Skip to content

framework: half-precision presets #566

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

Merged
Treece-Burgess merged 3 commits into from
Feb 19, 2026
Merged

framework: half-precision presets #566

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
0435607
framework: support for half-precision CPU presets
dbarry9 Feb 3, 2026
30ba0e4
presets: add HP for Intel SPR and EMR
dbarry9 Feb 18, 2026
179d932
framework: consistent number of derived terms
dbarry9 Feb 18, 2026
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: 4 additions & 0 deletions src/papiStdEventDefs.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -166,6 +166,8 @@ enum
PAPI_VEC_SP_idx, /* Single precision vector/SIMD instructions */
PAPI_VEC_DP_idx, /* Double precision vector/SIMD instructions */
PAPI_REF_CYC_idx, /* Reference clock cycles */
PAPI_HP_OPS_idx, /* Floating point operations executed; optimized to count scaled half precision vector operations */
PAPI_VEC_HP_idx, /* Half precision vector/SIMD instructions */
PAPI_END_idx /*This should always be last! */
};

Expand Down Expand Up @@ -277,6 +279,8 @@ enum
#define PAPI_VEC_SP (PAPI_VEC_SP_idx | PAPI_PRESET_MASK) /* Single precision vector/SIMD instructions */
#define PAPI_VEC_DP (PAPI_VEC_DP_idx | PAPI_PRESET_MASK) /* Double precision vector/SIMD instructions */
#define PAPI_REF_CYC (PAPI_REF_CYC_idx | PAPI_PRESET_MASK) /* Reference clock cycles */
#define PAPI_HP_OPS (PAPI_HP_OPS_idx | PAPI_PRESET_MASK) /* Floating point operations executed; optimized to count scaled half precision vector operations */
#define PAPI_VEC_HP (PAPI_VEC_HP_idx | PAPI_PRESET_MASK) /* Half precision vector/SIMD instructions */

#define PAPI_END (PAPI_END_idx | PAPI_PRESET_MASK) /*This should always be last! */

Expand Down
12 changes: 10 additions & 2 deletions src/papi_common_strings.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -544,8 +544,16 @@ hwi_presets_t _papi_hwi_presets[PAPI_MAX_PRESET_EVENTS] = {
"Reference clock cycles", 0,
0, PAPI_PRESET_BIT_MSC,
NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
/*108 */ {NULL, NULL, NULL, 0, 0, 0, NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
/*109 */ {NULL, NULL, NULL, 0, 0, 0, NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
/*108 */ {"PAPI_HP_OPS",
"HP operations",
"Floating point operations; optimized to count scaled half precision vector operations", 0,
0, PAPI_PRESET_BIT_INS + PAPI_PRESET_BIT_FP,
NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
/*109 */ {"PAPI_VEC_HP",
"HP Vector/SIMD instr",
"Half precision vector/SIMD instructions", 0,
0, PAPI_PRESET_BIT_INS + PAPI_PRESET_BIT_FP,
NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
/*110 */ {NULL, NULL, NULL, 0, 0, 0, NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
/*111 */ {NULL, NULL, NULL, 0, 0, 0, NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
/*112 */ {NULL, NULL, NULL, 0, 0, 0, NULL, {0}, {NULL}, {NULL}, {0}, {NULL}, NULL, 0, 0, {NULL}, {NULL}},
Expand Down
8 changes: 5 additions & 3 deletions src/papi_events.csv
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1041,11 +1041,13 @@ PRESET,PAPI_REF_CYC,NOT_DERIVED,UNHALTED_REFERENCE_CYCLES
# FLOPs
PRESET,PAPI_DP_OPS,DERIVED_POSTFIX,N0|N1|2|*|+|N2|4|*|+|N3|8|*|+|,FP_ARITH_INST_RETIRED:SCALAR_DOUBLE,FP_ARITH_INST_RETIRED:128B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:256B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:512B_PACKED_DOUBLE
PRESET,PAPI_SP_OPS,DERIVED_POSTFIX,N0|N1|4|*|+|N2|8|*|+|N3|16|*|+|,FP_ARITH_INST_RETIRED:SCALAR_SINGLE,FP_ARITH_INST_RETIRED:128B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:256B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:512B_PACKED_SINGLE
PRESET,PAPI_FP_OPS,DERIVED_POSTFIX,N0|N1|4|*|+|N2|8|*|+|N3|16|*|+|N4|+|N5|2|*|+|N6|4|*|+|N7|8|*|+|,FP_ARITH_INST_RETIRED:SCALAR_SINGLE,FP_ARITH_INST_RETIRED:128B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:256B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:512B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:SCALAR_DOUBLE,FP_ARITH_INST_RETIRED:128B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:256B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:512B_PACKED_DOUBLE
PRESET,PAPI_FP_INS,DERIVED_POSTFIX,N0|N1|N2|N3|N4|N5|N6|N7|+|+|+|+|+|+|+|,FP_ARITH_INST_RETIRED:SCALAR_SINGLE,FP_ARITH_INST_RETIRED:128B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:256B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:512B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:SCALAR_DOUBLE,FP_ARITH_INST_RETIRED:128B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:256B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:512B_PACKED_DOUBLE
PRESET,PAPI_HP_OPS,DERIVED_POSTFIX,N0|N1|8|*|+|N2|16|*|+|N3|32|*|+|,FP_ARITH_INST_RETIRED2:SCALAR_HALF,FP_ARITH_INST_RETIRED2:128B_PACKED_HALF,FP_ARITH_INST_RETIRED2:256B_PACKED_HALF,FP_ARITH_INST_RETIRED2:512B_PACKED_HALF
#PRESET,PAPI_FP_OPS,DERIVED_POSTFIX,N0|N1|4|*|+|N2|8|*|+|N3|16|*|+|N4|+|N5|2|*|+|N6|4|*|+|N7|8|*|+|,FP_ARITH_INST_RETIRED:SCALAR_SINGLE,FP_ARITH_INST_RETIRED:128B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:256B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:512B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:SCALAR_DOUBLE,FP_ARITH_INST_RETIRED:128B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:256B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:512B_PACKED_DOUBLE
PRESET,PAPI_FP_INS,DERIVED_POSTFIX,N0|N1|N2|N3|+|+|+|,FP_ARITH_INST_RETIRED:SCALAR,FP_ARITH_INST_RETIRED:VECTOR,FP_ARITH_INST_RETIRED2:SCALAR,FP_ARITH_INST_RETIRED2:VECTOR
PRESET,PAPI_VEC_DP,DERIVED_POSTFIX,N0|N1|N2|+|+|,FP_ARITH_INST_RETIRED:128B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:256B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:512B_PACKED_DOUBLE
PRESET,PAPI_VEC_SP,DERIVED_POSTFIX,N0|N1|N2|+|+|,FP_ARITH_INST_RETIRED:128B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:256B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:512B_PACKED_SINGLE
PRESET,PAPI_VEC_INS,DERIVED_POSTFIX,N0|N1|N2|N3|N4|N5|+|+|+|+|+|,FP_ARITH_INST_RETIRED:128B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:256B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:512B_PACKED_SINGLE,FP_ARITH_INST_RETIRED:128B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:256B_PACKED_DOUBLE,FP_ARITH_INST_RETIRED:512B_PACKED_DOUBLE
PRESET,PAPI_VEC_HP,DERIVED_POSTFIX,N0|N1|N2|+|+|,FP_ARITH_INST_RETIRED2:128B_PACKED_HALF,FP_ARITH_INST_RETIRED2:256B_PACKED_HALF,FP_ARITH_INST_RETIRED2:512B_PACKED_HALF
PRESET,PAPI_VEC_INS,DERIVED_ADD,FP_ARITH_INST_RETIRED:VECTOR,FP_ARITH_INST_RETIRED2:VECTOR
# Branches
PRESET,PAPI_BR_UCN,DERIVED_SUB,BR_INST_RETIRED:ALL_BRANCHES,BR_INST_RETIRED:COND
PRESET,PAPI_BR_CN,NOT_DERIVED,BR_INST_RETIRED:COND
Expand Down
2 changes: 1 addition & 1 deletion src/papi_internal.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -97,7 +97,7 @@ extern char **_papi_errlist;
#define NEED_CONTEXT 1
#define DONT_NEED_CONTEXT 0

#define PAPI_EVENTS_IN_DERIVED_EVENT 8
#define PAPI_EVENTS_IN_DERIVED_EVENT 12
#define PAPI_MAX_COMP_QUALS 8


Expand Down