public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: "Li Xu" <xuli1@eswincomputing.com>
To: kito.cheng <kito.cheng@gmail.com>,  juzhe.zhong <juzhe.zhong@rivai.ai>
Cc: gcc-patches <gcc-patches@gcc.gnu.org>
Subject: Re: Re: [PATCH] RISC-V: Add vcreate intrinsics for RVV tuple types
Date: Tue, 12 Sep 2023 11:21:27 +0800	[thread overview]
Message-ID: <20230912112127040396153@eswincomputing.com> (raw)
In-Reply-To: <CA+yXCZBeaOJJ+KOHAoL9Wbk61M84+WSOjD29WrEJeftzn1YLTA@mail.gmail.com>

Committed, thanks juzhe and kito.
--------------
Li Xu
>LGTM, thanks for implement this :)
>
>On Tue, Sep 12, 2023 at 10:16 AM juzhe.zhong@rivai.ai
><juzhe.zhong@rivai.ai> wrote:
>>
>> Thanks for support it.
>> LGTM from my side.
>> Wait for kito's more comments.
>>
>>
>>
>> juzhe.zhong@rivai.ai
>>
>> From: Li Xu
>> Date: 2023-09-12 10:08
>> To: gcc-patches
>> CC: kito.cheng; palmer; juzhe.zhong; pan2.li; gaofei; wangfeng; xuli
>> Subject: [PATCH] RISC-V: Add vcreate intrinsics for RVV tuple types
>> From: xuli <xuli1@eswincomputing.com>
>>
>> gcc/ChangeLog:
>>
>>         * config/riscv/riscv-vector-builtins-bases.cc (class vcreate):
>>         (BASE): New class.
>>         * config/riscv/riscv-vector-builtins-bases.h: Ditto.
>>         * config/riscv/riscv-vector-builtins-functions.def (vcreate): Add vcreate support.
>>         * config/riscv/riscv-vector-builtins-shapes.cc (struct vcreate_def): Ditto.
>>         (SHAPE): Ditto.
>>         * config/riscv/riscv-vector-builtins-shapes.h: Ditto.
>>         * config/riscv/riscv-vector-builtins.cc: Add args type.
>>
>> gcc/testsuite/ChangeLog:
>>
>>         * gcc.target/riscv/rvv/base/tuple_create.c: New test.
>> ---
>> .../riscv/riscv-vector-builtins-bases.cc      |  40 ++++++
>> .../riscv/riscv-vector-builtins-bases.h       |   1 +
>> .../riscv/riscv-vector-builtins-functions.def |   1 +
>> .../riscv/riscv-vector-builtins-shapes.cc     |  50 +++++++
>> .../riscv/riscv-vector-builtins-shapes.h      |   1 +
>> gcc/config/riscv/riscv-vector-builtins.cc     |  12 ++
>> .../gcc.target/riscv/rvv/base/tuple_create.c  | 123 ++++++++++++++++++
>> 7 files changed, 228 insertions(+)
>> create mode 100644 gcc/testsuite/gcc.target/riscv/rvv/base/tuple_create.c
>>
>> diff --git a/gcc/config/riscv/riscv-vector-builtins-bases.cc b/gcc/config/riscv/riscv-vector-builtins-bases.cc
>> index 8e679f72392..be3df2c1ea2 100644
>> --- a/gcc/config/riscv/riscv-vector-builtins-bases.cc
>> +++ b/gcc/config/riscv/riscv-vector-builtins-bases.cc
>> @@ -1824,6 +1824,44 @@ public:
>>    }
>> };
>> +class vcreate : public function_base
>> +{
>> +public:
>> +  gimple *fold (gimple_folder &f) const override
>> +  {
>> +    unsigned int nargs = gimple_call_num_args (f.call);
>> +    tree lhs_type = TREE_TYPE (f.lhs);
>> +
>> +    /* Replace the call with a clobber of the result (to prevent it from
>> +       becoming upwards exposed) followed by stores into each individual
>> +       vector of tuple.
>> +
>> +       The fold routines expect the replacement statement to have the
>> +       same lhs as the original call, so return the clobber statement
>> +       rather than the final vector store.  */
>> +    gassign *clobber = gimple_build_assign (f.lhs, build_clobber (lhs_type));
>> +
>> +    for (unsigned int i = nargs; i-- > 0; )
>> +      {
>> + tree rhs_vector = gimple_call_arg (f.call, i);
>> + tree field = tuple_type_field (TREE_TYPE (f.lhs));
>> + tree lhs_array = build3 (COMPONENT_REF, TREE_TYPE (field),
>> + unshare_expr (f.lhs), field, NULL_TREE);
>> + tree lhs_vector = build4 (ARRAY_REF, TREE_TYPE (rhs_vector),
>> +   lhs_array, size_int (i),
>> +   NULL_TREE, NULL_TREE);
>> + gassign *assign = gimple_build_assign (lhs_vector, rhs_vector);
>> + gsi_insert_after (f.gsi, assign, GSI_SAME_STMT);
>> +      }
>> +    return clobber;
>> +  }
>> +
>> +  rtx expand (function_expander &e) const override
>> +  {
>> +    return NULL_RTX;
>> +  }
>> +};
>> +
>> class read_vl : public function_base
>> {
>> public:
>> @@ -2285,6 +2323,7 @@ static CONSTEXPR const vlmul_ext vlmul_ext_obj;
>> static CONSTEXPR const vlmul_trunc vlmul_trunc_obj;
>> static CONSTEXPR const vset vset_obj;
>> static CONSTEXPR const vget vget_obj;
>> +static CONSTEXPR const vcreate vcreate_obj;
>> static CONSTEXPR const read_vl read_vl_obj;
>> static CONSTEXPR const vleff vleff_obj;
>> static CONSTEXPR const vlenb vlenb_obj;
>> @@ -2546,6 +2585,7 @@ BASE (vlmul_ext)
>> BASE (vlmul_trunc)
>> BASE (vset)
>> BASE (vget)
>> +BASE (vcreate)
>> BASE (read_vl)
>> BASE (vleff)
>> BASE (vlenb)
>> diff --git a/gcc/config/riscv/riscv-vector-builtins-bases.h b/gcc/config/riscv/riscv-vector-builtins-bases.h
>> index 69d4562091f..131041ea66f 100644
>> --- a/gcc/config/riscv/riscv-vector-builtins-bases.h
>> +++ b/gcc/config/riscv/riscv-vector-builtins-bases.h
>> @@ -267,6 +267,7 @@ extern const function_base *const vlmul_ext;
>> extern const function_base *const vlmul_trunc;
>> extern const function_base *const vset;
>> extern const function_base *const vget;
>> +extern const function_base *const vcreate;
>> extern const function_base *const read_vl;
>> extern const function_base *const vleff;
>> extern const function_base *const vlenb;
>> diff --git a/gcc/config/riscv/riscv-vector-builtins-functions.def b/gcc/config/riscv/riscv-vector-builtins-functions.def
>> index 3ce06dc60b7..18ed2c2b8f6 100644
>> --- a/gcc/config/riscv/riscv-vector-builtins-functions.def
>> +++ b/gcc/config/riscv/riscv-vector-builtins-functions.def
>> @@ -621,6 +621,7 @@ DEF_RVV_FUNCTION (vget, vget, none_preds, all_v_vget_lmul4_x2_ops)
>> // Tuple types
>> DEF_RVV_FUNCTION (vset, vset, none_preds, all_v_vset_tuple_ops)
>> DEF_RVV_FUNCTION (vget, vget, none_preds, all_v_vget_tuple_ops)
>> +DEF_RVV_FUNCTION (vcreate, vcreate, none_preds, all_v_vcreate_tuple_ops)
>> DEF_RVV_FUNCTION (vlseg, seg_loadstore, full_preds, tuple_v_scalar_const_ptr_ops)
>> DEF_RVV_FUNCTION (vsseg, seg_loadstore, none_m_preds, tuple_v_scalar_ptr_ops)
>> DEF_RVV_FUNCTION (vlsseg, seg_loadstore, full_preds, tuple_v_scalar_const_ptr_ptrdiff_ops)
>> diff --git a/gcc/config/riscv/riscv-vector-builtins-shapes.cc b/gcc/config/riscv/riscv-vector-builtins-shapes.cc
>> index f8fdec863e6..0bda934ae16 100644
>> --- a/gcc/config/riscv/riscv-vector-builtins-shapes.cc
>> +++ b/gcc/config/riscv/riscv-vector-builtins-shapes.cc
>> @@ -706,6 +706,55 @@ struct vget_def : public misc_def
>>    }
>> };
>> +/* vcreate_def class.  */
>> +struct vcreate_def : public build_base
>> +{
>> +  void build (function_builder &b,
>> +       const function_group_info &group) const override
>> +  {
>> +    for (unsigned int vec_type_idx = 0;
>> + group.ops_infos.types[vec_type_idx].index != NUM_VECTOR_TYPES;
>> + ++vec_type_idx)
>> +      {
>> +   auto_vec<tree, 8> argument_types;
>> +   function_instance function_instance (group.base_name, *group.base,
>> +       *group.shape,
>> +       group.ops_infos.types[vec_type_idx],
>> +       group.preds[0], &group.ops_infos);
>> +
>> +   tree return_type = group.ops_infos.ret.get_tree_type (
>> +     group.ops_infos.types[vec_type_idx].index);
>> +
>> +   if (!return_type)
>> +     continue;
>> +
>> +   machine_mode mode = TYPE_MODE (return_type);
>> +   unsigned int nf = get_nf (mode);
>> +
>> +   for (unsigned int i = 0; i < nf; i++)
>> +     argument_types.quick_push (
>> +       function_instance.op_info->args[0].get_tree_type (
>> +         function_instance.type.index));
>> +
>> +   b.add_unique_function (function_instance, (*group.shape), return_type,
>> +     argument_types);
>> +     }
>> +  }
>> +
>> +  char *get_name (function_builder &b, const function_instance &instance,
>> +   bool overloaded_p) const override
>> +  {
>> +    if (overloaded_p)
>> +      return nullptr;
>> +    b.append_base_name (instance.base_name);
>> +    b.append_name (operand_suffixes[instance.op_info->op]);
>> +    vector_type_index ret_type_idx
>> +      = instance.op_info->ret.get_function_type_index (instance.type.index);
>> +    b.append_name (type_suffixes[ret_type_idx].vector);
>> +    return b.finish_name ();
>> +  }
>> +};
>> +
>> /* read_vl_def class.  */
>> struct read_vl_def : public function_shape
>> {
>> @@ -942,6 +991,7 @@ SHAPE(vundefined, vundefined)
>> SHAPE(misc, misc)
>> SHAPE(vset, vset)
>> SHAPE(vget, vget)
>> +SHAPE(vcreate, vcreate)
>> SHAPE(read_vl, read_vl)
>> SHAPE(fault_load, fault_load)
>> SHAPE(vlenb, vlenb)
>> diff --git a/gcc/config/riscv/riscv-vector-builtins-shapes.h b/gcc/config/riscv/riscv-vector-builtins-shapes.h
>> index 92eb8bc9d71..df9884bb572 100644
>> --- a/gcc/config/riscv/riscv-vector-builtins-shapes.h
>> +++ b/gcc/config/riscv/riscv-vector-builtins-shapes.h
>> @@ -45,6 +45,7 @@ extern const function_shape *const vundefined;
>> extern const function_shape *const misc;
>> extern const function_shape *const vset;
>> extern const function_shape *const vget;
>> +extern const function_shape *const vcreate;
>> extern const function_shape *const read_vl;
>> extern const function_shape *const fault_load;
>> extern const function_shape *const vlenb;
>> diff --git a/gcc/config/riscv/riscv-vector-builtins.cc b/gcc/config/riscv/riscv-vector-builtins.cc
>> index 6d99f970ead..5318651138a 100644
>> --- a/gcc/config/riscv/riscv-vector-builtins.cc
>> +++ b/gcc/config/riscv/riscv-vector-builtins.cc
>> @@ -873,6 +873,10 @@ static CONSTEXPR const rvv_arg_type_info tuple_vset_args[]
>>    = {rvv_arg_type_info (RVV_BASE_vector), rvv_arg_type_info (RVV_BASE_size),
>>       rvv_arg_type_info (RVV_BASE_tuple_subpart), rvv_arg_type_info_end};
>> +/* A list of args for vector_type func (vector_type) function.  */
>> +static CONSTEXPR const rvv_arg_type_info tuple_vcreate_args[]
>> +  = {rvv_arg_type_info (RVV_BASE_tuple_subpart), rvv_arg_type_info_end};
>> +
>> /* A list of none preds that will be registered for intrinsic functions.  */
>> static CONSTEXPR const predication_type_index none_preds[]
>>    = {PRED_TYPE_none, NUM_PRED_TYPES};
>> @@ -2497,6 +2501,14 @@ static CONSTEXPR const rvv_op_info tuple_v_scalar_const_ptr_size_ptr_ops
>>       rvv_arg_type_info (RVV_BASE_vector), /* Return type */
>>       scalar_const_ptr_size_ptr_args /* Args */};
>> +/* A static operand information for vector_type func (vector_type)
>> + * function registration. */
>> +static CONSTEXPR const rvv_op_info all_v_vcreate_tuple_ops
>> +  = {tuple_ops,   /* Types */
>> +     OP_TYPE_v,   /* Suffix */
>> +     rvv_arg_type_info (RVV_BASE_vector), /* Return type */
>> +     tuple_vcreate_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/testsuite/gcc.target/riscv/rvv/base/tuple_create.c b/gcc/testsuite/gcc.target/riscv/rvv/base/tuple_create.c
>> new file mode 100644
>> index 00000000000..b252b2f3cd4
>> --- /dev/null
>> +++ b/gcc/testsuite/gcc.target/riscv/rvv/base/tuple_create.c
>> @@ -0,0 +1,123 @@
>> +/* { dg-do compile } */
>> +/* { dg-options "-march=rv64gcv_zvfh -mabi=lp64 -O3 -Wno-psabi" } */
>> +
>> +#include "riscv_vector.h"
>> +
>> +vfloat16mf4x2_t test_vcreate_v_f16mf4x2(vfloat16mf4_t v0, vfloat16mf4_t v1) {
>> +  return __riscv_vcreate_v_f16mf4x2(v0, v1);
>> +}
>> +
>> +vfloat16mf4x3_t test_vcreate_v_f16mf4x3(vfloat16mf4_t v0, vfloat16mf4_t v1, vfloat16mf4_t v2) {
>> +  return __riscv_vcreate_v_f16mf4x3(v0, v1, v2);
>> +}
>> +
>> +vfloat16mf4x4_t test_vcreate_v_f16mf4x4(vfloat16mf4_t v0, vfloat16mf4_t v1, vfloat16mf4_t v2, vfloat16mf4_t v3) {
>> +  return __riscv_vcreate_v_f16mf4x4(v0, v1, v2, v3);
>> +}
>> +
>> +vfloat16mf4x5_t test_vcreate_v_f16mf4x5(vfloat16mf4_t v0, vfloat16mf4_t v1, vfloat16mf4_t v2, vfloat16mf4_t v3, vfloat16mf4_t v4) {
>> +  return __riscv_vcreate_v_f16mf4x5(v0, v1, v2, v3, v4);
>> +}
>> +
>> +vfloat16mf4x6_t test_vcreate_v_f16mf4x6(vfloat16mf4_t v0, vfloat16mf4_t v1, vfloat16mf4_t v2, vfloat16mf4_t v3, vfloat16mf4_t v4, vfloat16mf4_t v5) {
>> +  return __riscv_vcreate_v_f16mf4x6(v0, v1, v2, v3, v4, v5);
>> +}
>> +
>> +vfloat16mf4x7_t test_vcreate_v_f16mf4x7(vfloat16mf4_t v0, vfloat16mf4_t v1, vfloat16mf4_t v2, vfloat16mf4_t v3, vfloat16mf4_t v4, vfloat16mf4_t v5, vfloat16mf4_t v6) {
>> +  return __riscv_vcreate_v_f16mf4x7(v0, v1, v2, v3, v4, v5, v6);
>> +}
>> +
>> +vfloat16mf4x8_t test_vcreate_v_f16mf4x8(vfloat16mf4_t v0, vfloat16mf4_t v1, vfloat16mf4_t v2, vfloat16mf4_t v3, vfloat16mf4_t v4, vfloat16mf4_t v5, vfloat16mf4_t v6, vfloat16mf4_t v7) {
>> +  return __riscv_vcreate_v_f16mf4x8(v0, v1, v2, v3, v4, v5, v6, v7);
>> +}
>> +
>> +vfloat32m1x2_t test_vcreate_v_f32m1x2(vfloat32m1_t v0, vfloat32m1_t v1) {
>> +  return __riscv_vcreate_v_f32m1x2(v0, v1);
>> +}
>> +
>> +vfloat32m1x3_t test_vcreate_v_f32m1x3(vfloat32m1_t v0, vfloat32m1_t v1, vfloat32m1_t v2) {
>> +  return __riscv_vcreate_v_f32m1x3(v0, v1, v2);
>> +}
>> +
>> +vfloat32m1x4_t test_vcreate_v_f32m1x4(vfloat32m1_t v0, vfloat32m1_t v1, vfloat32m1_t v2, vfloat32m1_t v3) {
>> +  return __riscv_vcreate_v_f32m1x4(v0, v1, v2, v3);
>> +}
>> +
>> +vfloat32m1x5_t test_vcreate_v_f32m1x5(vfloat32m1_t v0, vfloat32m1_t v1, vfloat32m1_t v2, vfloat32m1_t v3, vfloat32m1_t v4) {
>> +  return __riscv_vcreate_v_f32m1x5(v0, v1, v2, v3, v4);
>> +}
>> +
>> +vfloat32m1x6_t test_vcreate_v_f32m1x6(vfloat32m1_t v0, vfloat32m1_t v1, vfloat32m1_t v2, vfloat32m1_t v3, vfloat32m1_t v4, vfloat32m1_t v5) {
>> +  return __riscv_vcreate_v_f32m1x6(v0, v1, v2, v3, v4, v5);
>> +}
>> +
>> +vfloat32m1x7_t test_vcreate_v_f32m1x7(vfloat32m1_t v0, vfloat32m1_t v1, vfloat32m1_t v2, vfloat32m1_t v3, vfloat32m1_t v4, vfloat32m1_t v5, vfloat32m1_t v6) {
>> +  return __riscv_vcreate_v_f32m1x7(v0, v1, v2, v3, v4, v5, v6);
>> +}
>> +
>> +vfloat32m1x8_t test_vcreate_v_f32m1x8(vfloat32m1_t v0, vfloat32m1_t v1, vfloat32m1_t v2, vfloat32m1_t v3, vfloat32m1_t v4, vfloat32m1_t v5, vfloat32m1_t v6, vfloat32m1_t v7) {
>> +  return __riscv_vcreate_v_f32m1x8(v0, v1, v2, v3, v4, v5, v6, v7);
>> +}
>> +
>> +vfloat64m2x2_t test_vcreate_v_f64m2x2(vfloat64m2_t v0, vfloat64m2_t v1) {
>> +  return __riscv_vcreate_v_f64m2x2(v0, v1);
>> +}
>> +
>> +vfloat64m2x3_t test_vcreate_v_f64m2x3(vfloat64m2_t v0, vfloat64m2_t v1, vfloat64m2_t v2) {
>> +  return __riscv_vcreate_v_f64m2x3(v0, v1, v2);
>> +}
>> +
>> +vfloat64m2x4_t test_vcreate_v_f64m2x4(vfloat64m2_t v0, vfloat64m2_t v1, vfloat64m2_t v2, vfloat64m2_t v3) {
>> +  return __riscv_vcreate_v_f64m2x4(v0, v1, v2, v3);
>> +}
>> +
>> +vfloat64m4x2_t test_vcreate_v_f64m4x2(vfloat64m4_t v0, vfloat64m4_t v1) {
>> +  return __riscv_vcreate_v_f64m4x2(v0, v1);
>> +}
>> +
>> +vint8m2x2_t test_vcreate_v_i8m2x2(vint8m2_t v0, vint8m2_t v1) {
>> +  return __riscv_vcreate_v_i8m2x2(v0, v1);
>> +}
>> +
>> +vint8m2x3_t test_vcreate_v_i8m2x3(vint8m2_t v0, vint8m2_t v1, vint8m2_t v2) {
>> +  return __riscv_vcreate_v_i8m2x3(v0, v1, v2);
>> +}
>> +
>> +vint8m2x4_t test_vcreate_v_i8m2x4(vint8m2_t v0, vint8m2_t v1, vint8m2_t v2, vint8m2_t v3) {
>> +  return __riscv_vcreate_v_i8m2x4(v0, v1, v2, v3);
>> +}
>> +
>> +vint8m4x2_t test_vcreate_v_i8m4x2(vint8m4_t v0, vint8m4_t v1) {
>> +  return __riscv_vcreate_v_i8m4x2(v0, v1);
>> +}
>> +
>> +vint16m4x2_t test_vcreate_v_i16m4x2(vint16m4_t v0, vint16m4_t v1) {
>> +  return __riscv_vcreate_v_i16m4x2(v0, v1);
>> +}
>> +
>> +vint32m4x2_t test_vcreate_v_i32m4x2(vint32m4_t v0, vint32m4_t v1) {
>> +  return __riscv_vcreate_v_i32m4x2(v0, v1);
>> +}
>> +
>> +vint64m2x2_t test_vcreate_v_i64m2x2(vint64m2_t v0, vint64m2_t v1) {
>> +  return __riscv_vcreate_v_i64m2x2(v0, v1);
>> +}
>> +
>> +vint64m2x3_t test_vcreate_v_i64m2x3(vint64m2_t v0, vint64m2_t v1, vint64m2_t v2) {
>> +  return __riscv_vcreate_v_i64m2x3(v0, v1, v2);
>> +}
>> +
>> +vint64m2x4_t test_vcreate_v_i64m2x4(vint64m2_t v0, vint64m2_t v1, vint64m2_t v2, vint64m2_t v3) {
>> +  return __riscv_vcreate_v_i64m2x4(v0, v1, v2, v3);
>> +}
>> +
>> +/* { dg-final { scan-assembler-times {vsetvli\s+[a-x0-9]+,\s*zero,\s*e16,\s*mf4,\s*t[au],\s*m[au]} 7 } } */
>> +/* { dg-final { scan-assembler-times {vle16\.v\s+v[0-9]+,\s*0\([0-9ax]+\)} 35 } } */
>> +/* { dg-final { scan-assembler-times {vse16\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 35 } } */
>> +/* { dg-final { scan-assembler-times {vl2re8\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 9 } } */
>> +/* { dg-final { scan-assembler-times {vl1re32\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 35 } } */
>> +/* { dg-final { scan-assembler-times {vl2re64\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 18 } } */
>> +/* { dg-final { scan-assembler-times {vl4re64\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 2 } } */
>> +/* { dg-final { scan-assembler-times {vs1r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 35 } } */
>> +/* { dg-final { scan-assembler-times {vs2r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 27 } } */
>> +/* { dg-final { scan-assembler-times {vs4r\.v\s+v[0-9]+,\s*0\([a-x][0-9]+\)} 8 } } */
>> --
>> 2.17.1
>>
>>

      reply	other threads:[~2023-09-12  3:21 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-09-12  2:08 Li Xu
2023-09-12  2:16 ` juzhe.zhong
2023-09-12  2:46   ` Kito Cheng
2023-09-12  3:21     ` Li Xu [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230912112127040396153@eswincomputing.com \
    --to=xuli1@eswincomputing.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=juzhe.zhong@rivai.ai \
    --cc=kito.cheng@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).