Skip to content

ops: embiggen the ops framework #13015

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 2 commits into from
Closed

ops: embiggen the ops framework #13015

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
b047013
ops: embiggen the ops framework
hppritcha Jan 2, 2025
7664e23
pr feedback
hppritcha Jan 8, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion ompi/mca/op/aarch64/op_aarch64.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,8 @@
* reserved.
* Copyright (c) 2019 Arm Ltd. All rights reserved.
* Copyright (c) 2024 NVIDIA Corporation. All rights reserved.
* Copyright (c) 2025 Triad National Security, LLC. All rights
* reserved.
*
* $COPYRIGHT$
*
Expand Down Expand Up @@ -31,7 +33,7 @@ BEGIN_C_DECLS
*/
typedef struct {
/** The base op component struct */
ompi_op_base_component_1_0_0_t super;
ompi_op_base_component_2_0_0_t super;

/* What follows is aarch64-component-specific cached information. We
tend to use this scheme (caching information on the aarch64
Expand Down
10 changes: 6 additions & 4 deletions ompi/mca/op/aarch64/op_aarch64_component.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,8 @@
* Copyright (c) 2024 NVIDIA Corporation. All rights reserved.
* Copyright (c) 2024 Research Organization for Information Science
* and Technology (RIST). All rights reserved.
* Copyright (c) 2025 Triad National Security, LLC. All rights
* reserved.
*
* $COPYRIGHT$
*
Expand Down Expand Up @@ -34,7 +36,7 @@ static int mca_op_aarch64_component_open(void);
static int mca_op_aarch64_component_close(void);
static int mca_op_aarch64_component_init_query(bool enable_progress_threads,
bool enable_mpi_thread_multiple);
static struct ompi_op_base_module_1_0_0_t *
static struct ompi_op_base_module_2_0_0_t *
mca_op_aarch64_component_op_query(struct ompi_op_t *op, int *priority);
static int mca_op_aarch64_component_register(void);

Expand All @@ -43,7 +45,7 @@ ompi_op_aarch64_component_t mca_op_aarch64_component = {
information about the component itself */
{
.opc_version = {
OMPI_OP_BASE_VERSION_1_0_0,
OMPI_OP_BASE_VERSION_2_0_0,

.mca_component_name = "aarch64",
MCA_BASE_MAKE_VERSION(component, OMPI_MAJOR_VERSION, OMPI_MINOR_VERSION,
Expand Down Expand Up @@ -164,7 +166,7 @@ ompi_op_aarch64_3buff_functions_sve[OMPI_OP_BASE_FORTRAN_OP_MAX][OMPI_OP_BASE_TY
/*
* Query whether this component can be used for a specific op
*/
static struct ompi_op_base_module_1_0_0_t *
static struct ompi_op_base_module_2_0_0_t *
mca_op_aarch64_component_op_query(struct ompi_op_t *op, int *priority)
{
/* Sanity check -- although the framework should never invoke the
Expand Down Expand Up @@ -242,5 +244,5 @@ static struct ompi_op_base_module_1_0_0_t *
if (NULL != module) {
*priority = 50;
}
return (ompi_op_base_module_1_0_0_t *) module;
return (ompi_op_base_module_2_0_0_t *) module;
}
28 changes: 15 additions & 13 deletions ompi/mca/op/aarch64/op_aarch64_functions.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -6,6 +6,8 @@
* Copyright (c) 2024 NVIDIA Corporation. All rights reserved.
* Copyright (c) 2024 Research Organization for Information Science
* and Technology (RIST). All rights reserved.
* Copyright (c) 2025 Triad National Security, LLC. All rights
* reserved.
*
* $COPYRIGHT$
*
Expand Down Expand Up @@ -114,11 +116,11 @@ _Generic((*(out)), \
#if defined(GENERATE_NEON_CODE)
#define OP_AARCH64_FUNC(name, type_name, type_size, type_cnt, type, op) \
static void OP_CONCAT(ompi_op_aarch64_2buff_##name##_##type##type_size##_t, \
APPEND)(const void *_in, void *_out, int *count, \
APPEND)(const void *_in, void *_out, size_t *count, \
struct ompi_datatype_t **dtype, \
struct ompi_op_base_module_1_0_0_t *module) \
struct ompi_op_base_module_2_0_0_t *module) \
{ \
int left_over = *count; \
size_t left_over = *count; \
type##type_size##_t *in = (type##type_size##_t *) _in, \
*out = (type##type_size##_t *) _out; \
OP_CONCAT(OMPI_OP_TYPE_PREPEND, type##type_size##x##type_cnt##_t) vsrc, vdst; \
Expand All @@ -138,12 +140,12 @@ _Generic((*(out)), \
#elif defined(GENERATE_SVE_CODE)
#define OP_AARCH64_FUNC(name, type_name, type_size, type_cnt, type, op) \
static void OP_CONCAT(ompi_op_aarch64_2buff_##name##_##type##type_size##_t, APPEND) \
(const void *_in, void *_out, int *count, \
(const void *_in, void *_out, size_t *count, \
struct ompi_datatype_t **dtype, \
struct ompi_op_base_module_1_0_0_t *module) \
struct ompi_op_base_module_2_0_0_t *module) \
{ \
const int types_per_step = svcnt(*((type##type_size##_t *) _in)); \
const int cnt = *count; \
const size_t cnt = *count; \
type##type_size##_t *in = (type##type_size##_t *) _in, \
*out = (type##type_size##_t *) _out; \
OP_CONCAT(OMPI_OP_TYPE_PREPEND, type##type_size##_t) vsrc, vdst; \
Expand Down Expand Up @@ -279,11 +281,11 @@ _Generic((*(out)), \
#if defined(GENERATE_NEON_CODE)
#define OP_AARCH64_FUNC_3BUFF(name, type_name, type_size, type_cnt, type, op) \
static void OP_CONCAT(ompi_op_aarch64_3buff_##name##_##type##type_size##_t, APPEND) \
(const void *_in1, const void *_in2, void *_out, int *count, \
(const void *_in1, const void *_in2, void *_out, size_t *count, \
struct ompi_datatype_t **dtype, \
struct ompi_op_base_module_1_0_0_t *module) \
struct ompi_op_base_module_2_0_0_t *module) \
{ \
int left_over = *count; \
size_t left_over = *count; \
type##type_size##_t *in1 = (type##type_size##_t *) _in1, \
*in2 = (type##type_size##_t *) _in2, \
*out = (type##type_size##_t *) _out; \
Expand All @@ -304,17 +306,17 @@ static void OP_CONCAT(ompi_op_aarch64_3buff_##name##_##type##type_size##_t, APPE
#elif defined(GENERATE_SVE_CODE)
#define OP_AARCH64_FUNC_3BUFF(name, type_name, type_size, type_cnt, type, op) \
static void OP_CONCAT(ompi_op_aarch64_3buff_##name##_##type##type_size##_t, APPEND) \
(const void *_in1, const void *_in2, void *_out, int *count, \
(const void *_in1, const void *_in2, void *_out, size_t *count, \
struct ompi_datatype_t **dtype, \
struct ompi_op_base_module_1_0_0_t *module) \
struct ompi_op_base_module_2_0_0_t *module) \
{ \
const int types_per_step = svcnt(*((type##type_size##_t *) _in1)); \
type##type_size##_t *in1 = (type##type_size##_t *) _in1, \
*in2 = (type##type_size##_t *) _in2, \
*out = (type##type_size##_t *) _out; \
const int cnt = *count; \
const size_t cnt = *count; \
OP_CONCAT(OMPI_OP_TYPE_PREPEND, type##type_size##_t) vsrc, vdst; \
for (int idx=0; idx < cnt; idx += types_per_step) { \
for (size_t idx=0; idx < cnt; idx += types_per_step) { \
svbool_t pred = svwhilelt_b##type_size(idx, cnt); \
vsrc = svld1(pred, &in1[idx]); \
vdst = svld1(pred, &in2[idx]); \
Expand Down
4 changes: 3 additions & 1 deletion ompi/mca/op/avx/op_avx.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,6 +2,8 @@
* Copyright (c) 2019-2020 The University of Tennessee and The University
* of Tennessee Research Foundation. All rights
* reserved.
* Copyright (c) 2025 Triad National Security, LLC. All rights
* reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
Expand Down Expand Up @@ -37,7 +39,7 @@ BEGIN_C_DECLS
*/
typedef struct {
/** The base op component struct */
ompi_op_base_component_1_0_0_t super;
ompi_op_base_component_2_0_0_t super;

/* What follows is avx-component-specific cached information. We
tend to use this scheme (caching information on the avx
Expand Down
10 changes: 6 additions & 4 deletions ompi/mca/op/avx/op_avx_component.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,8 @@
* Copyright (c) 2020 Research Organization for Information Science
* and Technology (RIST). All rights reserved.
* Copyright (c) 2021 Cisco Systems, Inc. All rights reserved.
* Copyright (c) 2025 Triad National Security, LLC. All rights
* reserved.
* $COPYRIGHT$
*
* Additional copyrights may follow
Expand Down Expand Up @@ -32,7 +34,7 @@ static int avx_component_open(void);
static int avx_component_close(void);
static int avx_component_init_query(bool enable_progress_threads,
bool enable_mpi_thread_multiple);
static struct ompi_op_base_module_1_0_0_t *
static struct ompi_op_base_module_2_0_0_t *
avx_component_op_query(struct ompi_op_t *op, int *priority);
static int avx_component_register(void);

Expand Down Expand Up @@ -131,7 +133,7 @@ static uint32_t has_intel_AVX_features(void)
ompi_op_avx_component_t mca_op_avx_component = {
{
.opc_version = {
OMPI_OP_BASE_VERSION_1_0_0,
OMPI_OP_BASE_VERSION_2_0_0,

.mca_component_name = "avx",
MCA_BASE_MAKE_VERSION(component, OMPI_MAJOR_VERSION, OMPI_MINOR_VERSION,
Expand Down Expand Up @@ -250,7 +252,7 @@ avx_component_init_query(bool enable_progress_threads,
/*
* Query whether this component can be used for a specific op
*/
static struct ompi_op_base_module_1_0_0_t*
static struct ompi_op_base_module_2_0_0_t*
avx_component_op_query(struct ompi_op_t *op, int *priority)
{
ompi_op_base_module_t *module = NULL;
Expand Down Expand Up @@ -325,5 +327,5 @@ avx_component_op_query(struct ompi_op_t *op, int *priority)
if (NULL != module) {
*priority = 50;
}
return (ompi_op_base_module_1_0_0_t *) module;
return (ompi_op_base_module_2_0_0_t *) module;
}
Loading
Loading