[Bug target/61551] New: [NEON] alter costs to allow use of post-indexed addressing modes for VLD{2..4}/VST{2..4}

[Bug target/61551] New: [NEON] alter costs to allow use of post-indexed addressing modes for VLD{2..4}/VST{2..4}

[cbaylis at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61551

            Bug ID: 61551
           Summary: [NEON] alter costs to allow use of post-indexed
                    addressing modes for VLD{2..4}/VST{2..4}
           Product: gcc
           Version: 4.10.0
            Status: UNCONFIRMED
          Severity: enhancement
          Priority: P3
         Component: target
          Assignee: unassigned at gcc dot gnu.org
          Reporter: cbaylis at gcc dot gnu.org
                CC: ramana.radhakrishnan at arm dot com
            Target: arm-unknown-linux-gnueabi

Created attachment 32967
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=32967&action=edit
test for NEON addressing modes

The attached test case demonstrates that GCC does not exploit the post-indexed
addressing mode for NEON structure loads and stores: VLDn, VSTn where n>=2.

Generated code for VLD1/VST1 (using desired post-indexed addressing)

test_ld1:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 0, uses_anonymous_args = 0
        @ link register save eliminated.
        vld1.8  {d16}, [r0], r1
        vst1.8  {d16}, [r0], r1
        vld1.8  {d16}, [r0], r1
        vst1.8  {d16}, [r0], r1
        vld1.8  {d16}, [r0], r1
        vst1.8  {d16}, [r0]
        bx      lr

Generated code for VLD2:
test_ld2:
        @ args = 0, pretend = 0, frame = 0
        @ frame_needed = 0, uses_anonymous_args = 0
        @ link register save eliminated.
        adds    r3, r0, r1
        vld2.8  {d16-d17}, [r0]
        adds    r0, r3, r1
        adds    r2, r0, r1
        vst2.8  {d16-d17}, [r3]
        adds    r3, r2, r1
        vld2.8  {d16-d17}, [r0]
        add     r1, r1, r3
        vst2.8  {d16-d17}, [r2]
        vld2.8  {d16-d17}, [r3]
        vst2.8  {d16-d17}, [r1]
        bx      lr


A proof of concept patch is posted at:
https://gcc.gnu.org/ml/gcc-patches/2014-06/msg01361.html

[Bug target/61551] [NEON] alter costs to allow use of post-indexed addressing modes for VLD{2..4}/VST{2..4}

[ramana at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61551

Ramana Radhakrishnan <ramana at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |missed-optimization
             Status|UNCONFIRMED                 |NEW
   Last reconfirmed|                            |2014-07-25
                 CC|                            |ramana at gcc dot gnu.org
   Target Milestone|---                         |4.10.0
     Ever confirmed|0                           |1

[Bug target/61551] [NEON] alter costs to allow use of post-indexed addressing modes for VLD{2..4}/VST{2..4}

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61551

Jakub Jelinek <jakub at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|5.0                         |5.2

--- Comment #1 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
GCC 5.1 has been released.

[Bug target/61551] [NEON] alter costs to allow use of post-indexed addressing modes for VLD{2..4}/VST{2..4}

[ramana at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61551

Ramana Radhakrishnan <ramana at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|5.2                         |6.0

--- Comment #2 from Ramana Radhakrishnan <ramana at gcc dot gnu.org> ---
This really is 6.0 now - I can't see this as being eligible for the 5.x
branches.

[Bug target/61551] [NEON] alter costs to allow use of post-indexed addressing modes for VLD{2..4}/VST{2..4}

[cbaylis at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61551

cbaylis at gcc dot gnu.org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #6 from cbaylis at gcc dot gnu.org ---

This was fixed in
https://gcc.gnu.org/git/?p=gcc.git;a=commit;h=612ea540d2e1fe5eae5213823a3cec0d3d864fb2

 This patch adds support for modelling the varying costs of different
addressing modes.

This patch adds support for modelling the varying costs of
different addressing modes. The generic cost table treats
all addressing modes as having equal cost.

gcc/ChangeLog:

2017-11-23  Charles Baylis  <charles.baylis@linaro.org>

        * config/arm/arm-protos.h (enum arm_addr_mode_op): New.
        (struct addr_mode_cost_table): New.
        (struct tune_params): Add field addr_mode_costs.
        * config/arm/arm.c (generic_addr_mode_costs): New.
        (arm_slowmul_tune): Initialise addr_mode_costs field.
        (arm_fastmul_tune): Likewise.
        (arm_strongarm_tune): Likewise.
        (arm_xscale_tune): Likewise.
        (arm_9e_tune): Likewise.
        (arm_marvell_pj4_tune): Likewise.
        (arm_v6t2_tune): Likewise.
        (arm_cortex_tune): Likewise.
        (arm_cortex_a8_tune): Likewise.
        (arm_cortex_a7_tune): Likewise.
        (arm_cortex_a15_tune): Likewise.
        (arm_cortex_a35_tune): Likewise.
        (arm_cortex_a53_tune): Likewise.
        (arm_cortex_a57_tune): Likewise.
        (arm_exynosm1_tune): Likewise.
        (arm_xgene1_tune): Likewise.
        (arm_cortex_a5_tune): Likewise.
        (arm_cortex_a9_tune): Likewise.
        (arm_cortex_a12_tune): Likewise.
        (arm_cortex_a73_tune): Likewise.
        (arm_v7m_tune): Likewise.
        (arm_cortex_m7_tune): Likewise.
        (arm_v6m_tune): Likewise.
        (arm_fa726te_tune): Likewise.
        (arm_mem_costs): Use table lookup to calculate cost of addressing
        mode.

From-SVN: r255112