Wilco Dijkstra <wilco.dijks...@arm.com> writes:
> Use UZP1 instead of INS when combining low and high halves of vectors.
> UZP1 has 3 operands which improves register allocation, and is faster on
> some microarchitectures.
>
> Passes regress & bootstrap, OK for commit?
OK, thanks. We can add core-specific tuning later if a supported core
strongly prefers INS for some reason, but I agree that the three-address
nature of UZP1 makes it the better default choice.
Richard
>
> gcc:
> * config/aarch64/aarch64-simd.md (aarch64_combine_internal<mode>):
> Use UZP1 instead of INS.
> (aarch64_combine_internal_be<mode>): Likewise.
>
> gcc/testsuite:
> * gcc.target/aarch64/ldp_stp_16.c: Update to check for UZP1.
> * gcc.target/aarch64/pr109072_1.c: Likewise.
> * gcc.target/aarch64/vec-init-14.c: Likewise.
> * gcc.target/aarch64/vec-init-9.c: Likewise.
>
> ---
>
> diff --git a/gcc/config/aarch64/aarch64-simd.md
> b/gcc/config/aarch64/aarch64-simd.md
> index
> f8bb973a278c7964f3e3a4f7154a0ab62214b7cf..16b7445d9f72f77a98ab262e21fd24e6cc97eba0
> 100644
> --- a/gcc/config/aarch64/aarch64-simd.md
> +++ b/gcc/config/aarch64/aarch64-simd.md
> @@ -4388,7 +4388,7 @@
> && (register_operand (operands[0], <VDBL>mode)
> || register_operand (operands[2], <MODE>mode))"
> {@ [ cons: =0 , 1 , 2 ; attrs: type , arch ]
> - [ w , 0 , w ; neon_ins<dblq> , simd ]
> ins\t%0.<single_type>[1], %2.<single_type>[0]
> + [ w , w , w ; neon_permute<dblq> , simd ]
> uzp1\t%0.2<single_type>, %1.2<single_type>, %2.2<single_type>
> [ w , 0 , ?r ; neon_from_gp<dblq> , simd ]
> ins\t%0.<single_type>[1], %<single_wx>2
> [ w , 0 , ?r ; f_mcr , * ]
> fmov\t%0.d[1], %2
> [ w , 0 , Utv ; neon_load1_one_lane<dblq> , simd ]
> ld1\t{%0.<single_type>}[1], %2
> @@ -4407,7 +4407,7 @@
> && (register_operand (operands[0], <VDBL>mode)
> || register_operand (operands[2], <MODE>mode))"
> {@ [ cons: =0 , 1 , 2 ; attrs: type , arch ]
> - [ w , 0 , w ; neon_ins<dblq> , simd ]
> ins\t%0.<single_type>[1], %2.<single_type>[0]
> + [ w , w , w ; neon_permute<dblq> , simd ]
> uzp1\t%0.2<single_type>, %1.2<single_type>, %2.2<single_type>
> [ w , 0 , ?r ; neon_from_gp<dblq> , simd ]
> ins\t%0.<single_type>[1], %<single_wx>2
> [ w , 0 , ?r ; f_mcr , * ]
> fmov\t%0.d[1], %2
> [ w , 0 , Utv ; neon_load1_one_lane<dblq> , simd ]
> ld1\t{%0.<single_type>}[1], %2
> diff --git a/gcc/testsuite/gcc.target/aarch64/ldp_stp_16.c
> b/gcc/testsuite/gcc.target/aarch64/ldp_stp_16.c
> index
> f1f46e051a86d160a7f7f14872108da87b444ca1..95835aa2eb41c289e7b74f19bb56cf6fa23a3045
> 100644
> --- a/gcc/testsuite/gcc.target/aarch64/ldp_stp_16.c
> +++ b/gcc/testsuite/gcc.target/aarch64/ldp_stp_16.c
> @@ -80,16 +80,16 @@ CONS2_FN (2, float);
>
> /*
> ** cons2_4_float: { target aarch64_little_endian }
> -** ins v0.s\[1\], v1.s\[0\]
> -** stp d0, d0, \[x0\]
> -** stp d0, d0, \[x0, #?16\]
> +** uzp1 v([0-9])\.2s, v0\.2s, v1\.2s
> +** stp d\1, d\1, \[x0\]
> +** stp d\1, d\1, \[x0, #?16\]
> ** ret
> */
> /*
> ** cons2_4_float: { target aarch64_big_endian }
> -** ins v1.s\[1\], v0.s\[0\]
> -** stp d1, d1, \[x0\]
> -** stp d1, d1, \[x0, #?16\]
> +** uzp1 v([0-9])\.2s, v1\.2s, v0\.2s
> +** stp d\1, d\1, \[x0\]
> +** stp d\1, d\1, \[x0, #?16\]
> ** ret
> */
> CONS2_FN (4, float);
> @@ -125,8 +125,8 @@ CONS4_FN (2, float);
>
> /*
> ** cons4_4_float:
> -** ins v[0-9]+\.s[^\n]+
> -** ins v[0-9]+\.s[^\n]+
> +** uzp1 v[0-9]+\.2s[^\n]+
> +** uzp1 v[0-9]+\.2s[^\n]+
> ** zip1 v([0-9]+).4s, [^\n]+
> ** stp q\1, q\1, \[x0\]
> ** stp q\1, q\1, \[x0, #?32\]
> diff --git a/gcc/testsuite/gcc.target/aarch64/pr109072_1.c
> b/gcc/testsuite/gcc.target/aarch64/pr109072_1.c
> index
> 6c1d2b0bdccfb74b80d938a0d94413f0f9dda5ab..0fc195a598f3b82ff188b3151e77e1272254b78c
> 100644
> --- a/gcc/testsuite/gcc.target/aarch64/pr109072_1.c
> +++ b/gcc/testsuite/gcc.target/aarch64/pr109072_1.c
> @@ -54,7 +54,7 @@ f32x2_1 (float32_t x)
>
> /*
> ** f32x2_2:
> -** ins v0\.s\[1\], v1.s\[0\]
> +** uzp1 v0\.2s, v0\.2s, v1\.2s
> ** ret
> */
> float32x2_t
> @@ -165,7 +165,7 @@ f64x2_1 (float64_t x)
>
> /*
> ** f64x2_2:
> -** ins v0\.d\[1\], v1.d\[0\]
> +** uzp1 v0\.2d, v0\.2d, v1\.2d
> ** ret
> */
> float64x2_t
> diff --git a/gcc/testsuite/gcc.target/aarch64/vec-init-14.c
> b/gcc/testsuite/gcc.target/aarch64/vec-init-14.c
> index
> 02875088cd98833882cdf15b14dcb426951e428f..1a2cc9fbf473ad0de2d8ef97d7efdbe40d959866
> 100644
> --- a/gcc/testsuite/gcc.target/aarch64/vec-init-14.c
> +++ b/gcc/testsuite/gcc.target/aarch64/vec-init-14.c
> @@ -67,7 +67,7 @@ int32x2_t s32_6(int32_t a0, int32_t a1) {
>
> /*
> ** f32_1:
> -** ins v0\.s\[1\], v1\.s\[0\]
> +** uzp1 v0\.2s, v0\.2s, v1\.2s
> ** ret
> */
> float32x2_t f32_1(float32_t a0, float32_t a1) {
> @@ -90,7 +90,7 @@ float32x2_t f32_2(float32_t a0, float32_t *ptr) {
> /*
> ** f32_3:
> ** ldr s0, \[x0\]
> -** ins v0\.s\[1\], v1\.s\[0\]
> +** uzp1 v0\.2s, v0\.2s, v1\.2s
> ** ret
> */
> float32x2_t f32_3(float32_t a0, float32_t a1, float32_t *ptr) {
> diff --git a/gcc/testsuite/gcc.target/aarch64/vec-init-9.c
> b/gcc/testsuite/gcc.target/aarch64/vec-init-9.c
> index
> 8f68e06a55925b973a87723c7b5924264382e4b0..3cf05cf865e21fad482e5ffc8c769d0f15a57e74
> 100644
> --- a/gcc/testsuite/gcc.target/aarch64/vec-init-9.c
> +++ b/gcc/testsuite/gcc.target/aarch64/vec-init-9.c
> @@ -75,7 +75,7 @@ int64x2_t s64q_6(int64_t a0, int64_t a1) {
>
> /*
> ** f64q_1:
> -** ins v0\.d\[1\], v1\.d\[0\]
> +** uzp1 v0\.2d, v0\.2d, v1\.2d
> ** ret
> */
> float64x2_t f64q_1(float64_t a0, float64_t a1) {
> @@ -98,7 +98,7 @@ float64x2_t f64q_2(float64_t a0, float64_t *ptr) {
> /*
> ** f64q_3:
> ** ldr d0, \[x0\]
> -** ins v0\.d\[1\], v1\.d\[0\]
> +** uzp1 v0\.2d, v0\.2d, v1\.2d
> ** ret
> */
> float64x2_t f64q_3(float64_t a0, float64_t a1, float64_t *ptr) {
> @@ -140,7 +140,7 @@ float64x2_t f64q_6(float64_t a0, float64_t a1) {
>
> /*
> ** s32q_1:
> -** ins v0\.d\[1\], v1\.d\[0\]
> +** uzp1 v0\.2d, v0\.2d, v1\.2d
> ** ret
> */
> int32x4_t s32q_1(int32x2_t a0, int32x2_t a1) {
> @@ -157,7 +157,7 @@ int32x4_t s32q_2(int32x2_t a0, int32x2_t *ptr) {
> /*
> ** s32q_3:
> ** ldr d0, \[x0\]
> -** ins v0\.d\[1\], v1\.d\[0\]
> +** uzp1 v0\.2d, v0\.2d, v1\.2d
> ** ret
> */
> int32x4_t s32q_3(int32x2_t a0, int32x2_t a1, int32x2_t *ptr) {
> @@ -204,7 +204,7 @@ int32x4_t s32q_6(int32x2_t a0, int32x2_t a1) {
>
> /*
> ** f32q_1:
> -** ins v0\.d\[1\], v1\.d\[0\]
> +** uzp1 v0\.2d, v0\.2d, v1\.2d
> ** ret
> */
> float32x4_t f32q_1(float32x2_t a0, float32x2_t a1) {
> @@ -221,7 +221,7 @@ float32x4_t f32q_2(float32x2_t a0, float32x2_t *ptr) {
> /*
> ** f32q_3:
> ** ldr d0, \[x0\]
> -** ins v0\.d\[1\], v1\.d\[0\]
> +** uzp1 v0\.2d, v0\.2d, v1\.2d
> ** ret
> */
> float32x4_t f32q_3(float32x2_t a0, float32x2_t a1, float32x2_t *ptr) {
