In order to support vectorization of loops with multiple exits, this patch adds the implementation of the conditional branch optab for LoongArch LSX/LASX instructions.
This patch causes the gen-vect-{2,25}.c tests to fail. This is because the support for vectorizing loops with multiple exits has vectorized the loop checking the results. The failure is due to an issue in the test case's own implementation. gcc/ChangeLog: * config/loongarch/simd.md (cbranch<mode>4): New expander. gcc/testsuite/ChangeLog: * lib/target-supports.exp (check_effective_target_vect_early_break_hw, check_effective_target_vect_early_break): Support LoongArch LSX. * gcc.target/loongarch/vector/lasx/lasx-vseteqz.c: New test. * gcc.target/loongarch/vector/lsx/lsx-vseteqz.c: New test. Co-authored-by: Deng Jianbo <dengjia...@loongson.cn> --- v1-v2:Implement vector cbranch in simd.md instead of duplicating its implementation in lasx.md and lsx.md. v2-v3: Remove unnecessary commit messages. --- gcc/config/loongarch/simd.md | 30 +++++++++++++++++++ .../loongarch/vector/lasx/lasx-vseteqz.c | 14 +++++++++ .../loongarch/vector/lsx/lsx-vseteqz.c | 15 ++++++++++ gcc/testsuite/lib/target-supports.exp | 2 ++ 4 files changed, 61 insertions(+) create mode 100644 gcc/testsuite/gcc.target/loongarch/vector/lasx/lasx-vseteqz.c create mode 100644 gcc/testsuite/gcc.target/loongarch/vector/lsx/lsx-vseteqz.c diff --git a/gcc/config/loongarch/simd.md b/gcc/config/loongarch/simd.md index fc3d98a4340..4d006902753 100644 --- a/gcc/config/loongarch/simd.md +++ b/gcc/config/loongarch/simd.md @@ -516,6 +516,36 @@ (define_expand "vec_cmpu<mode><allmode_i>" DONE; }) +;; cbranch +(define_expand "cbranch<mode>4" + [(set (pc) + (if_then_else + (match_operator 0 "equality_operator" + [(match_operand:IVEC 1 "register_operand") + (match_operand:IVEC 2 "reg_or_vector_same_val_operand")]) + (label_ref (match_operand 3 "")) + (pc)))] + "" +{ + RTX_CODE code = GET_CODE (operands[0]); + rtx tmp = operands[1]; + rtx const0 = CONST0_RTX (SImode); + + /* If comparing against a non-zero vector we have to do a comparison first + so we can have a != 0 comparison with the result. */ + if (operands[2] != CONST0_RTX (<MODE>mode)) + { + tmp = gen_reg_rtx (<MODE>mode); + emit_insn (gen_xor<mode>3 (tmp, operands[1], operands[2])); + } + + if (code == NE) + emit_jump_insn (gen_<simd_isa>_<x>bnz_v_b (operands[3], tmp, const0)); + else + emit_jump_insn (gen_<simd_isa>_<x>bz_v_b (operands[3], tmp, const0)); + DONE; +}) + ; The LoongArch SX Instructions. (include "lsx.md") diff --git a/gcc/testsuite/gcc.target/loongarch/vector/lasx/lasx-vseteqz.c b/gcc/testsuite/gcc.target/loongarch/vector/lasx/lasx-vseteqz.c new file mode 100644 index 00000000000..1f69a80a784 --- /dev/null +++ b/gcc/testsuite/gcc.target/loongarch/vector/lasx/lasx-vseteqz.c @@ -0,0 +1,14 @@ +/* { dg-do compile } */ +/* { dg-options "-O3 -mlasx" } */ +/* { dg-final { scan-assembler "\txvset.*.v\t" } } */ +/* { dg-final { scan-assembler "bcnez" } } */ + +int +foo (int N) +{ + for (int i = 0; i <= N; i++) + if (i * i == N) + return i; + return -1; +} + diff --git a/gcc/testsuite/gcc.target/loongarch/vector/lsx/lsx-vseteqz.c b/gcc/testsuite/gcc.target/loongarch/vector/lsx/lsx-vseteqz.c new file mode 100644 index 00000000000..2536bb7945e --- /dev/null +++ b/gcc/testsuite/gcc.target/loongarch/vector/lsx/lsx-vseteqz.c @@ -0,0 +1,15 @@ +/* { dg-do compile } */ +/* { dg-options "-O3 -mlsx" } */ +/* { dg-final { scan-assembler "\tvset.*.v\t" } } */ +/* { dg-final { scan-assembler "bcnez" } } */ + +int +foo (int N) +{ + for (int i = 0; i <= N; i++) + if (i * i == N) + return i; + + return -1; +} + diff --git a/gcc/testsuite/lib/target-supports.exp b/gcc/testsuite/lib/target-supports.exp index a16e9534ccd..30ee528d230 100644 --- a/gcc/testsuite/lib/target-supports.exp +++ b/gcc/testsuite/lib/target-supports.exp @@ -4431,6 +4431,7 @@ proc check_effective_target_vect_early_break { } { || [check_effective_target_sse4] || [istarget amdgcn-*-*] || [check_effective_target_riscv_v] + || [check_effective_target_loongarch_sx] }}] } @@ -4447,6 +4448,7 @@ proc check_effective_target_vect_early_break_hw { } { || [check_sse4_hw_available] || [istarget amdgcn-*-*] || [check_effective_target_riscv_v_ok] + || [check_effective_target_loongarch_sx_hw] }}] } -- 2.20.1