From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from smtpbgbr2.qq.com (smtpbgbr2.qq.com [54.207.22.56]) by sourceware.org (Postfix) with ESMTPS id 1E2CF3858C5E for ; Fri, 12 May 2023 01:31:39 +0000 (GMT) DMARC-Filter: OpenDMARC Filter v1.4.2 sourceware.org 1E2CF3858C5E Authentication-Results: sourceware.org; dmarc=none (p=none dis=none) header.from=rivai.ai Authentication-Results: sourceware.org; spf=pass smtp.mailfrom=rivai.ai X-QQ-mid: bizesmtp69t1683855074tom427em Received: from server1.localdomain ( [58.60.1.22]) by bizesmtp.qq.com (ESMTP) with id ; Fri, 12 May 2023 09:31:13 +0800 (CST) X-QQ-SSF: 01400000000000F0Q000000A0000000 X-QQ-FEAT: QityeSR92A1brKU2IHc0K1HR7SPvRTyj+e2+HYQRHs8jmAkbCCNR/viT6olwU XvsYPk8XGh2EKuYzgOS1/C8oUET0/PtMsK+HbD+xe9n3GSK7DfzFee1UM99aF9GLep97YNf SEQaQh7vmZN6Me1MyU0dURIHtxoJPRZb8UEmNip0tOltQw9D2v3rpq7/o9qJlpyfidV73js XoOm18VahFBh/Fb1cLNjrcEvsDstAnHJLH2MDOehHpTDxKn9VQo8ki7OWUTjlHnVJ6C1T9k d1YqbyXUg6ntbTse45sJIQFFsAMmHG5xK66CS42z0IDzLLHmp58wCrWYDPKU4802K6LnW6f 001M26FfkCza6tLsxmCCwVM3T9JV9REIHYhIiJce4WOTqqnsMFwDPgwXT4eWTBnumtQ1Jjf X-QQ-GoodBg: 2 X-BIZMAIL-ID: 5962683737586881809 From: juzhe.zhong@rivai.ai To: gcc-patches@gcc.gnu.org Cc: kito.cheng@gmail.com, palmer@dabbelt.com, jeffreyalaw@gmail.com, rdapp.gcc@gmail.com, Juzhe-Zhong Subject: [PATCH V2] RISC-V: Add basic vec_init for VLS RVV auto-vectorization Date: Fri, 12 May 2023 09:31:12 +0800 Message-Id: <20230512013112.276462-1-juzhe.zhong@rivai.ai> X-Mailer: git-send-email 2.36.1 MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-QQ-SENDSIZE: 520 Feedback-ID: bizesmtp:rivai.ai:qybglogicsvrgz:qybglogicsvrgz7a-one-0 X-Spam-Status: No, score=-8.2 required=5.0 tests=BAYES_00,GIT_PATCH_0,KAM_ASCII_DIVIDERS,KAM_DMARC_STATUS,KAM_SHORT,RCVD_IN_BARRACUDACENTRAL,RCVD_IN_DNSWL_NONE,RCVD_IN_MSPIKE_H2,SCC_10_SHORT_WORD_LINES,SCC_20_SHORT_WORD_LINES,SCC_35_SHORT_WORD_LINES,SCC_5_SHORT_WORD_LINES,SPF_HELO_PASS,SPF_PASS,TXREP,T_SCC_BODY_TEXT_LINE autolearn=ham autolearn_force=no version=3.4.6 X-Spam-Checker-Version: SpamAssassin 3.4.6 (2021-04-09) on server2.sourceware.org List-Id: From: Juzhe-Zhong Rebase to trunk. This is patching is adding basic vec_init support for RVV auto-vectorization. This patch has been full coverage tested. This patch makes vec_init support common init vector handling (using vslide1down to insert element) which can handle any cases of initialization vec but it's not optimal for cases. And support Case 1 optimizaiton: https://godbolt.org/z/Yb9PK9jsz #include typedef int8_t vnx16qi __attribute__((vector_size (16))); #include typedef int8_t vnx16qi __attribute__ ((vector_size (16))); typedef int8_t vnx32qi __attribute__ ((vector_size (32))); typedef int8_t vnx64qi __attribute__ ((vector_size (64))); typedef int8_t vnx128qi __attribute__ ((vector_size (128))); __attribute__ ((noipa)) void f_vnx128qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t e, int8_t f, int8_t g, int8_t h, int8_t *out) { vnx128qi v = {a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h}; *(vnx128qi *) out = v; } LLVM codegen: https://godbolt.org/z/xsnavvWqx ... vslide1down.vx (x128 times) ... This patch codegen: f_vnx128qi: andi a1,a1,0xff andi a0,a0,0xff slli a1,a1,8 andi a2,a2,0xff or a1,a1,a0 slli a2,a2,16 andi a3,a3,0xff or a2,a2,a1 slli a3,a3,24 andi a4,a4,0xff or a3,a3,a2 slli a4,a4,32 andi a5,a5,0xff or a4,a4,a3 slli a5,a5,40 andi a6,a6,0xff or a5,a5,a4 slli a6,a6,48 or a6,a6,a5 vsetvli a5,zero,e64,m8,ta,ma ld a5,0(sp) slli a7,a7,56 or a7,a7,a6 vmv.v.x v8,a7 vs8r.v v8,0(a5) ret We support more optimizations cases in the future. But they are not included in this patch. --- gcc/config/riscv/autovec.md | 16 ++ gcc/config/riscv/riscv-protos.h | 1 + gcc/config/riscv/riscv-v.cc | 127 +++++++++++ gcc/config/riscv/vector-iterators.md | 9 + .../gcc.target/riscv/rvv/autovec/insert-1.c | 41 ++++ .../gcc.target/riscv/rvv/autovec/insert-2.c | 41 ++++ .../gcc.target/riscv/rvv/autovec/insert-3.c | 41 ++++ .../riscv/rvv/autovec/insert_run-1.c | 46 ++++ .../riscv/rvv/autovec/insert_run-2.c | 46 ++++ .../gcc.target/riscv/rvv/autovec/repeat-1.c | 75 +++++++ .../gcc.target/riscv/rvv/autovec/repeat-2.c | 61 ++++++ .../gcc.target/riscv/rvv/autovec/repeat-3.c | 53 +++++ .../gcc.target/riscv/rvv/autovec/repeat-4.c | 39 ++++ .../gcc.target/riscv/rvv/autovec/repeat-5.c | 74 +++++++ .../gcc.target/riscv/rvv/autovec/repeat-6.c | 78 +++++++ .../riscv/rvv/autovec/repeat_run-1.c | 125 +++++++++++ .../riscv/rvv/autovec/repeat_run-2.c | 145 +++++++++++++ .../riscv/rvv/autovec/repeat_run-3.c | 203 ++++++++++++++++++ .../riscv/rvv/autovec/repeat_run-4.c | 77 +++++++ .../riscv/rvv/autovec/repeat_run-5.c | 124 +++++++++++ .../riscv/rvv/autovec/repeat_run-6.c | 122 +++++++++++ 21 files changed, 1544 insertions(+) create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-1.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-2.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-3.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-1.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-2.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-1.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-2.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-3.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-4.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-5.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-6.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-1.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-2.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-3.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-4.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-5.c create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-6.c diff --git a/gcc/config/riscv/autovec.md b/gcc/config/riscv/autovec.md index ac0c939d277..ce0b46537ad 100644 --- a/gcc/config/riscv/autovec.md +++ b/gcc/config/riscv/autovec.md @@ -83,6 +83,22 @@ } ) +;; ------------------------------------------------------------------------- +;; ---- [INT,FP] Initialize from individual elements +;; ------------------------------------------------------------------------- +;; This is the pattern initialize the vector +;; ------------------------------------------------------------------------- + +(define_expand "vec_init" + [(match_operand:V 0 "register_operand") + (match_operand 1 "")] + "TARGET_VECTOR" + { + riscv_vector::expand_vec_init (operands[0], operands[1]); + DONE; + } +) + ;; ======================================================================== ;; == Vector operations ;; ========================================================================= diff --git a/gcc/config/riscv/riscv-protos.h b/gcc/config/riscv/riscv-protos.h index 55a43d6270e..bc71f9cbbba 100644 --- a/gcc/config/riscv/riscv-protos.h +++ b/gcc/config/riscv/riscv-protos.h @@ -222,6 +222,7 @@ void expand_tuple_move (machine_mode, rtx *); machine_mode preferred_simd_mode (scalar_mode); opt_machine_mode get_mask_mode (machine_mode); void expand_vec_series (rtx, rtx, rtx); +void expand_vec_init (rtx, rtx); } /* We classify builtin types into two classes: diff --git a/gcc/config/riscv/riscv-v.cc b/gcc/config/riscv/riscv-v.cc index 7cf5ec9f3ba..b8dc333f54e 100644 --- a/gcc/config/riscv/riscv-v.cc +++ b/gcc/config/riscv/riscv-v.cc @@ -1131,4 +1131,131 @@ preferred_simd_mode (scalar_mode mode) return word_mode; } +class rvv_builder : public rtx_vector_builder +{ +public: + rvv_builder () : rtx_vector_builder () {} + rvv_builder (machine_mode mode, unsigned int npatterns, + unsigned int nelts_per_pattern) + : rtx_vector_builder (mode, npatterns, nelts_per_pattern) + { + m_inner_mode = GET_MODE_INNER (mode); + m_inner_size = GET_MODE_BITSIZE (m_inner_mode).to_constant (); + } + + bool can_duplicate_repeating_sequence_p (); + rtx get_merged_repeating_sequence (); + + machine_mode new_mode () const { return m_new_mode; } + +private: + machine_mode m_inner_mode; + machine_mode m_new_mode; + scalar_int_mode m_new_inner_mode; + unsigned int m_inner_size; +}; + +/* Return true if the vector duplicated by a super element which is the fusion + of consecutive elements. + + v = { a, b, a, b } super element = ab, v = { ab, ab } */ +bool +rvv_builder::can_duplicate_repeating_sequence_p () +{ + poly_uint64 new_size = exact_div (full_nelts (), npatterns ()); + unsigned int new_inner_size = m_inner_size * npatterns (); + if (!int_mode_for_size (new_inner_size, 0).exists (&m_new_inner_mode) + || GET_MODE_SIZE (m_new_inner_mode) > UNITS_PER_WORD + || !get_vector_mode (m_new_inner_mode, new_size).exists (&m_new_mode)) + return false; + return repeating_sequence_p (0, full_nelts ().to_constant (), npatterns ()); +} + +/* Merge the repeating sequence into a single element and return the RTX. */ +rtx +rvv_builder::get_merged_repeating_sequence () +{ + scalar_int_mode mode = Pmode; + rtx target = gen_reg_rtx (mode); + emit_move_insn (target, const0_rtx); + rtx imm = gen_int_mode ((1ULL << m_inner_size) - 1, mode); + /* { a, b, a, b }: Generate duplicate element = b << bits | a. */ + for (unsigned int i = 0; i < npatterns (); i++) + { + unsigned int loc = m_inner_size * i; + rtx shift = gen_int_mode (loc, mode); + rtx ele = gen_lowpart (mode, elt (i)); + rtx tmp = expand_simple_binop (mode, AND, ele, imm, NULL_RTX, false, + OPTAB_DIRECT); + rtx tmp2 = expand_simple_binop (mode, ASHIFT, tmp, shift, NULL_RTX, false, + OPTAB_DIRECT); + rtx tmp3 = expand_simple_binop (mode, IOR, tmp2, target, NULL_RTX, false, + OPTAB_DIRECT); + emit_move_insn (target, tmp3); + } + if (GET_MODE_SIZE (m_new_inner_mode) < UNITS_PER_WORD) + return gen_lowpart (m_new_inner_mode, target); + return target; +} + +/* Subroutine of riscv_vector_expand_vector_init. + Works as follows: + (a) Initialize TARGET by broadcasting element NELTS_REQD - 1 of BUILDER. + (b) Skip leading elements from BUILDER, which are the same as + element NELTS_REQD - 1. + (c) Insert earlier elements in reverse order in TARGET using vslide1down. */ + +static void +expand_vector_init_insert_elems (rtx target, const rvv_builder &builder, + int nelts_reqd) +{ + machine_mode mode = GET_MODE (target); + scalar_mode elem_mode = GET_MODE_INNER (mode); + machine_mode mask_mode; + gcc_assert (get_mask_mode (mode).exists (&mask_mode)); + rtx dup = expand_vector_broadcast (mode, builder.elt (0)); + emit_move_insn (target, dup); + int ndups = builder.count_dups (0, nelts_reqd - 1, 1); + for (int i = ndups; i < nelts_reqd; i++) + { + unsigned int unspec + = FLOAT_MODE_P (mode) ? UNSPEC_VFSLIDE1DOWN : UNSPEC_VSLIDE1DOWN; + insn_code icode = code_for_pred_slide (unspec, mode); + emit_len_binop (icode, target, target, builder.elt (i), NULL, mask_mode, + elem_mode); + } +} + +/* Initialize register TARGET from the elements in PARALLEL rtx VALS. */ + +void +expand_vec_init (rtx target, rtx vals) +{ + machine_mode mode = GET_MODE (target); + int nelts = XVECLEN (vals, 0); + + rvv_builder v (mode, nelts, 1); + for (int i = 0; i < nelts; i++) + v.quick_push (XVECEXP (vals, 0, i)); + v.finalize (); + + if (nelts > 3) + { + /* Case 1: Convert v = { a, b, a, b } into v = { ab, ab }. */ + if (v.can_duplicate_repeating_sequence_p ()) + { + rtx ele = v.get_merged_repeating_sequence (); + rtx dup = expand_vector_broadcast (v.new_mode (), ele); + emit_move_insn (target, gen_lowpart (mode, dup)); + return; + } + /* TODO: We will support more Initialization of vector in the future. */ + } + + /* Handle common situation by vslide1down. This function can handle any + situation of vec_init. Only the cases that are not optimized above + will fall through here. */ + expand_vector_init_insert_elems (target, v, nelts); +} + } // namespace riscv_vector diff --git a/gcc/config/riscv/vector-iterators.md b/gcc/config/riscv/vector-iterators.md index 5cf958ba845..a282861335d 100644 --- a/gcc/config/riscv/vector-iterators.md +++ b/gcc/config/riscv/vector-iterators.md @@ -996,6 +996,15 @@ (VNx1DF "DF") (VNx2DF "DF") (VNx4DF "DF") (VNx8DF "DF") (VNx16DF "DF") ]) +(define_mode_attr vel [ + (VNx1QI "qi") (VNx2QI "qi") (VNx4QI "qi") (VNx8QI "qi") (VNx16QI "qi") (VNx32QI "qi") (VNx64QI "qi") (VNx128QI "qi") + (VNx1HI "hi") (VNx2HI "hi") (VNx4HI "hi") (VNx8HI "hi") (VNx16HI "hi") (VNx32HI "hi") (VNx64HI "hi") + (VNx1SI "si") (VNx2SI "si") (VNx4SI "si") (VNx8SI "si") (VNx16SI "si") (VNx32SI "si") + (VNx1DI "di") (VNx2DI "di") (VNx4DI "di") (VNx8DI "di") (VNx16DI "di") + (VNx1SF "sf") (VNx2SF "sf") (VNx4SF "sf") (VNx8SF "sf") (VNx16SF "sf") (VNx32SF "sf") + (VNx1DF "df") (VNx2DF "df") (VNx4DF "df") (VNx8DF "df") (VNx16DF "df") +]) + (define_mode_attr VSUBEL [ (VNx1HI "QI") (VNx2HI "QI") (VNx4HI "QI") (VNx8HI "QI") (VNx16HI "QI") (VNx32HI "QI") (VNx64HI "QI") (VNx1SI "HI") (VNx2SI "HI") (VNx4SI "HI") (VNx8SI "HI") (VNx16SI "HI") (VNx32SI "HI") diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-1.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-1.c new file mode 100644 index 00000000000..0154e7c87d4 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-1.c @@ -0,0 +1,41 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv64gcv -mabi=lp64d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef int64_t vnx2di __attribute__ ((vector_size (16))); +typedef int64_t vnx4di __attribute__ ((vector_size (32))); +typedef int64_t vnx8di __attribute__ ((vector_size (64))); +typedef int64_t vnx16di __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx2di (int64_t a, int64_t b, int64_t *out) +{ + vnx2di v = {a, b}; + *(vnx2di *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx4di (int64_t a, int64_t b, int64_t c, int64_t d, int64_t *out) +{ + vnx4di v = {a, b, c, d}; + *(vnx4di *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx8di (int64_t a, int64_t b, int64_t c, int64_t d, int64_t e, int64_t f, int64_t g, int64_t h, int64_t *out) +{ + vnx8di v = {a, b, c, d, e, f, g, h}; + *(vnx8di *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16di (int64_t a, int64_t b, int64_t c, int64_t d, int64_t e, int64_t f, + int64_t g, int64_t h, int64_t a2, int64_t b2, int64_t c2, int64_t d2, + int64_t e2, int64_t f2, int64_t g2, int64_t h2, int64_t *out) +{ + vnx16di v = {a, b, c, d, e, f, g, h, a2, b2, c2, d2, e2, f2, g2, h2}; + *(vnx16di *) out = v; +} + +/* { dg-final { scan-assembler-times {vslide1down\.vx\tv[0-9]+,\s*v[0-9]+,\s*[a-x0-9]+} 26 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-2.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-2.c new file mode 100644 index 00000000000..b42c216e421 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-2.c @@ -0,0 +1,41 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv32gcv -mabi=ilp32d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef int64_t vnx2di __attribute__ ((vector_size (16))); +typedef int64_t vnx4di __attribute__ ((vector_size (32))); +typedef int64_t vnx8di __attribute__ ((vector_size (64))); +typedef int64_t vnx16di __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx2di (int64_t a, int64_t b, int64_t *out) +{ + vnx2di v = {a, b}; + *(vnx2di *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx4di (int64_t a, int64_t b, int64_t c, int64_t d, int64_t *out) +{ + vnx4di v = {a, b, c, d}; + *(vnx4di *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx8di (int64_t a, int64_t b, int64_t c, int64_t d, int64_t e, int64_t f, int64_t g, int64_t h, int64_t *out) +{ + vnx8di v = {a, b, c, d, e, f, g, h}; + *(vnx8di *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16di (int64_t a, int64_t b, int64_t c, int64_t d, int64_t e, int64_t f, + int64_t g, int64_t h, int64_t a2, int64_t b2, int64_t c2, int64_t d2, + int64_t e2, int64_t f2, int64_t g2, int64_t h2, int64_t *out) +{ + vnx16di v = {a, b, c, d, e, f, g, h, a2, b2, c2, d2, e2, f2, g2, h2}; + *(vnx16di *) out = v; +} + +/* { dg-final { scan-assembler-times {vslide1down\.vx\tv[0-9]+,\s*v[0-9]+,\s*[a-x0-9]+} 52 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-3.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-3.c new file mode 100644 index 00000000000..be0aaea8203 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert-3.c @@ -0,0 +1,41 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv64gcv -mabi=lp64d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef double vnx2df __attribute__ ((vector_size (16))); +typedef double vnx4df __attribute__ ((vector_size (32))); +typedef double vnx8df __attribute__ ((vector_size (64))); +typedef double vnx16df __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx2df (double a, double b, double *out) +{ + vnx2df v = {a, b}; + *(vnx2df *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx4df (double a, double b, double c, double d, double *out) +{ + vnx4df v = {a, b, c, d}; + *(vnx4df *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx8df (double a, double b, double c, double d, double e, double f, double g, double h, double *out) +{ + vnx8df v = {a, b, c, d, e, f, g, h}; + *(vnx8df *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16df (double a, double b, double c, double d, double e, double f, + double g, double h, double a2, double b2, double c2, double d2, + double e2, double f2, double g2, double h2, double *out) +{ + vnx16df v = {a, b, c, d, e, f, g, h, a2, b2, c2, d2, e2, f2, g2, h2}; + *(vnx16df *) out = v; +} + +/* { dg-final { scan-assembler-times {vfslide1down\.vf\tv[0-9]+,\s*v[0-9]+,\s*[a-x0-9]+} 26 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-1.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-1.c new file mode 100644 index 00000000000..450151445eb --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-1.c @@ -0,0 +1,46 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "insert-1.c" + +int +main () +{ + int64_t in[16] = {-458615, -1551, -552, -458615, -45815, -551, -2, -4585, + -134, -99294, -789, -51284, -7324, -3333, -6666, -11}; + + int64_t v_vnx2di[sizeof (vnx2di) / sizeof (int64_t)]; + f_vnx2di (in[0], in[1], v_vnx2di); + for (int i = 0; i < sizeof (vnx2di) / sizeof (int64_t); i++) + { + if (v_vnx2di[i] != in[i]) + __builtin_abort (); + } + + int64_t v_vnx4di[sizeof (vnx4di) / sizeof (int64_t)]; + f_vnx4di (in[0], in[1], in[2], in[3], v_vnx4di); + for (int i = 0; i < sizeof (vnx2di) / sizeof (int64_t); i++) + { + if (v_vnx2di[i] != in[i]) + __builtin_abort (); + } + + int64_t v_vnx8di[sizeof (vnx8di) / sizeof (int64_t)]; + f_vnx8di (in[0], in[1], in[2], in[3], in[4], in[5], in[6], in[7], v_vnx8di); + for (int i = 0; i < sizeof (vnx2di) / sizeof (int64_t); i++) + { + if (v_vnx2di[i] != in[i]) + __builtin_abort (); + } + + int64_t v_vnx16di[sizeof (vnx16di) / sizeof (int64_t)]; + f_vnx16di (in[0], in[1], in[2], in[3], in[4], in[5], in[6], in[7], in[8], + in[9], in[10], in[11], in[12], in[13], in[14], in[15], v_vnx16di); + for (int i = 0; i < sizeof (vnx2di) / sizeof (int64_t); i++) + { + if (v_vnx2di[i] != in[i]) + __builtin_abort (); + } + + return 0; +} diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-2.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-2.c new file mode 100644 index 00000000000..b0e2d6c257f --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/insert_run-2.c @@ -0,0 +1,46 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "insert-3.c" + +int +main () +{ + double in[16] = {-458.615, -1.551, -55.2, -4586.15, -4581.5, -55.1, -2, -4.585, + -13.4, -9929.4, -7.89, -512.84, -73.24, -33.33, -666.6, -1.1}; + + double v_vnx2df[sizeof (vnx2df) / sizeof (double)]; + f_vnx2df (in[0], in[1], v_vnx2df); + for (int i = 0; i < sizeof (vnx2df) / sizeof (double); i++) + { + if (v_vnx2df[i] != in[i]) + __builtin_abort (); + } + + double v_vnx4df[sizeof (vnx4df) / sizeof (double)]; + f_vnx4df (in[0], in[1], in[2], in[3], v_vnx4df); + for (int i = 0; i < sizeof (vnx2df) / sizeof (double); i++) + { + if (v_vnx2df[i] != in[i]) + __builtin_abort (); + } + + double v_vnx8df[sizeof (vnx8df) / sizeof (double)]; + f_vnx8df (in[0], in[1], in[2], in[3], in[4], in[5], in[6], in[7], v_vnx8df); + for (int i = 0; i < sizeof (vnx2df) / sizeof (double); i++) + { + if (v_vnx2df[i] != in[i]) + __builtin_abort (); + } + + double v_vnx16df[sizeof (vnx16df) / sizeof (double)]; + f_vnx16df (in[0], in[1], in[2], in[3], in[4], in[5], in[6], in[7], in[8], + in[9], in[10], in[11], in[12], in[13], in[14], in[15], v_vnx16df); + for (int i = 0; i < sizeof (vnx2df) / sizeof (double); i++) + { + if (v_vnx2df[i] != in[i]) + __builtin_abort (); + } + + return 0; +} diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-1.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-1.c new file mode 100644 index 00000000000..b0153a82418 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-1.c @@ -0,0 +1,75 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv32gcv -mabi=ilp32d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef int8_t vnx2qi __attribute__ ((vector_size (2))); +typedef int8_t vnx4qi __attribute__ ((vector_size (4))); +typedef int8_t vnx8qi __attribute__ ((vector_size (8))); +typedef int8_t vnx16qi __attribute__ ((vector_size (16))); +typedef int8_t vnx32qi __attribute__ ((vector_size (32))); +typedef int8_t vnx64qi __attribute__ ((vector_size (64))); +typedef int8_t vnx128qi __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx2qi (int8_t a, int8_t b, int8_t *out) +{ + vnx2qi v = {a, b}; + *(vnx2qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx4qi (int8_t a, int8_t b, int8_t *out) +{ + vnx4qi v = {a, b, a, b}; + *(vnx4qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx8qi (int8_t a, int8_t b, int8_t *out) +{ + vnx8qi v = {a, b, a, b, a, b, a, b}; + *(vnx8qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16qi (int8_t a, int8_t b, int8_t *out) +{ + vnx16qi v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx16qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx32qi (int8_t a, int8_t b, int8_t *out) +{ + vnx32qi v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx32qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx64qi (int8_t a, int8_t b, int8_t *out) +{ + vnx64qi v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx64qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx128qi (int8_t a, int8_t b, int8_t *out) +{ + vnx128qi v + = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx128qi *) out = v; +} + +/* { dg-final { scan-assembler-times {vmv.v.x\tv[0-9]+,\s*[a-x0-9]+} 7 } } */ +/* { dg-final { scan-assembler-times {slli\t[a-x0-9]+,\s*[a-x0-9]+,\s*8} 6 } } */ +/* { dg-final { scan-assembler-times {or\t[a-x0-9]+,\s*[a-x0-9]+,\s*[a-x0-9]+} 6 } } */ +/* { dg-final { scan-assembler-times {vslide1down\.vx\tv[0-9]+,\s*v[0-9]+,\s*[a-x0-9]+} 1 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-2.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-2.c new file mode 100644 index 00000000000..ac5599510a3 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-2.c @@ -0,0 +1,61 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv32gcv -mabi=ilp32d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef int8_t vnx8qi __attribute__ ((vector_size (8))); +typedef int8_t vnx16qi __attribute__ ((vector_size (16))); +typedef int8_t vnx32qi __attribute__ ((vector_size (32))); +typedef int8_t vnx64qi __attribute__ ((vector_size (64))); +typedef int8_t vnx128qi __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx8qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t *out) +{ + vnx8qi v = {a, b, c, d, a, b, c, d}; + *(vnx8qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t *out) +{ + vnx16qi v = {a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d}; + *(vnx16qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx32qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t *out) +{ + vnx32qi v = {a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d}; + *(vnx32qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx64qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t *out) +{ + vnx64qi v = {a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d}; + *(vnx64qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx128qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t *out) +{ + vnx128qi v + = {a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d, + a, b, c, d, a, b, c, d, a, b, c, d, a, b, c, d}; + *(vnx128qi *) out = v; +} + +/* { dg-final { scan-assembler-times {vmv.v.x\tv[0-9]+,\s*[a-x0-9]+} 5 } } */ +/* { dg-final { scan-assembler-times {slli} 15 } } */ +/* { dg-final { scan-assembler-times {or} 15 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-3.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-3.c new file mode 100644 index 00000000000..ee5cf8eed09 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-3.c @@ -0,0 +1,53 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv64gcv -mabi=lp64d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef int8_t vnx16qi __attribute__ ((vector_size (16))); +typedef int8_t vnx32qi __attribute__ ((vector_size (32))); +typedef int8_t vnx64qi __attribute__ ((vector_size (64))); +typedef int8_t vnx128qi __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx16qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t e, int8_t f, int8_t g, int8_t h, int8_t *out) +{ + vnx16qi v = {a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h}; + *(vnx16qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx32qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t e, int8_t f, int8_t g, int8_t h, int8_t *out) +{ + vnx32qi v = {a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h}; + *(vnx32qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx64qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t e, int8_t f, int8_t g, int8_t h, int8_t *out) +{ + vnx64qi v = {a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h}; + *(vnx64qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx128qi (int8_t a, int8_t b, int8_t c, int8_t d, int8_t e, int8_t f, int8_t g, int8_t h, int8_t *out) +{ + vnx128qi v + = {a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h, + a, b, c, d, e, f, g, h, a, b, c, d, e, f, g, h}; + *(vnx128qi *) out = v; +} + +/* { dg-final { scan-assembler-times {vmv\.v\.x\tv[0-9]+,\s*[a-x0-9]+} 4 } } */ +/* { dg-final { scan-assembler-times {slli} 28 } } */ +/* { dg-final { scan-assembler-times {or} 28 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-4.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-4.c new file mode 100644 index 00000000000..52d84449bdf --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-4.c @@ -0,0 +1,39 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv64gcv -mabi=lp64d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef float vnx4sf __attribute__ ((vector_size (16))); +typedef float vnx8sf __attribute__ ((vector_size (32))); +typedef float vnx16sf __attribute__ ((vector_size (64))); +typedef float vnx32sf __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx4sf (float a, float b, float *out) +{ + vnx4sf v = {a, b, a, b}; + *(vnx4sf *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx8sf (float a, float b, float *out) +{ + vnx8sf v = {a, b, a, b, a, b, a, b}; + *(vnx8sf *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16sf (float a, float b, float *out) +{ + vnx16sf v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx16sf *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx32sf (float a, float b, float *out) +{ + vnx32sf v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx32sf *) out = v; +} + +/* { dg-final { scan-assembler-times {vmv\.v\.x\tv[0-9]+,\s*[a-x0-9]+} 4 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-5.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-5.c new file mode 100644 index 00000000000..b3443f700be --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-5.c @@ -0,0 +1,74 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv32gcv -mabi=ilp32d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef int8_t vnx2qi __attribute__ ((vector_size (2))); +typedef int8_t vnx4qi __attribute__ ((vector_size (4))); +typedef int8_t vnx8qi __attribute__ ((vector_size (8))); +typedef int8_t vnx16qi __attribute__ ((vector_size (16))); +typedef int8_t vnx32qi __attribute__ ((vector_size (32))); +typedef int8_t vnx64qi __attribute__ ((vector_size (64))); +typedef int8_t vnx128qi __attribute__ ((vector_size (128))); + +__attribute__ ((noipa)) void +f_vnx2qi (int8_t a, int8_t *out) +{ + vnx2qi v = {a, -88}; + *(vnx2qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx4qi (int8_t a, int8_t *out) +{ + vnx4qi v = {a, -88, a, -88}; + *(vnx4qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx8qi (int8_t a, int8_t *out) +{ + vnx8qi v = {a, -88, a, -88, a, -88, a, -88}; + *(vnx8qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16qi (int8_t a, int8_t *out) +{ + vnx16qi v = {a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88}; + *(vnx16qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx32qi (int8_t a, int8_t *out) +{ + vnx32qi v = {a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88}; + *(vnx32qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx64qi (int8_t a, int8_t *out) +{ + vnx64qi v = {a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88}; + *(vnx64qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx128qi (int8_t a, int8_t *out) +{ + vnx128qi v + = {a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, + a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88, a, -88}; + *(vnx128qi *) out = v; +} + +/* { dg-final { scan-assembler-times {vmv.v.x\tv[0-9]+,\s*[a-x0-9]+} 7 } } */ +/* { dg-final { scan-assembler-times {or\t[a-x0-9]+,\s*[a-x0-9]+,\s*[a-x0-9]+} 6 } } */ +/* { dg-final { scan-assembler-times {vslide1down\.vx\tv[0-9]+,\s*v[0-9]+,\s*[a-x0-9]+} 1 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-6.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-6.c new file mode 100644 index 00000000000..f1788a8759f --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat-6.c @@ -0,0 +1,78 @@ +/* { dg-do compile } */ +/* { dg-options "-march=rv32gcv -mabi=ilp32d --param riscv-autovec-preference=fixed-vlmax" } */ + +#include + +typedef int8_t vnx2qi __attribute__ ((vector_size (2))); +typedef int8_t vnx4qi __attribute__ ((vector_size (4))); +typedef int8_t vnx8qi __attribute__ ((vector_size (8))); +typedef int8_t vnx16qi __attribute__ ((vector_size (16))); +typedef int8_t vnx32qi __attribute__ ((vector_size (32))); +typedef int8_t vnx64qi __attribute__ ((vector_size (64))); +typedef int8_t vnx128qi __attribute__ ((vector_size (128))); + +int8_t a = -33; +int8_t b = -123; + +__attribute__ ((noipa)) void +f_vnx2qi (int8_t *out) +{ + vnx2qi v = {a, b}; + *(vnx2qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx4qi (int8_t *out) +{ + vnx4qi v = {a, b, a, b}; + *(vnx4qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx8qi (int8_t *out) +{ + vnx8qi v = {a, b, a, b, a, b, a, b}; + *(vnx8qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx16qi (int8_t *out) +{ + vnx16qi v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx16qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx32qi (int8_t *out) +{ + vnx32qi v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx32qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx64qi (int8_t *out) +{ + vnx64qi v = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx64qi *) out = v; +} + +__attribute__ ((noipa)) void +f_vnx128qi (int8_t *out) +{ + vnx128qi v + = {a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, + a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b, a, b}; + *(vnx128qi *) out = v; +} + +/* { dg-final { scan-assembler-times {vmv.v.x\tv[0-9]+,\s*[a-x0-9]+} 6 } } */ +/* { dg-final { scan-assembler-times {slli\t[a-x0-9]+,\s*[a-x0-9]+,\s*8} 6 } } */ +/* { dg-final { scan-assembler-times {or\t[a-x0-9]+,\s*[a-x0-9]+,\s*[a-x0-9]+} 6 } } */ +/* { dg-final { scan-assembler-times {vslide1down\.vx\tv[0-9]+,\s*v[0-9]+,\s*[a-x0-9]+} 1 } } */ diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-1.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-1.c new file mode 100644 index 00000000000..ee94bebf0b1 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-1.c @@ -0,0 +1,125 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "repeat-1.c" + +int +main () +{ + int8_t a = -17; + int8_t b = 89; + + int8_t v_vnx2qi[sizeof (vnx2qi) / sizeof (int8_t)]; + f_vnx2qi (a, b, v_vnx2qi); + for (int i = 0; i < sizeof (vnx2qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx2qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx2qi[i] != b) + __builtin_abort (); + } + } + + int8_t v_vnx4qi[sizeof (vnx4qi) / sizeof (int8_t)]; + f_vnx4qi (a, b, v_vnx4qi); + for (int i = 0; i < sizeof (vnx4qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx4qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx4qi[i] != b) + __builtin_abort (); + } + } + + int8_t v_vnx8qi[sizeof (vnx8qi) / sizeof (int8_t)]; + f_vnx8qi (a, b, v_vnx8qi); + for (int i = 0; i < sizeof (vnx8qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx8qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx8qi[i] != b) + __builtin_abort (); + } + } + + int8_t v_vnx16qi[sizeof (vnx16qi) / sizeof (int8_t)]; + f_vnx16qi (a, b, v_vnx16qi); + for (int i = 0; i < sizeof (vnx16qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx16qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx16qi[i] != b) + __builtin_abort (); + } + } + + int8_t v_vnx32qi[sizeof (vnx32qi) / sizeof (int8_t)]; + f_vnx32qi (a, b, v_vnx32qi); + for (int i = 0; i < sizeof (vnx32qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx32qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx32qi[i] != b) + __builtin_abort (); + } + } + + int8_t v_vnx64qi[sizeof (vnx64qi) / sizeof (int8_t)]; + f_vnx64qi (a, b, v_vnx64qi); + for (int i = 0; i < sizeof (vnx64qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx64qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx64qi[i] != b) + __builtin_abort (); + } + } + + int8_t v_vnx128qi[sizeof (vnx128qi) / sizeof (int8_t)]; + f_vnx128qi (a, b, v_vnx128qi); + for (int i = 0; i < sizeof (vnx128qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx128qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx128qi[i] != b) + __builtin_abort (); + } + } + + return 0; +} diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-2.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-2.c new file mode 100644 index 00000000000..5c6600a73d8 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-2.c @@ -0,0 +1,145 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "repeat-2.c" + +int +main () +{ + int8_t a = -17; + int8_t b = -120; + int8_t c = 111; + int8_t d = -11; + + int8_t v_vnx8qi[sizeof (vnx8qi) / sizeof (int8_t)]; + f_vnx8qi (a, b, c, d, v_vnx8qi); + for (int i = 0; i < sizeof (vnx8qi) / sizeof (int8_t); i++) + { + if (i % 4 == 0) + { + if (v_vnx8qi[i] != a) + __builtin_abort (); + } + else if (i % 4 == 1) + { + if (v_vnx8qi[i] != b) + __builtin_abort (); + } + else if (i % 4 == 2) + { + if (v_vnx8qi[i] != c) + __builtin_abort (); + } + else + { + if (v_vnx8qi[i] != d) + __builtin_abort (); + } + } + + int8_t v_vnx16qi[sizeof (vnx16qi) / sizeof (int8_t)]; + f_vnx16qi (a, b, c, d, v_vnx16qi); + for (int i = 0; i < sizeof (vnx16qi) / sizeof (int8_t); i++) + { + if (i % 4 == 0) + { + if (v_vnx16qi[i] != a) + __builtin_abort (); + } + else if (i % 4 == 1) + { + if (v_vnx16qi[i] != b) + __builtin_abort (); + } + else if (i % 4 == 2) + { + if (v_vnx16qi[i] != c) + __builtin_abort (); + } + else + { + if (v_vnx16qi[i] != d) + __builtin_abort (); + } + } + + int8_t v_vnx32qi[sizeof (vnx32qi) / sizeof (int8_t)]; + f_vnx32qi (a, b, c, d, v_vnx32qi); + for (int i = 0; i < sizeof (vnx32qi) / sizeof (int8_t); i++) + { + if (i % 4 == 0) + { + if (v_vnx32qi[i] != a) + __builtin_abort (); + } + else if (i % 4 == 1) + { + if (v_vnx32qi[i] != b) + __builtin_abort (); + } + else if (i % 4 == 2) + { + if (v_vnx32qi[i] != c) + __builtin_abort (); + } + else + { + if (v_vnx32qi[i] != d) + __builtin_abort (); + } + } + + int8_t v_vnx64qi[sizeof (vnx64qi) / sizeof (int8_t)]; + f_vnx64qi (a, b, c, d, v_vnx64qi); + for (int i = 0; i < sizeof (vnx64qi) / sizeof (int8_t); i++) + { + if (i % 4 == 0) + { + if (v_vnx64qi[i] != a) + __builtin_abort (); + } + else if (i % 4 == 1) + { + if (v_vnx64qi[i] != b) + __builtin_abort (); + } + else if (i % 4 == 2) + { + if (v_vnx64qi[i] != c) + __builtin_abort (); + } + else + { + if (v_vnx64qi[i] != d) + __builtin_abort (); + } + } + + int8_t v_vnx128qi[sizeof (vnx128qi) / sizeof (int8_t)]; + f_vnx128qi (a, b, c, d, v_vnx128qi); + for (int i = 0; i < sizeof (vnx128qi) / sizeof (int8_t); i++) + { + if (i % 4 == 0) + { + if (v_vnx128qi[i] != a) + __builtin_abort (); + } + else if (i % 4 == 1) + { + if (v_vnx128qi[i] != b) + __builtin_abort (); + } + else if (i % 4 == 2) + { + if (v_vnx128qi[i] != c) + __builtin_abort (); + } + else + { + if (v_vnx128qi[i] != d) + __builtin_abort (); + } + } + + return 0; +} diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-3.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-3.c new file mode 100644 index 00000000000..53927f6cc12 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-3.c @@ -0,0 +1,203 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "repeat-3.c" + +int +main () +{ + int8_t a = -17; + int8_t b = -120; + int8_t c = 111; + int8_t d = -11; + int8_t e = -34; + int8_t f = -101; + int8_t g = 33; + int8_t h = 9; + + int8_t v_vnx16qi[sizeof (vnx16qi) / sizeof (int8_t)]; + f_vnx16qi (a, b, c, d, e, f, g, h, v_vnx16qi); + for (int i = 0; i < sizeof (vnx16qi) / sizeof (int8_t); i++) + { + if (i % 8 == 0) + { + if (v_vnx16qi[i] != a) + __builtin_abort (); + } + else if (i % 8 == 1) + { + if (v_vnx16qi[i] != b) + __builtin_abort (); + } + else if (i % 8 == 2) + { + if (v_vnx16qi[i] != c) + __builtin_abort (); + } + else if (i % 8 == 3) + { + if (v_vnx16qi[i] != d) + __builtin_abort (); + } + else if (i % 8 == 4) + { + if (v_vnx16qi[i] != e) + __builtin_abort (); + } + else if (i % 8 == 5) + { + if (v_vnx16qi[i] != f) + __builtin_abort (); + } + else if (i % 8 == 6) + { + if (v_vnx16qi[i] != g) + __builtin_abort (); + } + else if (i % 8 == 7) + { + if (v_vnx16qi[i] != h) + __builtin_abort (); + } + } + + int8_t v_vnx32qi[sizeof (vnx32qi) / sizeof (int8_t)]; + f_vnx32qi (a, b, c, d, e, f, g, h, v_vnx32qi); + for (int i = 0; i < sizeof (vnx32qi) / sizeof (int8_t); i++) + { + if (i % 8 == 0) + { + if (v_vnx32qi[i] != a) + __builtin_abort (); + } + else if (i % 8 == 1) + { + if (v_vnx32qi[i] != b) + __builtin_abort (); + } + else if (i % 8 == 2) + { + if (v_vnx32qi[i] != c) + __builtin_abort (); + } + else if (i % 8 == 3) + { + if (v_vnx32qi[i] != d) + __builtin_abort (); + } + else if (i % 8 == 4) + { + if (v_vnx32qi[i] != e) + __builtin_abort (); + } + else if (i % 8 == 5) + { + if (v_vnx32qi[i] != f) + __builtin_abort (); + } + else if (i % 8 == 6) + { + if (v_vnx32qi[i] != g) + __builtin_abort (); + } + else if (i % 8 == 7) + { + if (v_vnx32qi[i] != h) + __builtin_abort (); + } + } + + int8_t v_vnx64qi[sizeof (vnx64qi) / sizeof (int8_t)]; + f_vnx64qi (a, b, c, d, e, f, g, h, v_vnx64qi); + for (int i = 0; i < sizeof (vnx64qi) / sizeof (int8_t); i++) + { + if (i % 8 == 0) + { + if (v_vnx64qi[i] != a) + __builtin_abort (); + } + else if (i % 8 == 1) + { + if (v_vnx64qi[i] != b) + __builtin_abort (); + } + else if (i % 8 == 2) + { + if (v_vnx64qi[i] != c) + __builtin_abort (); + } + else if (i % 8 == 3) + { + if (v_vnx64qi[i] != d) + __builtin_abort (); + } + else if (i % 8 == 4) + { + if (v_vnx64qi[i] != e) + __builtin_abort (); + } + else if (i % 8 == 5) + { + if (v_vnx64qi[i] != f) + __builtin_abort (); + } + else if (i % 8 == 6) + { + if (v_vnx64qi[i] != g) + __builtin_abort (); + } + else if (i % 8 == 7) + { + if (v_vnx64qi[i] != h) + __builtin_abort (); + } + } + + int8_t v_vnx128qi[sizeof (vnx128qi) / sizeof (int8_t)]; + f_vnx128qi (a, b, c, d, e, f, g, h, v_vnx128qi); + for (int i = 0; i < sizeof (vnx128qi) / sizeof (int8_t); i++) + { + if (i % 8 == 0) + { + if (v_vnx128qi[i] != a) + __builtin_abort (); + } + else if (i % 8 == 1) + { + if (v_vnx128qi[i] != b) + __builtin_abort (); + } + else if (i % 8 == 2) + { + if (v_vnx128qi[i] != c) + __builtin_abort (); + } + else if (i % 8 == 3) + { + if (v_vnx128qi[i] != d) + __builtin_abort (); + } + else if (i % 8 == 4) + { + if (v_vnx128qi[i] != e) + __builtin_abort (); + } + else if (i % 8 == 5) + { + if (v_vnx128qi[i] != f) + __builtin_abort (); + } + else if (i % 8 == 6) + { + if (v_vnx128qi[i] != g) + __builtin_abort (); + } + else if (i % 8 == 7) + { + if (v_vnx128qi[i] != h) + __builtin_abort (); + } + } + + return 0; +} diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-4.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-4.c new file mode 100644 index 00000000000..65c134b2262 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-4.c @@ -0,0 +1,77 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "repeat-4.c" + +int +main () +{ + float a = -9523.33; + float b = 8156.55; + + float v_vnx4sf[sizeof (vnx4sf) / sizeof (float)]; + f_vnx4sf (a, b, v_vnx4sf); + for (int i = 0; i < sizeof (vnx4sf) / sizeof (float); i++) + { + if (i % 2 == 0) + { + if (v_vnx4sf[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx4sf[i] != b) + __builtin_abort (); + } + } + + float v_vnx8sf[sizeof (vnx8sf) / sizeof (float)]; + f_vnx8sf (a, b, v_vnx8sf); + for (int i = 0; i < sizeof (vnx8sf) / sizeof (float); i++) + { + if (i % 2 == 0) + { + if (v_vnx8sf[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx8sf[i] != b) + __builtin_abort (); + } + } + + float v_vnx16sf[sizeof (vnx16sf) / sizeof (float)]; + f_vnx16sf (a, b, v_vnx16sf); + for (int i = 0; i < sizeof (vnx16sf) / sizeof (float); i++) + { + if (i % 2 == 0) + { + if (v_vnx16sf[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx16sf[i] != b) + __builtin_abort (); + } + } + + float v_vnx32sf[sizeof (vnx32sf) / sizeof (float)]; + f_vnx32sf (a, b, v_vnx32sf); + for (int i = 0; i < sizeof (vnx32sf) / sizeof (float); i++) + { + if (i % 2 == 0) + { + if (v_vnx32sf[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx32sf[i] != b) + __builtin_abort (); + } + } + + return 0; +} diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-5.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-5.c new file mode 100644 index 00000000000..aec4a9168b9 --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-5.c @@ -0,0 +1,124 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "repeat-5.c" + +int +main () +{ + int8_t a = -122; + + int8_t v_vnx2qi[sizeof (vnx2qi) / sizeof (int8_t)]; + f_vnx2qi (a, v_vnx2qi); + for (int i = 0; i < sizeof (vnx2qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx2qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx2qi[i] != -88) + __builtin_abort (); + } + } + + int8_t v_vnx4qi[sizeof (vnx4qi) / sizeof (int8_t)]; + f_vnx4qi (a, v_vnx4qi); + for (int i = 0; i < sizeof (vnx4qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx4qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx4qi[i] != -88) + __builtin_abort (); + } + } + + int8_t v_vnx8qi[sizeof (vnx8qi) / sizeof (int8_t)]; + f_vnx8qi (a, v_vnx8qi); + for (int i = 0; i < sizeof (vnx8qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx8qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx8qi[i] != -88) + __builtin_abort (); + } + } + + int8_t v_vnx16qi[sizeof (vnx16qi) / sizeof (int8_t)]; + f_vnx16qi (a, v_vnx16qi); + for (int i = 0; i < sizeof (vnx16qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx16qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx16qi[i] != -88) + __builtin_abort (); + } + } + + int8_t v_vnx32qi[sizeof (vnx32qi) / sizeof (int8_t)]; + f_vnx32qi (a, v_vnx32qi); + for (int i = 0; i < sizeof (vnx32qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx32qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx32qi[i] != -88) + __builtin_abort (); + } + } + + int8_t v_vnx64qi[sizeof (vnx64qi) / sizeof (int8_t)]; + f_vnx64qi (a, v_vnx64qi); + for (int i = 0; i < sizeof (vnx64qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx64qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx64qi[i] != -88) + __builtin_abort (); + } + } + + int8_t v_vnx128qi[sizeof (vnx128qi) / sizeof (int8_t)]; + f_vnx128qi (a, v_vnx128qi); + for (int i = 0; i < sizeof (vnx128qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx128qi[i] != a) + __builtin_abort (); + } + else + { + if (v_vnx128qi[i] != -88) + __builtin_abort (); + } + } + + return 0; +} diff --git a/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-6.c b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-6.c new file mode 100644 index 00000000000..6685a0d331a --- /dev/null +++ b/gcc/testsuite/gcc.target/riscv/rvv/autovec/repeat_run-6.c @@ -0,0 +1,122 @@ +/* { dg-do run { target { riscv_vector } } } */ +/* { dg-options "--param riscv-autovec-preference=fixed-vlmax" } */ + +#include "repeat-6.c" + +int +main () +{ + int8_t v_vnx2qi[sizeof (vnx2qi) / sizeof (int8_t)]; + f_vnx2qi (v_vnx2qi); + for (int i = 0; i < sizeof (vnx2qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx2qi[i] != -33) + __builtin_abort (); + } + else + { + if (v_vnx2qi[i] != -123) + __builtin_abort (); + } + } + + int8_t v_vnx4qi[sizeof (vnx4qi) / sizeof (int8_t)]; + f_vnx4qi (v_vnx4qi); + for (int i = 0; i < sizeof (vnx4qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx4qi[i] != -33) + __builtin_abort (); + } + else + { + if (v_vnx4qi[i] != -123) + __builtin_abort (); + } + } + + int8_t v_vnx8qi[sizeof (vnx8qi) / sizeof (int8_t)]; + f_vnx8qi (v_vnx8qi); + for (int i = 0; i < sizeof (vnx8qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx8qi[i] != -33) + __builtin_abort (); + } + else + { + if (v_vnx8qi[i] != -123) + __builtin_abort (); + } + } + + int8_t v_vnx16qi[sizeof (vnx16qi) / sizeof (int8_t)]; + f_vnx16qi (v_vnx16qi); + for (int i = 0; i < sizeof (vnx16qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx16qi[i] != -33) + __builtin_abort (); + } + else + { + if (v_vnx16qi[i] != -123) + __builtin_abort (); + } + } + + int8_t v_vnx32qi[sizeof (vnx32qi) / sizeof (int8_t)]; + f_vnx32qi (v_vnx32qi); + for (int i = 0; i < sizeof (vnx32qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx32qi[i] != -33) + __builtin_abort (); + } + else + { + if (v_vnx32qi[i] != -123) + __builtin_abort (); + } + } + + int8_t v_vnx64qi[sizeof (vnx64qi) / sizeof (int8_t)]; + f_vnx64qi (v_vnx64qi); + for (int i = 0; i < sizeof (vnx64qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx64qi[i] != -33) + __builtin_abort (); + } + else + { + if (v_vnx64qi[i] != -123) + __builtin_abort (); + } + } + + int8_t v_vnx128qi[sizeof (vnx128qi) / sizeof (int8_t)]; + f_vnx128qi (v_vnx128qi); + for (int i = 0; i < sizeof (vnx128qi) / sizeof (int8_t); i++) + { + if (i % 2 == 0) + { + if (v_vnx128qi[i] != -33) + __builtin_abort (); + } + else + { + if (v_vnx128qi[i] != -123) + __builtin_abort (); + } + } + + return 0; +} -- 2.36.1