Re: [Bug target/65614] [5 Regression] PowerPC VSX systems should use XSCPSGNDP to copy scalar fp data to/from Altivec registers

Re: [Bug target/65614] [5 Regression] PowerPC VSX systems should use XSCPSGNDP to copy scalar fp data to/from Altivec registers

[Michael Meissner]

On Fri, Apr 03, 2015 at 07:19:24PM +0000, dje at gcc dot gnu.org wrote:
> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65614
> 
> --- Comment #4 from David Edelsohn <dje at gcc dot gnu.org> ---
> Comment on attachment 35227
>   --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35227
> Proposed patch to fix the problem
> 
> The FLOAT_EXTEND cost should be based on the processor tuning, not the ISA.

This patch moves the decision to the cost structure based on the processor
tuning.  Is it ok to install?

[gcc]
2015-04-06  Michael Meissner  <meissner@linux.vnet.ibm.com>

	PR target/65614
	* config/rs6000/rs6000.c (struct processor_costs): Add cost field
	for SF->DF conversions to make FLOAT_EXTEND more expensive, so
	that LFD is used to load double constants instead of LFS.  Add
	defaults for all costs structures.  Add comments for missing
	initialization fields.
	(size32_cost): Likewise.
	(size64_cost): Likewise.
	(rs64a_cost): Likewise.
	(mpccore_cost): Likewise.
	(ppc403_cost): Likewise.
	(ppc405_cost): Likewise.
	(ppc440_cost): Likewise.
	(ppc476_cost): Likewise.
	(ppc601_cost): Likewise.
	(ppc603_cost): Likewise.
	(ppc604_cost): Likewise.
	(ppc604e_cost): Likewise.
	(ppc620_cost): Likewise.
	(ppc630_cost): Likewise.
	(ppccell_cost): Likewise.
	(ppc750_cost): Likewise.
	(ppc7450_cost): Likewise.
	(ppc8540_cost): Likewise.
	(ppce300c2c3_cost): Likewise.
	(ppce500mc_cost): Likewise.
	(ppce500mc64_cost): Likewise.
	(ppce5500_cost): Likewise.
	(ppce6500_cost): Likewise.
	(titan_cost): Likewise.
	(power4_cost): Likewise.
	(power6_cost): Likewise.
	(power7_cost): Likewise.
	(power8_cost): Likewise.
	(ppca2_cost): Likewise.
	(rs6000_rtx_costs): Make FLOAT_EXTEND use SFDF_convert field.

	* config/rs6000/rs6000.md (extendsfdf2_fpr): Generate XSCPSGNDP
	instead of XXLOR to copy SFmode to clear out dirty bits created
	when SFmode denormals are generated.
	(mov<mode>_hardfloat, FMOVE32 case): Likewise.
	(truncdfsf2_fpr): Add support for ISA 2.07 XSRSP instruction.

[gcc/testsuite]
2015-04-06  Michael Meissner  <meissner@linux.vnet.ibm.com>

	PR target/65614
	* gcc.target/powerpc/compress-float-ppc-pic.c: Run test on power5
	to get floating point compression.
	* gcc.target/powerpc/compress-foat-ppc.c: Likewise.

-- 
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: meissner@linux.vnet.ibm.com, phone: +1 (978) 899-4797

Re: [Bug target/65614] [5 Regression] PowerPC VSX systems should use XSCPSGNDP to copy scalar fp data to/from Altivec registers

[David Edelsohn]

On Mon, Apr 6, 2015 at 1:10 PM, Michael Meissner
<meissner@linux.vnet.ibm.com> wrote:
> On Fri, Apr 03, 2015 at 07:19:24PM +0000, dje at gcc dot gnu.org wrote:
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65614
>>
>> --- Comment #4 from David Edelsohn <dje at gcc dot gnu.org> ---
>> Comment on attachment 35227
>>   --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=35227
>> Proposed patch to fix the problem
>>
>> The FLOAT_EXTEND cost should be based on the processor tuning, not the ISA.
>
> This patch moves the decision to the cost structure based on the processor
> tuning.  Is it ok to install?
>
> [gcc]
> 2015-04-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
>
>         PR target/65614
>         * config/rs6000/rs6000.c (struct processor_costs): Add cost field
>         for SF->DF conversions to make FLOAT_EXTEND more expensive, so
>         that LFD is used to load double constants instead of LFS.  Add
>         defaults for all costs structures.  Add comments for missing
>         initialization fields.
>         (size32_cost): Likewise.
>         (size64_cost): Likewise.
>         (rs64a_cost): Likewise.
>         (mpccore_cost): Likewise.
>         (ppc403_cost): Likewise.
>         (ppc405_cost): Likewise.
>         (ppc440_cost): Likewise.
>         (ppc476_cost): Likewise.
>         (ppc601_cost): Likewise.
>         (ppc603_cost): Likewise.
>         (ppc604_cost): Likewise.
>         (ppc604e_cost): Likewise.
>         (ppc620_cost): Likewise.
>         (ppc630_cost): Likewise.
>         (ppccell_cost): Likewise.
>         (ppc750_cost): Likewise.
>         (ppc7450_cost): Likewise.
>         (ppc8540_cost): Likewise.
>         (ppce300c2c3_cost): Likewise.
>         (ppce500mc_cost): Likewise.
>         (ppce500mc64_cost): Likewise.
>         (ppce5500_cost): Likewise.
>         (ppce6500_cost): Likewise.
>         (titan_cost): Likewise.
>         (power4_cost): Likewise.
>         (power6_cost): Likewise.
>         (power7_cost): Likewise.
>         (power8_cost): Likewise.
>         (ppca2_cost): Likewise.
>         (rs6000_rtx_costs): Make FLOAT_EXTEND use SFDF_convert field.
>
>         * config/rs6000/rs6000.md (extendsfdf2_fpr): Generate XSCPSGNDP
>         instead of XXLOR to copy SFmode to clear out dirty bits created
>         when SFmode denormals are generated.
>         (mov<mode>_hardfloat, FMOVE32 case): Likewise.
>         (truncdfsf2_fpr): Add support for ISA 2.07 XSRSP instruction.
>
> [gcc/testsuite]
> 2015-04-06  Michael Meissner  <meissner@linux.vnet.ibm.com>
>
>         PR target/65614
>         * gcc.target/powerpc/compress-float-ppc-pic.c: Run test on power5
>         to get floating point compression.
>         * gcc.target/powerpc/compress-foat-ppc.c: Likewise.

Okay.

What's the source of COSTS_N_INSNS(3) value?

Thanks, David

Re: [Bug target/65614] [5 Regression] PowerPC VSX systems should use XSCPSGNDP to copy scalar fp data to/from Altivec registers

[Michael Meissner]

On Mon, Apr 06, 2015 at 07:56:53PM -0400, David Edelsohn wrote:
> Okay.
> 
> What's the source of COSTS_N_INSNS(3) value?

It is the basic cost of a FP insn (i.e. it duplicates what rs6000_cost->fp
would be -- all of the simple FP ops are this value, XSADDDP, XSMULDP,
etc. including the two move insns XSCPSGNDP and FMR).  On the power7 and
power8, each basic FP op is 6 cycles, and all of the costs use the fact that
COSTS_N_INSNS multiplies by 2.

Obviously in the grand scheme of things, we need to rewrite the cost
infrastructure, and bring it up to date, but for now, I used the values for the
other insns.

-- 
Michael Meissner, IBM
IBM, M/S 2506R, 550 King Street, Littleton, MA 01460-6245, USA
email: meissner@linux.vnet.ibm.com, phone: +1 (978) 899-4797