On Fri, 25 Oct 2019 at 14:18, Richard Sandiford <richard.sandif...@arm.com> wrote: > > Hi Prathamesh, > > I've just committed a patch that fixes a large number of SVE > reduction-related failures. Could you rebase and retest on top of that? > Sorry for messing you around, but regression testing based on the state > before the patch wouldn't have been that meaningful. In particular... > > Prathamesh Kulkarni <prathamesh.kulka...@linaro.org> writes: > > diff --git a/gcc/tree-vect-loop.c b/gcc/tree-vect-loop.c > > index a70d52eb2ca..82814e2c2af 100644 > > --- a/gcc/tree-vect-loop.c > > +++ b/gcc/tree-vect-loop.c > > @@ -6428,6 +6428,7 @@ vectorizable_reduction (stmt_vec_info stmt_info, > > slp_tree slp_node, > > if (loop_vinfo && LOOP_VINFO_CAN_FULLY_MASK_P (loop_vinfo)) > > { > > if (reduction_type != FOLD_LEFT_REDUCTION > > + && reduction_type != EXTRACT_LAST_REDUCTION > > && !mask_by_cond_expr > > && (cond_fn == IFN_LAST > > || !direct_internal_fn_supported_p (cond_fn, vectype_in, > > ...after today's patch, it's instead necessary to remove: > > if (loop_vinfo > && LOOP_VINFO_CAN_FULLY_MASK_P (loop_vinfo) > && reduction_type == EXTRACT_LAST_REDUCTION) > { > if (dump_enabled_p ()) > dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location, > "can't yet use a fully-masked loop for" > " EXTRACT_LAST_REDUCTION.\n"); > LOOP_VINFO_CAN_FULLY_MASK_P (loop_vinfo) = false; > } > > from vectorizable_condition. We no longer need any changes to > vectorizable_reduction itself. > > > @@ -10180,18 +10181,29 @@ vectorizable_condition (stmt_vec_info stmt_info, > > gimple_stmt_iterator *gsi, > > vec != { 0, ... } (masked in the MASK_LOAD, > > unmasked in the VEC_COND_EXPR). */ > > > > - if (loop_mask) > > + if (masks) > > { > > - if (COMPARISON_CLASS_P (vec_compare)) > > + unsigned vec_num = vec_oprnds0.length (); > > + loop_mask = vect_get_loop_mask (gsi, masks, vec_num * ncopies, > > + vectype, vec_num * j + i); > > Ah... now that the two cases are merged (good!), just "if (masks)" isn't > right after all, sorry for the misleading comment. I think this should > instead be: > > /* Force vec_compare to be an SSA_NAME rather than a comparison, > in cases where that's necessary. */ > if (masks || reduction_type == EXTRACT_LAST_REDUCTION) > { > > Not doing that would break unmasked EXTRACT_LAST_REDUCTIONs. Ah right, thanks for pointing out! > > Then make the existing: > > tree tmp2 = make_ssa_name (vec_cmp_type); > gassign *g = gimple_build_assign (tmp2, BIT_AND_EXPR, > vec_compare, loop_mask); > vect_finish_stmt_generation (stmt_info, g, gsi); > vec_compare = tmp2; > > conditional on "if (masks)" only, and defer the calculation of loop_mask > to this point too. > > [ It ould be good to spot-check that aarch64-sve.exp passes after making > the changes to the stmt-generation part of vectorizable_condition, > but before removing the: > > LOOP_VINFO_CAN_FULLY_MASK_P (loop_vinfo) = false; > > quoted above. That would show that unmasked fold-left reductions > still work after the changes. > > There are still some lingering cases in which we can test unmasked > SVE loops directly, but they're becoming rarer and should eventually > go away altogether. So I don't think it's worth trying to construct > an unmasked test for the testsuite. ] > > > + > > + if (!is_gimple_val (vec_compare)) > > + { > > + tree vec_compare_name = make_ssa_name (vec_cmp_type); > > + gassign *new_stmt = gimple_build_assign > > (vec_compare_name, > > + vec_compare); > > + vect_finish_stmt_generation (stmt_info, new_stmt, gsi); > > + vec_compare = vec_compare_name; > > + } > > Should use tab-based indentation. Thanks for the suggestions, does the attached version look OK ? Comparing aarch64-sve.exp before/after patch shows no regressions, bootstrap+test in progress.
Thanks, Prathamesh > > Thanks, > Richard
diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_1.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_1.c index d4f9b0b6a94..d3ea52dea47 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_1.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_1.c @@ -1,5 +1,5 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */ #define N 32 @@ -17,4 +17,5 @@ condition_reduction (int *a, int min_v) return last; } -/* { dg-final { scan-assembler {\tclastb\tw[0-9]+, p[0-7], w[0-9]+, z[0-9]+\.s} } } */ +/* { dg-final { scan-tree-dump "using a fully-masked loop." "vect" } } */ +/* { dg-final { scan-assembler {\tclastb\ts[0-9]+, p[0-7], s[0-9]+, z[0-9]+\.s} } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_2.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_2.c index 2c49bd3b0f0..c222b707912 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_2.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_2.c @@ -1,5 +1,5 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */ #include <stdint.h> @@ -23,4 +23,5 @@ condition_reduction (TYPE *a, TYPE min_v) return last; } -/* { dg-final { scan-assembler {\tclastb\tw[0-9]+, p[0-7]+, w[0-9]+, z[0-9]+\.s} } } */ +/* { dg-final { scan-tree-dump "using a fully-masked loop." "vect" } } */ +/* { dg-final { scan-assembler {\tclastb\ts[0-9]+, p[0-7], s[0-9]+, z[0-9]+\.s} } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_3.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_3.c index 35344f446c6..5aaa71f948d 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_3.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_3.c @@ -1,8 +1,9 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */ #define TYPE uint8_t #include "clastb_2.c" -/* { dg-final { scan-assembler {\tclastb\tw[0-9]+, p[0-7]+, w[0-9]+, z[0-9]+\.b} } } */ +/* { dg-final { scan-tree-dump "using a fully-masked loop." "vect" } } */ +/* { dg-final { scan-assembler {\tclastb\tb[0-9]+, p[0-7], b[0-9]+, z[0-9]+\.b} } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_4.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_4.c index ce58abd6161..b4db170ea06 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_4.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_4.c @@ -1,8 +1,9 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */ #define TYPE int16_t #include "clastb_2.c" +/* { dg-final { scan-tree-dump "using a fully-masked loop." "vect" } } */ /* { dg-final { scan-assembler {\tclastb\tw[0-9]+, p[0-7], w[0-9]+, z[0-9]+\.h} } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_5.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_5.c index 2b9783d6627..28d40a01f93 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_5.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_5.c @@ -1,8 +1,9 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */ #define TYPE uint64_t #include "clastb_2.c" -/* { dg-final { scan-assembler {\tclastb\tx[0-9]+, p[0-7], x[0-9]+, z[0-9]+\.d} } } */ +/* { dg-final { scan-tree-dump "using a fully-masked loop." "vect" } } */ +/* { dg-final { scan-assembler {\tclastb\td[0-9]+, p[0-7], d[0-9]+, z[0-9]+\.d} } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_6.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_6.c index c47d303f730..38632a21be1 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_6.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_6.c @@ -1,5 +1,5 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */ #define N 32 @@ -21,4 +21,5 @@ condition_reduction (TYPE *a, TYPE min_v) return last; } +/* { dg-final { scan-tree-dump "using a fully-masked loop." "vect" } } */ /* { dg-final { scan-assembler {\tclastb\ts[0-9]+, p[0-7], s[0-9]+, z[0-9]+\.s} } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_7.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_7.c index 3345f874a39..e5307d2edc8 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_7.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_7.c @@ -1,7 +1,8 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details --save-temps" } */ #define TYPE double #include "clastb_6.c" +/* { dg-final { scan-tree-dump "using a fully-masked loop." "vect" } } */ /* { dg-final { scan-assembler {\tclastb\td[0-9]+, p[0-7], d[0-9]+, z[0-9]+\.d} } } */ diff --git a/gcc/testsuite/gcc.target/aarch64/sve/clastb_8.c b/gcc/testsuite/gcc.target/aarch64/sve/clastb_8.c index d86a428a7fa..583fc8d8d6d 100644 --- a/gcc/testsuite/gcc.target/aarch64/sve/clastb_8.c +++ b/gcc/testsuite/gcc.target/aarch64/sve/clastb_8.c @@ -1,5 +1,5 @@ /* { dg-do assemble { target aarch64_asm_sve_ok } } */ -/* { dg-options "-O2 -ftree-vectorize -msve-vector-bits=256 --save-temps" } */ +/* { dg-options "-O2 -ftree-vectorize -fdump-tree-vect-details -msve-vector-bits=256 --save-temps" } */ #include <stdint.h> @@ -19,6 +19,7 @@ TEST_TYPE (uint16_t); TEST_TYPE (uint32_t); TEST_TYPE (uint64_t); +/* { dg-final { scan-tree-dump-times "using a fully-masked loop." 4 "vect" } } */ /* { dg-final { scan-assembler {\tclastb\t(b[0-9]+), p[0-7], \1, z[0-9]+\.b\n} } } */ /* { dg-final { scan-assembler {\tclastb\t(h[0-9]+), p[0-7], \1, z[0-9]+\.h\n} } } */ /* { dg-final { scan-assembler {\tclastb\t(s[0-9]+), p[0-7], \1, z[0-9]+\.s\n} } } */ diff --git a/gcc/tree-vect-stmts.c b/gcc/tree-vect-stmts.c index 19ac82fe4e3..588145ebdce 100644 --- a/gcc/tree-vect-stmts.c +++ b/gcc/tree-vect-stmts.c @@ -10050,16 +10050,6 @@ vectorizable_condition (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, return false; } } - if (loop_vinfo - && LOOP_VINFO_CAN_FULLY_MASK_P (loop_vinfo) - && reduction_type == EXTRACT_LAST_REDUCTION) - { - if (dump_enabled_p ()) - dump_printf_loc (MSG_MISSED_OPTIMIZATION, vect_location, - "can't yet use a fully-masked loop for" - " EXTRACT_LAST_REDUCTION.\n"); - LOOP_VINFO_CAN_FULLY_MASK_P (loop_vinfo) = false; - } if (expand_vec_cond_expr_p (vectype, comp_vectype, cond_code)) { @@ -10089,31 +10079,31 @@ vectorizable_condition (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, /* Handle cond expr. */ for (j = 0; j < ncopies; j++) { - tree loop_mask = NULL_TREE; bool swap_cond_operands = false; /* See whether another part of the vectorized code applies a loop mask to the condition, or to its inverse. */ + vec_loop_masks *masks = NULL; if (loop_vinfo && LOOP_VINFO_FULLY_MASKED_P (loop_vinfo)) { - scalar_cond_masked_key cond (cond_expr, ncopies); - if (loop_vinfo->scalar_cond_masked_set.contains (cond)) - { - vec_loop_masks *masks = &LOOP_VINFO_MASKS (loop_vinfo); - loop_mask = vect_get_loop_mask (gsi, masks, ncopies, vectype, j); - } + if (reduction_type == EXTRACT_LAST_REDUCTION) + masks = &LOOP_VINFO_MASKS (loop_vinfo); else { - bool honor_nans = HONOR_NANS (TREE_TYPE (cond.op0)); - cond.code = invert_tree_comparison (cond.code, honor_nans); + scalar_cond_masked_key cond (cond_expr, ncopies); if (loop_vinfo->scalar_cond_masked_set.contains (cond)) + masks = &LOOP_VINFO_MASKS (loop_vinfo); + else { - vec_loop_masks *masks = &LOOP_VINFO_MASKS (loop_vinfo); - loop_mask = vect_get_loop_mask (gsi, masks, ncopies, - vectype, j); - cond_code = cond.code; - swap_cond_operands = true; + bool honor_nans = HONOR_NANS (TREE_TYPE (cond.op0)); + cond.code = invert_tree_comparison (cond.code, honor_nans); + if (loop_vinfo->scalar_cond_masked_set.contains (cond)) + { + masks = &LOOP_VINFO_MASKS (loop_vinfo); + cond_code = cond.code; + swap_cond_operands = true; + } } } } @@ -10248,28 +10238,10 @@ vectorizable_condition (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, vec != { 0, ... } (masked in the MASK_LOAD, unmasked in the VEC_COND_EXPR). */ - if (loop_mask) - { - if (COMPARISON_CLASS_P (vec_compare)) - { - tree tmp = make_ssa_name (vec_cmp_type); - tree op0 = TREE_OPERAND (vec_compare, 0); - tree op1 = TREE_OPERAND (vec_compare, 1); - gassign *g = gimple_build_assign (tmp, - TREE_CODE (vec_compare), - op0, op1); - vect_finish_stmt_generation (stmt_info, g, gsi); - vec_compare = tmp; - } - - tree tmp2 = make_ssa_name (vec_cmp_type); - gassign *g = gimple_build_assign (tmp2, BIT_AND_EXPR, - vec_compare, loop_mask); - vect_finish_stmt_generation (stmt_info, g, gsi); - vec_compare = tmp2; - } + /* Force vec_compare to be an SSA_NAME rather than a comparison, + in cases where that's necessary. */ - if (reduction_type == EXTRACT_LAST_REDUCTION) + if (masks || reduction_type == EXTRACT_LAST_REDUCTION) { if (!is_gimple_val (vec_compare)) { @@ -10279,6 +10251,7 @@ vectorizable_condition (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, vect_finish_stmt_generation (stmt_info, new_stmt, gsi); vec_compare = vec_compare_name; } + if (must_invert_cmp_result) { tree vec_compare_name = make_ssa_name (vec_cmp_type); @@ -10288,6 +10261,23 @@ vectorizable_condition (stmt_vec_info stmt_info, gimple_stmt_iterator *gsi, vect_finish_stmt_generation (stmt_info, new_stmt, gsi); vec_compare = vec_compare_name; } + + if (masks) + { + unsigned vec_num = vec_oprnds0.length (); + tree loop_mask + = vect_get_loop_mask (gsi, masks, vec_num * ncopies, + vectype, vec_num * j + i); + tree tmp2 = make_ssa_name (vec_cmp_type); + gassign *g = gimple_build_assign (tmp2, BIT_AND_EXPR, + vec_compare, loop_mask); + vect_finish_stmt_generation (stmt_info, g, gsi); + vec_compare = tmp2; + } + } + + if (reduction_type == EXTRACT_LAST_REDUCTION) + { gcall *new_stmt = gimple_build_call_internal (IFN_FOLD_EXTRACT_LAST, 3, else_clause, vec_compare, vec_then_clause);