From: Yunze Zhu <yunze...@linux.alibaba.com>
This commit add support for xtheadvector-specific unit-stride segment
load/store intrinsics with b/h/w suffix. We also defined enum to be used
in thead-vector.md
https://github.com/XUANTIE-RV/thead-extension-spec/pull/66
V2:
Change to reuse existed thead function base th_loadstore_width. Fix indentation
error.
gcc/ChangeLog:
* config/riscv/riscv-vector-builtins-bases.cc (BASE): New base_name.
* config/riscv/riscv-vector-builtins-bases.h: New function_base.
* config/riscv/riscv-vector-builtins-shapes.cc (build_th_loadstore):
Define new builtin shapes.
(struct th_seg_loadstore_def): Ditto.
(build): Ditto.
(SHAPE): Ditto.
* config/riscv/riscv-vector-builtins-shapes.h: Ditto.
* config/riscv/riscv-vector-builtins-types.def
(DEF_RVV_TH_INT_TUPLE_OPS): New type.
(DEF_RVV_TH_UINT_TUPLE_OPS): Ditto.
(vint8m1x2_t): Ditto.
(vint8m1x3_t): Ditto.
(vint8m1x4_t): Ditto.
(vint8m1x5_t): Ditto.
(vint8m1x6_t): Ditto.
(vint8m1x7_t): Ditto.
(vint8m1x8_t): Ditto.
(vint8m2x2_t): Ditto.
(vint8m2x3_t): Ditto.
(vint8m2x4_t): Ditto.
(vint8m4x2_t): Ditto.
(vint16m1x2_t): Ditto.
(vint16m1x3_t): Ditto.
(vint16m1x4_t): Ditto.
(vint16m1x5_t): Ditto.
(vint16m1x6_t): Ditto.
(vint16m1x7_t): Ditto.
(vint16m1x8_t): Ditto.
(vint16m2x2_t): Ditto.
(vint16m2x3_t): Ditto.
(vint16m2x4_t): Ditto.
(vint16m4x2_t): Ditto.
(vint32m1x2_t): Ditto.
(vint32m1x3_t): Ditto.
(vint32m1x4_t): Ditto.
(vint32m1x5_t): Ditto.
(vint32m1x6_t): Ditto.
(vint32m1x7_t): Ditto.
(vint32m1x8_t): Ditto.
(vint32m2x2_t): Ditto.
(vint32m2x3_t): Ditto.
(vint32m2x4_t): Ditto.
(vint32m4x2_t): Ditto.
(vint64m1x2_t): Ditto.
(vint64m1x3_t): Ditto.
(vint64m1x4_t): Ditto.
(vint64m1x5_t): Ditto.
(vint64m1x6_t): Ditto.
(vint64m1x7_t): Ditto.
(vint64m1x8_t): Ditto.
(vint64m2x2_t): Ditto.
(vint64m2x3_t): Ditto.
(vint64m2x4_t): Ditto.
(vint64m4x2_t): Ditto.
(vuint8m1x2_t): Ditto.
(vuint8m1x3_t): Ditto.
(vuint8m1x4_t): Ditto.
(vuint8m1x5_t): Ditto.
(vuint8m1x6_t): Ditto.
(vuint8m1x7_t): Ditto.
(vuint8m1x8_t): Ditto.
(vuint8m2x2_t): Ditto.
(vuint8m2x3_t): Ditto.
(vuint8m2x4_t): Ditto.
(vuint8m4x2_t): Ditto.
(vuint16m1x2_t): Ditto.
(vuint16m1x3_t): Ditto.
(vuint16m1x4_t): Ditto.
(vuint16m1x5_t): Ditto.
(vuint16m1x6_t): Ditto.
(vuint16m1x7_t): Ditto.
(vuint16m1x8_t): Ditto.
(vuint16m2x2_t): Ditto.
(vuint16m2x3_t): Ditto.
(vuint16m2x4_t): Ditto.
(vuint16m4x2_t): Ditto.
(vuint32m1x2_t): Ditto.
(vuint32m1x3_t): Ditto.
(vuint32m1x4_t): Ditto.
(vuint32m1x5_t): Ditto.
(vuint32m1x6_t): Ditto.
(vuint32m1x7_t): Ditto.
(vuint32m1x8_t): Ditto.
(vuint32m2x2_t): Ditto.
(vuint32m2x3_t): Ditto.
(vuint32m2x4_t): Ditto.
(vuint32m4x2_t): Ditto.
(vuint64m1x2_t): Ditto.
(vuint64m1x3_t): Ditto.
(vuint64m1x4_t): Ditto.
(vuint64m1x5_t): Ditto.
(vuint64m1x6_t): Ditto.
(vuint64m1x7_t): Ditto.
(vuint64m1x8_t): Ditto.
(vuint64m2x2_t): Ditto.
(vuint64m2x3_t): Ditto.
(vuint64m2x4_t): Ditto.
(vuint64m4x2_t): Ditto.
* config/riscv/riscv-vector-builtins.cc (DEF_RVV_TH_INT_TUPLE_OPS): New
builtins def.
(DEF_RVV_TH_UINT_TUPLE_OPS): Ditto.
* config/riscv/t-riscv: Add include for Thead files.
* config/riscv/thead-vector-builtins-functions.def (vlsegb): New
intrinsics def.
(vlsegh): Ditto.
(vlsegw): Ditto.
(vlsegbu): Ditto.
(vlseghu): Ditto.
(vlsegwu): Ditto.
(vssegb): Ditto.
(vssegh): Ditto.
(vssegw): Ditto.
* config/riscv/thead-vector.md
(@pred_th_unit_seg_load<vlmem_op_attr><mode>): New RTL mode.
(@pred_th_unit_seg_store<vlmem_op_attr><mode>): Ditto.
* config/riscv/thead.cc (th_asm_output_opcode): Implement Thead
function to add assembler insn code prefix/suffix.
gcc/testsuite/ChangeLog:
* gcc.target/riscv/rvv/xtheadvector/vlseg-vsseg.c: New test.
* gcc.target/riscv/rvv/xtheadvector/vlsegu-vsseg.c: New test.
---
.../riscv/riscv-vector-builtins-bases.cc | 132 ++++++++++-------
.../riscv/riscv-vector-builtins-bases.h | 9 ++
.../riscv/riscv-vector-builtins-shapes.cc | 75 +++++++++-
.../riscv/riscv-vector-builtins-shapes.h | 2 +
.../riscv/riscv-vector-builtins-types.def | 104 ++++++++++++++
gcc/config/riscv/riscv-vector-builtins.cc | 43 ++++++
gcc/config/riscv/t-riscv | 4 +
.../riscv/thead-vector-builtins-functions.def | 9 ++
gcc/config/riscv/thead-vector.md | 133 ++++++++++++++++++
gcc/config/riscv/thead.cc | 60 +++++---
.../riscv/rvv/xtheadvector/vlseg-vsseg.c | 118 ++++++++++++++++
.../riscv/rvv/xtheadvector/vlsegu-vsseg.c | 116 +++++++++++++++
12 files changed, 733 insertions(+), 72 deletions(-)
create mode 100644
gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlseg-vsseg.c
create mode 100644
gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlsegu-vsseg.c
diff --git a/gcc/config/riscv/riscv-vector-builtins-bases.cc
b/gcc/config/riscv/riscv-vector-builtins-bases.cc
index bf5172c6e04..4133a7846c4 100644
--- a/gcc/config/riscv/riscv-vector-builtins-bases.cc
+++ b/gcc/config/riscv/riscv-vector-builtins-bases.cc
@@ -2136,7 +2136,7 @@ public:
* th.vl(b/h/w)[u].v/th.vs(b/h/w)[u].v/th.vls(b/h/w)[u].v/th.vss(b/h/w)[u].v/
* th.vlx(b/h/w)[u].v/th.vs[u]x(b/h/w).v
* codegen. */
-template<bool STORE_P, lst_type LST_TYPE, int UNSPEC>
+template<bool STORE_P, lst_type LST_TYPE, bool SEG_P, int UNSPEC>
class th_loadstore_width : public function_base
{
public:
@@ -2161,33 +2161,45 @@ public:
rtx expand (function_expander &e) const override
{
gcc_assert (TARGET_XTHEADVECTOR);
- if (LST_TYPE == LST_INDEXED)
- {
- if (STORE_P)
- return e.use_exact_insn (
- code_for_pred_indexed_store_width (UNSPEC, UNSPEC,
- e.vector_mode ()));
- else
- return e.use_exact_insn (
- code_for_pred_indexed_load_width (UNSPEC, e.vector_mode ()));
- }
- else if (LST_TYPE == LST_STRIDED)
+ if (!SEG_P)
{
- if (STORE_P)
- return e.use_contiguous_store_insn (
- code_for_pred_strided_store_width (UNSPEC, e.vector_mode ()));
+ if (LST_TYPE == LST_INDEXED)
+ {
+ if (STORE_P)
+ return e.use_exact_insn (
+ code_for_pred_indexed_store_width (UNSPEC, UNSPEC,
+ e.vector_mode ()));
+ else
+ return e.use_exact_insn (
+ code_for_pred_indexed_load_width (UNSPEC, e.vector_mode ()));
+ }
+ else if (LST_TYPE == LST_STRIDED)
+ {
+ if (STORE_P)
+ return e.use_contiguous_store_insn (
+ code_for_pred_strided_store_width (UNSPEC, e.vector_mode ()));
+ else
+ return e.use_contiguous_load_insn (
+ code_for_pred_strided_load_width (UNSPEC, e.vector_mode ()));
+ }
else
- return e.use_contiguous_load_insn (
- code_for_pred_strided_load_width (UNSPEC, e.vector_mode ()));
+ {
+ if (STORE_P)
+ return e.use_contiguous_store_insn (
+ code_for_pred_store_width (UNSPEC, e.vector_mode ()));
+ else
+ return e.use_contiguous_load_insn (
+ code_for_pred_mov_width (UNSPEC, e.vector_mode ()));
+ }
}
else
{
if (STORE_P)
- return e.use_contiguous_store_insn (
- code_for_pred_store_width (UNSPEC, e.vector_mode ()));
+ return e.use_exact_insn (
+ code_for_pred_th_unit_seg_store (UNSPEC, e.vector_mode ()));
else
- return e.use_contiguous_load_insn (
- code_for_pred_mov_width (UNSPEC, e.vector_mode ()));
+ return e.use_exact_insn (
+ code_for_pred_th_unit_seg_load (UNSPEC, e.vector_mode ()));
}
}
};
@@ -2725,37 +2737,46 @@ static CONSTEXPR const seg_indexed_load<UNSPEC_ORDERED>
vloxseg_obj;
static CONSTEXPR const seg_indexed_store<UNSPEC_UNORDERED> vsuxseg_obj;
static CONSTEXPR const seg_indexed_store<UNSPEC_ORDERED> vsoxseg_obj;
static CONSTEXPR const vlsegff vlsegff_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE,
UNSPEC_TH_VLB> vlb_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE,
UNSPEC_TH_VLBU> vlbu_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE,
UNSPEC_TH_VLH> vlh_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE,
UNSPEC_TH_VLHU> vlhu_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE,
UNSPEC_TH_VLW> vlw_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE,
UNSPEC_TH_VLWU> vlwu_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE,
UNSPEC_TH_VLB> vsb_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE,
UNSPEC_TH_VLH> vsh_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE,
UNSPEC_TH_VLW> vsw_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, UNSPEC_TH_VLSB>
vlsb_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, UNSPEC_TH_VLSBU>
vlsbu_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, UNSPEC_TH_VLSH>
vlsh_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, UNSPEC_TH_VLSHU>
vlshu_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, UNSPEC_TH_VLSW>
vlsw_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, UNSPEC_TH_VLSWU>
vlswu_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_STRIDED, UNSPEC_TH_VLSB>
vssb_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_STRIDED, UNSPEC_TH_VLSH>
vssh_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_STRIDED, UNSPEC_TH_VLSW>
vssw_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, UNSPEC_TH_VLXB>
vlxb_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, UNSPEC_TH_VLXBU>
vlxbu_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, UNSPEC_TH_VLXH>
vlxh_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, UNSPEC_TH_VLXHU>
vlxhu_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, UNSPEC_TH_VLXW>
vlxw_obj;
-static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, UNSPEC_TH_VLXWU>
vlxwu_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, UNSPEC_TH_VLXB>
vsxb_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, UNSPEC_TH_VLXH>
vsxh_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, UNSPEC_TH_VLXW>
vsxw_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, UNSPEC_TH_VSUXB>
vsuxb_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, UNSPEC_TH_VSUXH>
vsuxh_obj;
-static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, UNSPEC_TH_VSUXW>
vsuxw_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLB> vlb_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLBU> vlbu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLH> vlh_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLHU> vlhu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLW> vlw_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLWU> vlwu_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLB> vsb_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLH> vsh_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE, false,
UNSPEC_TH_VLW> vsw_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, false,
UNSPEC_TH_VLSB> vlsb_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, false,
UNSPEC_TH_VLSBU> vlsbu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, false,
UNSPEC_TH_VLSH> vlsh_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, false,
UNSPEC_TH_VLSHU> vlshu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, false,
UNSPEC_TH_VLSW> vlsw_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_STRIDED, false,
UNSPEC_TH_VLSWU> vlswu_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_STRIDED, false,
UNSPEC_TH_VLSB> vssb_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_STRIDED, false,
UNSPEC_TH_VLSH> vssh_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_STRIDED, false,
UNSPEC_TH_VLSW> vssw_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, false,
UNSPEC_TH_VLXB> vlxb_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, false,
UNSPEC_TH_VLXBU> vlxbu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, false,
UNSPEC_TH_VLXH> vlxh_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, false,
UNSPEC_TH_VLXHU> vlxhu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, false,
UNSPEC_TH_VLXW> vlxw_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_INDEXED, false,
UNSPEC_TH_VLXWU> vlxwu_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, false,
UNSPEC_TH_VLXB> vsxb_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, false,
UNSPEC_TH_VLXH> vsxh_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, false,
UNSPEC_TH_VLXW> vsxw_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, false,
UNSPEC_TH_VSUXB> vsuxb_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, false,
UNSPEC_TH_VSUXH> vsuxh_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_INDEXED, false,
UNSPEC_TH_VSUXW> vsuxw_obj;
static CONSTEXPR const th_extract vext_x_v_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGB> vlsegb_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGH> vlsegh_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGW> vlsegw_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGBU> vlsegbu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGHU> vlseghu_obj;
+static CONSTEXPR const th_loadstore_width<false, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGWU> vlsegwu_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGB> vssegb_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGH> vssegh_obj;
+static CONSTEXPR const th_loadstore_width<true, LST_UNIT_STRIDE, true,
UNSPEC_TH_VLSEGW> vssegw_obj;
/* Crypto Vector */
static CONSTEXPR const vandn vandn_obj;
@@ -3086,6 +3107,15 @@ BASE (vsuxb)
BASE (vsuxh)
BASE (vsuxw)
BASE (vext_x_v)
+BASE (vlsegb)
+BASE (vlsegh)
+BASE (vlsegw)
+BASE (vlsegbu)
+BASE (vlseghu)
+BASE (vlsegwu)
+BASE (vssegb)
+BASE (vssegh)
+BASE (vssegw)
/* Crypto vector */
BASE (vandn)
BASE (vbrev)
diff --git a/gcc/config/riscv/riscv-vector-builtins-bases.h
b/gcc/config/riscv/riscv-vector-builtins-bases.h
index 4a2f6b4bdf7..be62fb1781e 100644
--- a/gcc/config/riscv/riscv-vector-builtins-bases.h
+++ b/gcc/config/riscv/riscv-vector-builtins-bases.h
@@ -317,6 +317,15 @@ extern const function_base *const vsuxb;
extern const function_base *const vsuxh;
extern const function_base *const vsuxw;
extern const function_base *const vext_x_v;
+extern const function_base *const vlsegb;
+extern const function_base *const vlsegh;
+extern const function_base *const vlsegw;
+extern const function_base *const vlsegbu;
+extern const function_base *const vlseghu;
+extern const function_base *const vlsegwu;
+extern const function_base *const vssegb;
+extern const function_base *const vssegh;
+extern const function_base *const vssegw;
/* Below function_base are Vectro Crypto*/
extern const function_base *const vandn;
extern const function_base *const vbrev;
diff --git a/gcc/config/riscv/riscv-vector-builtins-shapes.cc
b/gcc/config/riscv/riscv-vector-builtins-shapes.cc
index b855d4c5fa5..4aa8824451d 100644
--- a/gcc/config/riscv/riscv-vector-builtins-shapes.cc
+++ b/gcc/config/riscv/riscv-vector-builtins-shapes.cc
@@ -259,6 +259,11 @@ build_th_loadstore (function_builder &b, const
function_group_info &group,
return;
tree type = builtin_types[group.ops_infos.types[vec_type_idx].index].vector;
+ machine_mode mode = TYPE_MODE (type);
+ if (riscv_v_ext_tuple_mode_p (mode))
+ type = group.ops_infos.ret.get_tuple_subpart_type
+ (group.ops_infos.types[vec_type_idx].index);
+
if (strstr (group.base_name, "l")
&& strstr (group.base_name, "u")
&& !TYPE_UNSIGNED (TREE_TYPE (type)))
@@ -269,7 +274,6 @@ build_th_loadstore (function_builder &b, const
function_group_info &group,
&& TYPE_UNSIGNED (TREE_TYPE (type)))
return;
- machine_mode mode = TYPE_MODE (type);
int sew = GET_MODE_BITSIZE (GET_MODE_INNER (mode));
if (strstr (group.base_name, "h") && sew == 8)
return;
@@ -1389,6 +1393,74 @@ struct sf_vcix_def : public build_base
};
+/* th_seg_loadstore_def class. */
+struct th_seg_loadstore_def : public build_base {
+void build (function_builder &b,
+ const function_group_info &group) const override
+ {
+ for (unsigned int pred_idx = 0; group.preds[pred_idx] != NUM_PRED_TYPES;
+ ++pred_idx)
+ {
+ for (unsigned int vec_type_idx = 0;
+ group.ops_infos.types[vec_type_idx].index != NUM_VECTOR_TYPES;
+ ++vec_type_idx)
+ {
+ build_th_loadstore (b, group, pred_idx, vec_type_idx);
+ }
+ }
+ }
+
+ char *get_name (function_builder &b, const function_instance &instance,
+ bool overloaded_p) const override {
+ /* Return nullptr if it can not be overloaded. */
+ if (overloaded_p && !instance.base->can_be_overloaded_p (instance.pred))
+ return nullptr;
+
+ if (strstr (instance.base_name, "vlseg"))
+ b.append_name ("__riscv_th_vlseg");
+ else if (strstr (instance.base_name, "vsseg"))
+ b.append_name ("__riscv_th_vsseg");
+ else
+ gcc_unreachable ();
+
+ tree type = builtin_types[instance.type.index].vector;
+ machine_mode mode = TYPE_MODE (type);
+
+ int nf = get_nf (mode);
+ /* vop --> vop<nf>. */
+ b.append_nf (nf);
+
+ /* vop<nf> --> vop<nf><b/h/w>. */
+ if (strstr (instance.base_name, "segb"))
+ b.append_name ("b");
+ else if (strstr (instance.base_name, "segh"))
+ b.append_name ("h");
+ else if (strstr (instance.base_name, "segw"))
+ b.append_name ("w");
+ else
+ gcc_unreachable ();
+
+ if (strstr (instance.base_name, "l")
+ && TYPE_UNSIGNED (builtin_types[instance.type.index].scalar))
+ b.append_name ("u");
+
+ if (!overloaded_p)
+ {
+ /* vop<nf><b/h/w> --> vop<nf><b/h/w>_v. */
+ b.append_name (operand_suffixes[instance.op_info->op]);
+ /* vop<nf><b/h/w>_v --> vop<nf><b/h/w>_v_<type>. */
+ b.append_name (type_suffixes[instance.type.index].vector);
+ }
+
+ /* According to rvv-intrinsic-doc, it does not add "_m" suffix
+ for vop_m C++ overloaded API. */
+ if (overloaded_p && instance.pred == PRED_TYPE_m)
+ return b.finish_name ();
+ b.append_name (predication_suffixes[instance.pred]);
+ return b.finish_name ();
+ }
+};
+
SHAPE(vsetvl, vsetvl)
SHAPE(vsetvl, vsetvlmax)
SHAPE(loadstore, loadstore)
@@ -1427,4 +1499,5 @@ SHAPE (sf_vqmacc, sf_vqmacc)
SHAPE (sf_vfnrclip, sf_vfnrclip)
SHAPE(sf_vcix_se, sf_vcix_se)
SHAPE(sf_vcix, sf_vcix)
+SHAPE (th_seg_loadstore, th_seg_loadstore)
} // end namespace riscv_vector
diff --git a/gcc/config/riscv/riscv-vector-builtins-shapes.h
b/gcc/config/riscv/riscv-vector-builtins-shapes.h
index 2f2636ee386..f432d9edbff 100644
--- a/gcc/config/riscv/riscv-vector-builtins-shapes.h
+++ b/gcc/config/riscv/riscv-vector-builtins-shapes.h
@@ -64,6 +64,8 @@ extern const function_shape *const sf_vqmacc;
extern const function_shape *const sf_vfnrclip;
extern const function_shape *const sf_vcix_se;
extern const function_shape *const sf_vcix;
+/* Xtheadvector extension. */
+extern const function_shape *const th_seg_loadstore;
}
} // end namespace riscv_vector
diff --git a/gcc/config/riscv/riscv-vector-builtins-types.def
b/gcc/config/riscv/riscv-vector-builtins-types.def
index ade6644b56e..7b6cbb118bc 100644
--- a/gcc/config/riscv/riscv-vector-builtins-types.def
+++ b/gcc/config/riscv/riscv-vector-builtins-types.def
@@ -381,6 +381,18 @@ along with GCC; see the file COPYING3. If not see
#define DEF_RVV_X2_WU_OPS(TYPE, REQUIRE)
#endif
+/* Use "DEF_RVV_TH_INT_TUPLE_OPS" macro include all signed tuple types
+ which will be iterated and registered as intrinsic functions. */
+#ifndef DEF_RVV_TH_INT_TUPLE_OPS
+#define DEF_RVV_TH_INT_TUPLE_OPS(TYPE, REQUIRE)
+#endif
+
+/* Use "DEF_RVV_TH_UINT_TUPLE_OPS" macro include all unsigned tuple types
+ which will be iterated and registered as intrinsic functions. */
+#ifndef DEF_RVV_TH_UINT_TUPLE_OPS
+#define DEF_RVV_TH_UINT_TUPLE_OPS(TYPE, REQUIRE)
+#endif
+
DEF_RVV_I_OPS (vint8mf8_t, RVV_REQUIRE_ELEN_64)
DEF_RVV_I_OPS (vint8mf4_t, 0)
DEF_RVV_I_OPS (vint8mf2_t, 0)
@@ -1501,6 +1513,96 @@ DEF_RVV_X2_WU_OPS (vuint32m1_t, 0)
DEF_RVV_X2_WU_OPS (vuint32m2_t, 0)
DEF_RVV_X2_WU_OPS (vuint32m4_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m1x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m1x3_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m1x4_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m1x5_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m1x6_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m1x7_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m1x8_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m2x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m2x3_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m2x4_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint8m4x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m1x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m1x3_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m1x4_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m1x5_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m1x6_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m1x7_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m1x8_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m2x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m2x3_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m2x4_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint16m4x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m1x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m1x3_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m1x4_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m1x5_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m1x6_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m1x7_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m1x8_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m2x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m2x3_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m2x4_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint32m4x2_t, 0)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m1x2_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m1x3_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m1x4_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m1x5_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m1x6_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m1x7_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m1x8_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m2x2_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m2x3_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m2x4_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_INT_TUPLE_OPS (vint64m4x2_t, RVV_REQUIRE_ELEN_64)
+
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m1x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m1x3_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m1x4_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m1x5_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m1x6_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m1x7_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m1x8_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m2x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m2x3_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m2x4_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint8m4x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m1x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m1x3_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m1x4_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m1x5_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m1x6_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m1x7_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m1x8_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m2x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m2x3_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m2x4_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint16m4x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m1x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m1x3_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m1x4_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m1x5_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m1x6_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m1x7_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m1x8_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m2x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m2x3_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m2x4_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint32m4x2_t, 0)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m1x2_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m1x3_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m1x4_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m1x5_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m1x6_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m1x7_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m1x8_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m2x2_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m2x3_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m2x4_t, RVV_REQUIRE_ELEN_64)
+DEF_RVV_TH_UINT_TUPLE_OPS (vuint64m4x2_t, RVV_REQUIRE_ELEN_64)
+
#undef DEF_RVV_I_OPS
#undef DEF_RVV_U_OPS
#undef DEF_RVV_F_OPS
@@ -1559,3 +1661,5 @@ DEF_RVV_X2_WU_OPS (vuint32m4_t, 0)
#undef DEF_RVV_XFQF_OPS
#undef DEF_RVV_X2_U_OPS
#undef DEF_RVV_X2_WU_OPS
+#undef DEF_RVV_TH_INT_TUPLE_OPS
+#undef DEF_RVV_TH_UINT_TUPLE_OPS
diff --git a/gcc/config/riscv/riscv-vector-builtins.cc
b/gcc/config/riscv/riscv-vector-builtins.cc
index f3c706bfba9..00105749cad 100644
--- a/gcc/config/riscv/riscv-vector-builtins.cc
+++ b/gcc/config/riscv/riscv-vector-builtins.cc
@@ -571,6 +571,25 @@ static const rvv_type_info xfqf_ops[] = {
#include "riscv-vector-builtins-types.def"
{NUM_VECTOR_TYPES, 0}};
+/* A list of Int Tuple types will be registered for intrinsic functions. */
+static const rvv_type_info th_tuple_int_ops[] = {
+#define DEF_RVV_TH_INT_TUPLE_OPS(TYPE, REQUIRE) {VECTOR_TYPE_##TYPE, REQUIRE},
+#define DEF_RVV_TH_UINT_TUPLE_OPS(TYPE, REQUIRE) {VECTOR_TYPE_##TYPE, REQUIRE},
+#include "riscv-vector-builtins-types.def"
+ {NUM_VECTOR_TYPES, 0}};
+
+/* A list of Int Tuple types will be registered for intrinsic functions. */
+static const rvv_type_info th_tuple_sint_ops[] = {
+#define DEF_RVV_TH_INT_TUPLE_OPS(TYPE, REQUIRE) {VECTOR_TYPE_##TYPE, REQUIRE},
+#include "riscv-vector-builtins-types.def"
+ {NUM_VECTOR_TYPES, 0}};
+
+/* A list of Int Tuple types will be registered for intrinsic functions. */
+static const rvv_type_info th_tuple_uint_ops[] = {
+#define DEF_RVV_TH_UINT_TUPLE_OPS(TYPE, REQUIRE) {VECTOR_TYPE_##TYPE, REQUIRE},
+#include "riscv-vector-builtins-types.def"
+ {NUM_VECTOR_TYPES, 0}};
+
static CONSTEXPR const rvv_arg_type_info rvv_arg_type_info_end
= rvv_arg_type_info (NUM_BASE_TYPES);
@@ -3343,6 +3362,30 @@ static CONSTEXPR const rvv_op_info sf_vc_v_fvw_ops
rvv_arg_type_info (RVV_BASE_x2_vector), /* Return type */
sf_vc_fvw_args /* Args */};
+/* A static operand information for vector_type func (const scalar_type *)
+ * function registration. */
+static CONSTEXPR const rvv_op_info th_tuple_v_sint_scalar_const_ptr_ops
+ = {th_tuple_sint_ops, /* Types */
+ OP_TYPE_v, /* Suffix */
+ rvv_arg_type_info (RVV_BASE_vector),/* Return type */
+ scalar_const_ptr_args /* Args */};
+
+/* A static operand information for vector_type func (const scalar_type *)
+ * function registration. */
+static CONSTEXPR const rvv_op_info th_tuple_v_uint_scalar_const_ptr_ops
+ = {th_tuple_uint_ops, /* Types */
+ OP_TYPE_v, /* Suffix */
+ rvv_arg_type_info (RVV_BASE_vector),/* Return type */
+ scalar_const_ptr_args /* Args */};
+
+/* A static operand information for void func (scalar_type *, vector_type)
+ * function registration. */
+static CONSTEXPR const rvv_op_info th_tuple_v_int_scalar_ptr_ops
+ = {th_tuple_int_ops, /* Types */
+ OP_TYPE_v, /* Suffix */
+ rvv_arg_type_info (RVV_BASE_void), /* Return type */
+ scalar_ptr_args /* Args */};
+
/* A list of all RVV base function types. */
static CONSTEXPR const function_type_info function_types[] = {
#define DEF_RVV_TYPE_INDEX(
\
diff --git a/gcc/config/riscv/t-riscv b/gcc/config/riscv/t-riscv
index 12e2b6e999a..90d01380d70 100644
--- a/gcc/config/riscv/t-riscv
+++ b/gcc/config/riscv/t-riscv
@@ -11,6 +11,7 @@ riscv-builtins.o: $(srcdir)/config/riscv/riscv-builtins.cc
$(CONFIG_H) \
$(srcdir)/config/riscv/riscv-ftypes.def \
$(srcdir)/config/riscv/riscv-vector-builtins-types.def \
$(srcdir)/config/riscv/sifive-vector-builtins-functions.def \
+ $(srcdir)/config/riscv/thead-vector-builtins-functions.def \
$(srcdir)/config/riscv/riscv-modes.def \
$(srcdir)/config/riscv/riscv-cmo.def \
$(srcdir)/config/riscv/riscv-scalar-crypto.def
@@ -28,6 +29,7 @@ riscv-vector-builtins.o:
$(srcdir)/config/riscv/riscv-vector-builtins.cc \
$(srcdir)/config/riscv/sifive-vector-builtins-bases.h \
$(srcdir)/config/riscv/riscv-vector-builtins-types.def \
$(srcdir)/config/riscv/sifive-vector-builtins-functions.def \
+ $(srcdir)/config/riscv/thead-vector-builtins-functions.def \
$(RISCV_BUILTINS_H)
$(COMPILER) -c $(ALL_COMPILERFLAGS) $(ALL_CPPFLAGS) $(INCLUDES) \
$(srcdir)/config/riscv/riscv-vector-builtins.cc
@@ -178,6 +180,8 @@ $(srcdir)/config/riscv/riscv-vector-builtins.def:
riscv-vector-type-indexer.gen.
$(srcdir)/config/riscv/riscv-vector-builtins.h:
$(srcdir)/config/riscv/riscv-vector-builtins.def
$(srcdir)/config/riscv/sifive-vector-builtins-functions.def:
riscv-vector-type-indexer.gen.def
$(srcdir)/config/riscv/riscv-vector-builtins.h:
$(srcdir)/config/riscv/sifive-vector-builtins-functions.def
+$(srcdir)/config/riscv/thead-vector-builtins-functions.def:
riscv-vector-type-indexer.gen.def
+$(srcdir)/config/riscv/riscv-vector-builtins.h:
$(srcdir)/config/riscv/thead-vector-builtins-functions.def
riscv-vector-type-indexer.gen.def: s-riscv-vector-type-indexer.gen.defs ; @true
diff --git a/gcc/config/riscv/thead-vector-builtins-functions.def
b/gcc/config/riscv/thead-vector-builtins-functions.def
index fd3ba29bae9..c7702f4be48 100644
--- a/gcc/config/riscv/thead-vector-builtins-functions.def
+++ b/gcc/config/riscv/thead-vector-builtins-functions.def
@@ -34,6 +34,15 @@ DEF_RVV_FUNCTION (vsuxb, th_indexed_loadstore_width,
none_m_preds, all_v_scalar_
DEF_RVV_FUNCTION (vsuxh, th_indexed_loadstore_width, none_m_preds,
all_v_scalar_ptr_index_ops)
DEF_RVV_FUNCTION (vsuxw, th_indexed_loadstore_width, none_m_preds,
all_v_scalar_ptr_index_ops)
DEF_RVV_FUNCTION (vext_x_v, th_extract, none_preds, iu_x_s_u_ops)
+DEF_RVV_FUNCTION (vlsegb, th_seg_loadstore, full_preds,
th_tuple_v_sint_scalar_const_ptr_ops)
+DEF_RVV_FUNCTION (vlsegh, th_seg_loadstore, full_preds,
th_tuple_v_sint_scalar_const_ptr_ops)
+DEF_RVV_FUNCTION (vlsegw, th_seg_loadstore, full_preds,
th_tuple_v_sint_scalar_const_ptr_ops)
+DEF_RVV_FUNCTION (vlsegbu, th_seg_loadstore, full_preds,
th_tuple_v_uint_scalar_const_ptr_ops)
+DEF_RVV_FUNCTION (vlseghu, th_seg_loadstore, full_preds,
th_tuple_v_uint_scalar_const_ptr_ops)
+DEF_RVV_FUNCTION (vlsegwu, th_seg_loadstore, full_preds,
th_tuple_v_uint_scalar_const_ptr_ops)
+DEF_RVV_FUNCTION (vssegb, th_seg_loadstore, none_m_preds,
th_tuple_v_int_scalar_ptr_ops)
+DEF_RVV_FUNCTION (vssegh, th_seg_loadstore, none_m_preds,
th_tuple_v_int_scalar_ptr_ops)
+DEF_RVV_FUNCTION (vssegw, th_seg_loadstore, none_m_preds,
th_tuple_v_int_scalar_ptr_ops)
#undef REQUIRED_EXTENSIONS
#undef DEF_RVV_FUNCTION
diff --git a/gcc/config/riscv/thead-vector.md b/gcc/config/riscv/thead-vector.md
index 5a02debdd20..007682c3af5 100644
--- a/gcc/config/riscv/thead-vector.md
+++ b/gcc/config/riscv/thead-vector.md
@@ -25,6 +25,13 @@ (define_c_enum "unspec" [
UNSPEC_TH_VSUXW
UNSPEC_TH_VWLDST
+
+ UNSPEC_TH_VLSEGB
+ UNSPEC_TH_VLSEGBU
+ UNSPEC_TH_VLSEGH
+ UNSPEC_TH_VLSEGHU
+ UNSPEC_TH_VLSEGW
+ UNSPEC_TH_VLSEGWU
])
(define_int_iterator UNSPEC_TH_VLMEM_OP [
@@ -45,6 +52,12 @@ (define_int_iterator UNSPEC_TH_VLXMEM_OP [
UNSPEC_TH_VLXW UNSPEC_TH_VLXWU
])
+(define_int_iterator UNSPEC_TH_VLSEGMEM_OP[
+ UNSPEC_TH_VLSEGB UNSPEC_TH_VLSEGBU
+ UNSPEC_TH_VLSEGH UNSPEC_TH_VLSEGHU
+ UNSPEC_TH_VLSEGW UNSPEC_TH_VLSEGWU
+])
+
(define_int_attr vlmem_op_attr [
(UNSPEC_TH_VLB "b") (UNSPEC_TH_VLBU "bu")
(UNSPEC_TH_VLH "h") (UNSPEC_TH_VLHU "hu")
@@ -58,6 +71,9 @@ (define_int_attr vlmem_op_attr [
(UNSPEC_TH_VSUXB "b")
(UNSPEC_TH_VSUXH "h")
(UNSPEC_TH_VSUXW "w")
+ (UNSPEC_TH_VLSEGB "b") (UNSPEC_TH_VLSEGBU "bu")
+ (UNSPEC_TH_VLSEGH "h") (UNSPEC_TH_VLSEGHU "hu")
+ (UNSPEC_TH_VLSEGW "w") (UNSPEC_TH_VLSEGWU "wu")
])
(define_int_attr vlmem_order_attr [
@@ -90,9 +106,89 @@ (define_int_iterator UNSPEC_TH_VSXMEM_OP [
UNSPEC_TH_VSUXW
])
+(define_int_iterator UNSPEC_TH_VSSEGMEM_OP[
+ UNSPEC_TH_VLSEGB
+ UNSPEC_TH_VLSEGH
+ UNSPEC_TH_VLSEGW
+])
+
(define_mode_iterator V_VLS_VT [V VLS VT])
(define_mode_iterator V_VB_VLS_VT [V VB VLS VT])
+(define_mode_attr th_width[
+ (RVVM8QI "b") (RVVM4QI "b") (RVVM2QI "b") (RVVM1QI "b") (RVVMF2QI "b")
(RVVMF4QI "b") (RVVMF8QI "b")
+
+ (RVVM8HI "h") (RVVM4HI "h") (RVVM2HI "h") (RVVM1HI "h") (RVVMF2HI "h")
(RVVMF4HI "h")
+
+ (RVVM8BF "h") (RVVM4BF "h") (RVVM2BF "h") (RVVM1BF "h") (RVVMF2BF "h")
(RVVMF4BF "h")
+
+ (RVVM8HF "h") (RVVM4HF "h") (RVVM2HF "h") (RVVM1HF "h") (RVVMF2HF "h")
(RVVMF4HF "h")
+
+ (RVVM8SI "w") (RVVM4SI "w") (RVVM2SI "w") (RVVM1SI "w") (RVVMF2SI "w")
+
+ (RVVM8SF "w") (RVVM4SF "w") (RVVM2SF "w") (RVVM1SF "w") (RVVMF2SF "w")
+
+ (RVVM1x8QI "b") (RVVMF2x8QI "b") (RVVMF4x8QI "b") (RVVMF8x8QI "b")
+ (RVVM1x7QI "b") (RVVMF2x7QI "b") (RVVMF4x7QI "b") (RVVMF8x7QI "b")
+ (RVVM1x6QI "b") (RVVMF2x6QI "b") (RVVMF4x6QI "b") (RVVMF8x6QI "b")
+ (RVVM1x5QI "b") (RVVMF2x5QI "b") (RVVMF4x5QI "b") (RVVMF8x5QI "b")
+ (RVVM2x4QI "b") (RVVM1x4QI "b") (RVVMF2x4QI "b") (RVVMF4x4QI "b")
(RVVMF8x4QI "b")
+ (RVVM2x3QI "b") (RVVM1x3QI "b") (RVVMF2x3QI "b") (RVVMF4x3QI "b")
(RVVMF8x3QI "b")
+ (RVVM4x2QI "b") (RVVM2x2QI "b") (RVVM1x2QI "b") (RVVMF2x2QI "b") (RVVMF4x2QI
"b") (RVVMF8x2QI "b")
+
+ (RVVM1x8HI "h") (RVVMF2x8HI "h") (RVVMF4x8HI "h")
+ (RVVM1x7HI "h") (RVVMF2x7HI "h") (RVVMF4x7HI "h")
+ (RVVM1x6HI "h") (RVVMF2x6HI "h") (RVVMF4x6HI "h")
+ (RVVM1x5HI "h") (RVVMF2x5HI "h") (RVVMF4x5HI "h")
+ (RVVM2x4HI "h") (RVVM1x4HI "h") (RVVMF2x4HI "h") (RVVMF4x4HI "h")
+ (RVVM2x3HI "h") (RVVM1x3HI "h") (RVVMF2x3HI "h") (RVVMF4x3HI "h")
+ (RVVM4x2HI "h") (RVVM2x2HI "h") (RVVM1x2HI "h") (RVVMF2x2HI "h") (RVVMF4x2HI
"h")
+
+ (RVVM1x8BF "h") (RVVMF2x8BF "h") (RVVMF4x8BF "h")
+ (RVVM1x7BF "h") (RVVMF2x7BF "h") (RVVMF4x7BF "h")
+ (RVVM1x6BF "h") (RVVMF2x6BF "h") (RVVMF4x6BF "h")
+ (RVVM1x5BF "h") (RVVMF2x5BF "h") (RVVMF4x5BF "h")
+ (RVVM2x4BF "h") (RVVM1x4BF "h") (RVVMF2x4BF "h") (RVVMF4x4BF "h")
+ (RVVM2x3BF "h") (RVVM1x3BF "h") (RVVMF2x3BF "h") (RVVMF4x3BF "h")
+ (RVVM4x2BF "h") (RVVM2x2BF "h") (RVVM1x2BF "h") (RVVMF2x2BF "h") (RVVMF4x2BF
"h")
+
+ (RVVM1x8HF "h") (RVVMF2x8HF "h") (RVVMF4x8HF "h")
+ (RVVM1x7HF "h") (RVVMF2x7HF "h") (RVVMF4x7HF "h")
+ (RVVM1x6HF "h") (RVVMF2x6HF "h") (RVVMF4x6HF "h")
+ (RVVM1x5HF "h") (RVVMF2x5HF "h") (RVVMF4x5HF "h")
+ (RVVM2x4HF "h") (RVVM1x4HF "h") (RVVMF2x4HF "h") (RVVMF4x4HF "h")
+ (RVVM2x3HF "h") (RVVM1x3HF "h") (RVVMF2x3HF "h") (RVVMF4x3HF "h")
+ (RVVM4x2HF "h") (RVVM2x2HF "h") (RVVM1x2HF "h") (RVVMF2x2HF "h") (RVVMF4x2HF
"h")
+
+ (RVVM1x8SI "w") (RVVMF2x8SI "w")
+ (RVVM1x7SI "w") (RVVMF2x7SI "w")
+ (RVVM1x6SI "w") (RVVMF2x6SI "w")
+ (RVVM1x5SI "w") (RVVMF2x5SI "w")
+ (RVVM2x4SI "w") (RVVM1x4SI "w") (RVVMF2x4SI "w")
+ (RVVM2x3SI "w") (RVVM1x3SI "w") (RVVMF2x3SI "w")
+ (RVVM4x2SI "w") (RVVM2x2SI "w") (RVVM1x2SI "w") (RVVMF2x2SI "w")
+
+ (RVVM1x8SF "w") (RVVMF2x8SF "w")
+ (RVVM1x7SF "w") (RVVMF2x7SF "w")
+ (RVVM1x6SF "w") (RVVMF2x6SF "w")
+ (RVVM1x5SF "w") (RVVMF2x5SF "w")
+ (RVVM2x4SF "w") (RVVM1x4SF "w") (RVVMF2x4SF "w")
+ (RVVM2x3SF "w") (RVVM1x3SF "w") (RVVMF2x3SF "w")
+ (RVVM4x2SF "w") (RVVM2x2SF "w") (RVVM1x2SF "w") (RVVMF2x2SF "w")
+
+ ;; VLS modes.
+ (V1QI "b") (V2QI "b") (V4QI "b") (V8QI "b") (V16QI "b") (V32QI "b") (V64QI
"b") (V128QI "b") (V256QI "b") (V512QI "b")
+ (V1024QI "b") (V2048QI "b") (V4096QI "b")
+ (V1HI "h") (V2HI "h") (V4HI "h") (V8HI "h") (V16HI "h") (V32HI "h") (V64HI
"h") (V128HI "h") (V256HI "h")
+ (V512HI "h") (V1024HI "h") (V2048HI "h")
+ (V1SI "w") (V2SI "w") (V4SI "w") (V8SI "w") (V16SI "w") (V32SI "w") (V64SI
"w") (V128SI "w") (V256SI "w")
+ (V512SI "w") (V1024SI "w")
+ (V1HF "h") (V2HF "h") (V4HF "h") (V8HF "h") (V16HF "h") (V32HF "h") (V64HF
"h") (V128HF "h") (V256HF "h")
+ (V512HF "h") (V1024HF "h") (V2048HF "h")
+ (V1SF "w") (V2SF "w") (V4SF "w") (V8SF "w") (V16SF "w") (V32SF "w") (V64SF
"w") (V128SF "w") (V256SF "w")
+ (V512SF "w") (V1024SF "w")
+])
+
(define_split
[(set (match_operand:V_VB_VLS_VT 0 "reg_or_mem_operand")
(match_operand:V_VB_VLS_VT 1 "reg_or_mem_operand"))]
@@ -350,3 +446,40 @@ (define_insn "*pred_th_extract<mode>"
"vext.x.v\t%0,%1,%2"
[(set_attr "type" "vimovvx")
(set_attr "mode" "<MODE>")])
+
+(define_insn "@pred_th_unit_seg_load<vlmem_op_attr><mode>"
+ [(set (match_operand:VT 0 "register_operand" "=vr, vr, vd")
+ (if_then_else:VT
+ (unspec:<VM>
+ [(match_operand:<VM> 1 "vector_mask_operand" "vmWc1, Wc1, vm")
+ (match_operand 4 "vector_length_operand" " rK, rK, rK")
+ (match_operand 5 "const_int_operand" " i, i, i")
+ (match_operand 6 "const_int_operand" " i, i, i")
+ (match_operand 7 "const_int_operand" " i, i, i")
+ (reg:SI VL_REGNUM)
+ (reg:SI VTYPE_REGNUM)] UNSPEC_TH_VLSEGMEM_OP)
+ (unspec:VT
+ [(match_operand 3 "pmode_reg_or_0_operand" " rJ, rJ, rJ")
+ (mem:BLK (scratch))] UNSPEC_TH_VLSEGMEM_OP)
+ (match_operand:VT 2 "vector_merge_operand" " 0, vu,
vu")))]
+ "TARGET_XTHEADVECTOR"
+ "vlseg<nf><vlmem_op_attr>.v\t%0,(%z3)%p1"
+ [(set_attr "type" "vlsegde")
+ (set_attr "mode" "<MODE>")])
+
+(define_insn "@pred_th_unit_seg_store<vlmem_op_attr><mode>"
+ [(set (mem:BLK (scratch))
+ (unspec:BLK
+ [(unspec:<VM>
+ [(match_operand:<VM> 0 "vector_mask_operand" "vmWc1")
+ (match_operand 3 "vector_length_operand" " rK")
+ (match_operand 4 "const_int_operand" " i")
+ (reg:SI VL_REGNUM)
+ (reg:SI VTYPE_REGNUM)] UNSPEC_TH_VSSEGMEM_OP)
+ (match_operand 1 "pmode_reg_or_0_operand" " rJ")
+ (match_operand:VT 2 "register_operand" " vr")
+ (mem:BLK (scratch))] UNSPEC_TH_VSSEGMEM_OP))]
+ "TARGET_XTHEADVECTOR"
+ "vsseg<nf><vlmem_op_attr>.v\t%2,(%z1)%p0"
+ [(set_attr "type" "vssegte")
+ (set_attr "mode" "<MODE>")])
diff --git a/gcc/config/riscv/thead.cc b/gcc/config/riscv/thead.cc
index de23e410d4c..628f50d012c 100644
--- a/gcc/config/riscv/thead.cc
+++ b/gcc/config/riscv/thead.cc
@@ -1059,12 +1059,17 @@ th_asm_output_opcode (FILE *asm_out_file, const char *p)
get_attr_type (current_output_insn) == TYPE_VSSEGTE
? fputs ("th.vsseg", asm_out_file)
: fputs ("th.vlseg", asm_out_file);
- asm_fprintf (asm_out_file, "%c", p[5]);
- fputs ("e", asm_out_file);
- if (strstr (p, "e8"))
- return p+8;
+ if (strstr (p, "b") || strstr (p, "h") || strstr (p, "w"))
+ return p+5;
else
- return p+9;
+ {
+ asm_fprintf (asm_out_file, "%c", p[5]);
+ fputs ("e", asm_out_file);
+ if (strstr (p, "e8"))
+ return p+8;
+ else
+ return p+9;
+ }
}
if (get_attr_type (current_output_insn) == TYPE_VLSEGDS ||
@@ -1073,36 +1078,51 @@ th_asm_output_opcode (FILE *asm_out_file, const char *p)
get_attr_type (current_output_insn) == TYPE_VSSEGTS
? fputs ("th.vssseg", asm_out_file)
: fputs ("th.vlsseg", asm_out_file);
- asm_fprintf (asm_out_file, "%c", p[6]);
- fputs ("e", asm_out_file);
- if (strstr (p, "e8"))
- return p+9;
+ if (strstr (p, "b") || strstr (p, "h") || strstr (p, "w"))
+ return p+6;
else
- return p+10;
+ {
+ asm_fprintf (asm_out_file, "%c", p[6]);
+ fputs ("e", asm_out_file);
+ if (strstr (p, "e8"))
+ return p+9;
+ else
+ return p+10;
+ }
}
if (get_attr_type (current_output_insn) == TYPE_VLSEGDUX ||
get_attr_type (current_output_insn) == TYPE_VLSEGDOX)
{
fputs ("th.vlxseg", asm_out_file);
- asm_fprintf (asm_out_file, "%c", p[7]);
- fputs ("e", asm_out_file);
- if (strstr (p, "ei8"))
- return p+11;
+ if (strstr (p, "b") || strstr (p, "h") || strstr (p, "w"))
+ return p+6;
else
- return p+12;
+ {
+ asm_fprintf (asm_out_file, "%c", p[7]);
+ fputs ("e", asm_out_file);
+ if (strstr (p, "ei8"))
+ return p+11;
+ else
+ return p+12;
+ }
}
if (get_attr_type (current_output_insn) == TYPE_VSSEGTUX ||
get_attr_type (current_output_insn) == TYPE_VSSEGTOX)
{
fputs ("th.vsxseg", asm_out_file);
- asm_fprintf (asm_out_file, "%c", p[7]);
- fputs ("e", asm_out_file);
- if (strstr (p, "ei8"))
- return p+11;
+ if (strstr (p, "b") || strstr (p, "h") || strstr (p, "w"))
+ return p+6;
else
- return p+12;
+ {
+ asm_fprintf (asm_out_file, "%c", p[7]);
+ fputs ("e", asm_out_file);
+ if (strstr (p, "ei8"))
+ return p+11;
+ else
+ return p+12;
+ }
}
if (get_attr_type (current_output_insn) == TYPE_VNSHIFT)
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlseg-vsseg.c
b/gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlseg-vsseg.c
new file mode 100644
index 00000000000..1911824bbfc
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlseg-vsseg.c
@@ -0,0 +1,118 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcxtheadvector -mabi=ilp32d -O3" { target rv32 } }
*/
+/* { dg-final { check-function-bodies "**" "" } } */
+#include "riscv_th_vector.h"
+
+/*
+** f1:
+** li\s+[a-x0-9]+,4
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vlseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vsseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** ret
+*/
+void f1 (void * in, void *out)
+{
+ vint16m1x2_t v = __riscv_th_vlseg2h_v_i16m1x2 (in, 4);
+ vint16m1x2_t v2 = __riscv_th_vlseg2h_v_i16m1x2_tu (v, in, 4);
+ vint16m1_t v2_0 = __riscv_vget_i16m1 (v2, 0);
+ vint16m1_t v2_1 = __riscv_vget_i16m1 (v2, 1);
+ vint16m1_t v3_0 = __riscv_vadd_vv_i16m1 (v2_0, v2_0, 4);
+ vint16m1_t v3_1 = __riscv_vadd_vv_i16m1 (v2_1, v2_1, 4);
+ vint16m1_t v4_0 = __riscv_vadd_vv_i16m1 (v3_0, v2_0, 4);
+ vint16m1_t v4_1 = __riscv_vadd_vv_i16m1 (v3_1, v2_1, 4);
+ vint16m1x2_t v4 = __riscv_vset (v4, 0, v4_0);
+ v4 = __riscv_vset (v4, 1, v4_1);
+ __riscv_th_vsseg2h_v_i16m1x2 (out, v4, 4);
+}
+
+/*
+** f2:
+** li\s+[a-x0-9]+,4
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2h\.v\s+v[0-9]+,\([a-x0-9]+\),v0\.t
+** th\.vlseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vsseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** ret
+*/
+void f2 (void * in, void *out, vbool16_t mask)
+{
+ vint16m1x2_t v = __riscv_th_vlseg2h_v_i16m1x2 (in, 4);
+ vint16m1x2_t v2 = __riscv_th_vlseg2h_v_i16m1x2_m (mask, in, 4);
+ vint16m1_t v_0 = __riscv_vget_i16m1(v, 0);
+ vint16m1_t v_1 = __riscv_vget_i16m1(v, 1);
+ vint16m1_t v2_0 = __riscv_vget_i16m1 (v2, 0);
+ vint16m1_t v2_1 = __riscv_vget_i16m1 (v2, 1);
+ vint16m1_t v3_0 = __riscv_vadd_vv_i16m1 (v_0, v2_0, 4);
+ vint16m1_t v3_1 = __riscv_vadd_vv_i16m1 (v_1, v2_1, 4);
+ vint16m1_t v4_0 = __riscv_vadd_vv_i16m1 (v3_0, v2_0, 4);
+ vint16m1_t v4_1 = __riscv_vadd_vv_i16m1 (v3_1, v2_1, 4);
+ vint16m1x2_t v4 = __riscv_vset (v4, 0, v4_0);
+ v4 = __riscv_vset (v4, 1, v4_1);
+ __riscv_th_vsseg2h_v_i16m1x2 (out, v4, 4);
+}
+
+/*
+** f3:
+** li\s+[a-x0-9]+,4
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vsetvli\s+[a-x0-9]+,zero,e16,m1
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2h\.v\s+v[0-9]+,\([a-x0-9]+\),v0\.t
+** th\.vsetvli\s+[a-x0-9]+,zero,e16,m1
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vsseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** ret
+*/
+void f3 (void * in, void *out, vbool16_t mask)
+{
+ vint16m1x2_t v = __riscv_th_vlseg2h_v_i16m1x2 (in, 4);
+ vint16m1x2_t v2 = __riscv_th_vlseg2h_v_i16m1x2_tumu (mask, v, in, 4);
+ vint16m1_t v2_0 = __riscv_vget_i16m1 (v2, 0);
+ vint16m1_t v2_1 = __riscv_vget_i16m1 (v2, 1);
+ vint16m1_t v3_0 = __riscv_vadd_vv_i16m1_tumu (mask, v3_0, v2_0, v2_0, 4);
+ vint16m1_t v3_1 = __riscv_vadd_vv_i16m1_tumu (mask, v3_1, v2_1, v2_1, 4);
+ vint16m1_t v4_0 = __riscv_vadd_vv_i16m1_tumu (mask, v4_0, v3_0, v2_0, 4);
+ vint16m1_t v4_1 = __riscv_vadd_vv_i16m1_tumu (mask, v4_1, v3_1, v2_1, 4);
+ vint16m1x2_t v4 = __riscv_vset (v4, 0, v4_0);
+ v4 = __riscv_vset (v4, 1, v4_1);
+ __riscv_th_vsseg2h_v_i16m1x2 (out, v4, 4);
+}
diff --git a/gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlsegu-vsseg.c
b/gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlsegu-vsseg.c
new file mode 100644
index 00000000000..c3791f7872f
--- /dev/null
+++ b/gcc/testsuite/gcc.target/riscv/rvv/xtheadvector/vlsegu-vsseg.c
@@ -0,0 +1,116 @@
+/* { dg-do compile } */
+/* { dg-options "-march=rv32gcxtheadvector -mabi=ilp32d -O3" { target rv32 } }
+ */
+/* { dg-final { check-function-bodies "**" "" } } */
+#include "riscv_th_vector.h"
+
+/*
+** f1:
+** li\s+[a-x0-9]+,4
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2hu\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vlseg2hu\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vsseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** ret
+*/
+void f1(void *in, void *out) {
+ vuint16m1x2_t v = __riscv_th_vlseg2hu_v_u16m1x2(in, 4);
+ vuint16m1x2_t v2 = __riscv_th_vlseg2hu_v_u16m1x2_tu(v, in, 4);
+ vuint16m1_t v2_0 = __riscv_vget_u16m1(v2, 0);
+ vuint16m1_t v2_1 = __riscv_vget_u16m1(v2, 1);
+ vuint16m1_t v3_0 = __riscv_vadd_vv_u16m1(v2_0, v2_0, 4);
+ vuint16m1_t v3_1 = __riscv_vadd_vv_u16m1(v2_1, v2_1, 4);
+ vuint16m1_t v4_0 = __riscv_vadd_vv_u16m1(v3_0, v2_0, 4);
+ vuint16m1_t v4_1 = __riscv_vadd_vv_u16m1(v3_1, v2_1, 4);
+ vuint16m1x2_t v4 = __riscv_vset(v4, 0, v4_0);
+ v4 = __riscv_vset(v4, 1, v4_1);
+ __riscv_th_vsseg2h_v_u16m1x2(out, v4, 4);
+}
+
+/*
+** f2:
+** li\s+[a-x0-9]+,4
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2hu\.v\s+v[0-9]+,\([a-x0-9]+\),v0\.t
+** th\.vlseg2hu\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vsseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** ret
+*/
+void f2(void *in, void *out, vbool16_t mask) {
+ vuint16m1x2_t v = __riscv_th_vlseg2hu_v_u16m1x2(in, 4);
+ vuint16m1x2_t v2 = __riscv_th_vlseg2hu_v_u16m1x2_m(mask, in, 4);
+ vuint16m1_t v_0 = __riscv_vget_u16m1(v, 0);
+ vuint16m1_t v_1 = __riscv_vget_u16m1(v, 1);
+ vuint16m1_t v2_0 = __riscv_vget_u16m1(v2, 0);
+ vuint16m1_t v2_1 = __riscv_vget_u16m1(v2, 1);
+ vuint16m1_t v3_0 = __riscv_vadd_vv_u16m1(v_0, v2_0, 4);
+ vuint16m1_t v3_1 = __riscv_vadd_vv_u16m1(v_1, v2_1, 4);
+ vuint16m1_t v4_0 = __riscv_vadd_vv_u16m1(v3_0, v2_0, 4);
+ vuint16m1_t v4_1 = __riscv_vadd_vv_u16m1(v3_1, v2_1, 4);
+ vuint16m1x2_t v4 = __riscv_vset(v4, 0, v4_0);
+ v4 = __riscv_vset(v4, 1, v4_1);
+ __riscv_th_vsseg2h_v_u16m1x2(out, v4, 4);
+}
+
+/*
+** f3:
+** li\s+[a-x0-9]+,4
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2hu\.v\s+v[0-9]+,\([a-x0-9]+\)
+** th\.vsetvli\s+[a-x0-9]+,zero,e16,m1
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vlseg2hu\.v\s+v[0-9]+,\([a-x0-9]+\),v0\.t
+** th\.vsetvli\s+[a-x0-9]+,zero,e16,m1
+** th\.vmv\.v\.i\s+v[0-9]+,0
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vadd\.vv\s+v[0-9]+,v[0-9]+,v[0-9]+,v0\.t
+** th\.vsetvli\s+zero,zero,e16,m1
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vmv\.v\.v\s+v[0-9]+,v[0-9]+
+** th\.vsetvli\s+zero,[a-x0-9]+,e16,m1
+** th\.vsseg2h\.v\s+v[0-9]+,\([a-x0-9]+\)
+** ret
+*/
+void f3(void *in, void *out, vbool16_t mask) {
+ vuint16m1x2_t v = __riscv_th_vlseg2hu_v_u16m1x2(in, 4);
+ vuint16m1x2_t v2 = __riscv_th_vlseg2hu_v_u16m1x2_tumu(mask, v, in, 4);
+ vuint16m1_t v2_0 = __riscv_vget_u16m1(v2, 0);
+ vuint16m1_t v2_1 = __riscv_vget_u16m1(v2, 1);
+ vuint16m1_t v3_0 = __riscv_vadd_vv_u16m1_tumu(mask, v3_0, v2_0, v2_0, 4);
+ vuint16m1_t v3_1 = __riscv_vadd_vv_u16m1_tumu(mask, v3_1, v2_1, v2_1, 4);
+ vuint16m1_t v4_0 = __riscv_vadd_vv_u16m1_tumu(mask, v4_0, v3_0, v2_0, 4);
+ vuint16m1_t v4_1 = __riscv_vadd_vv_u16m1_tumu(mask, v4_1, v3_1, v2_1, 4);
+ vuint16m1x2_t v4 = __riscv_vset(v4, 0, v4_0);
+ v4 = __riscv_vset(v4, 1, v4_1);
+ __riscv_th_vsseg2h_v_u16m1x2(out, v4, 4);
+}
--
2.47.1
