't think you need to add an option here.
>
> I think you should change the default according to the vsetvl intrinsic
> builtin
> sematic. For example, set SEW/LMUL as e32m2 if it is __riscv_vsetvl_e32m2, or
> set it as e64m1 if it is __riscv_vsetvl_e64m1 which will make codegen more
> straightforward.
>
>
> if (TARGET_XTHEADVECTOR)
> {
> machine_mode inner_mode = GET_MODE_INNER (mode);
> /* SEW. */
> e.add_input_operand (Pmode,
> gen_int_mode (GET_MODE_BITSIZE (inner_mode), Pmode));
> /* LMUL. */
> e.add_input_operand (Pmode,
> gen_int_mode (get_vlmul (mode), Pmode));
> }
> else
> {
> /* Normalize same RATO (SEW/LMUL) into same vsetvl instruction.
>
> - e8,mf8/e16,mf4/e32,mf2/e64,m1 --> e8mf8
> - e8,mf4/e16,mf2/e32,m1/e64,m2 --> e8mf4
> - e8,mf2/e16,m1/e32,m2/e64,m4 --> e8mf2
> - e8,m1/e16,m2/e32,m4/e64,m8 --> e8m1
> - e8,m2/e16,m4/e32,m8 --> e8m2
> - e8,m4/e16,m8 --> e8m4
> - e8,m8 --> e8m8
> */
> /* SEW. */
> e.add_input_operand (Pmode, gen_int_mode (8, Pmode));
>
> /* LMUL. */
> machine_mode e8_mode
> = get_vector_mode (QImode, GET_MODE_NUNITS (mode)).require ();
> e.add_input_operand (Pmode, gen_int_mode (get_vlmul (e8_mode), Pmode));
> }
>
> into:
>
> machine_mode inner_mode = GET_MODE_INNER (mode);
> /* SEW. */
> e.add_input_operand (Pmode,
> gen_int_mode (GET_MODE_BITSIZE (inner_mode), Pmode));
> /* LMUL. */
> e.add_input_operand (Pmode,
> gen_int_mode (get_vlmul (mode), Pmode));
I'm closing the issue as this looks to me the best way to fix it. Thanks for
all the support.
