[ 4 / 4 ] Application profile is long enough (582.19 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 1.18 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 4 / 4 ] Enough time of the experiment time spent in analyzed loops (31.18%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 3 / 4 ] A significant amount of threads are idle (22.55%)

On average, more than 10% of observed threads are idle. Such threads are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 3 / 4 ] CPU activity is below 90% (77.46%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 4 / 4 ] Loop profile is not flat

At least one loop coverage is greater than 4% (20.38%), representing an hotspot for the application

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (10.62%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 0 / 4 ] Affinity stability is lower than 90% (17.20%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 3 / 3 ] Functions mostly use all threads

Functions running on a reduced number of threads (typically sequential code) cover less than 10% of application walltime (3.50%)

[ 0 / 3 ] Cumulative Outermost/In between loops coverage (20.55%) greater than cumulative innermost loop coverage (10.62%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.00%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (332.88 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 1.12 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (26.93%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 4 / 4 ] Threads activity is good

On average, more than 143.31% of observed threads are actually active

[ 3 / 4 ] CPU activity is below 90% (71.70%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 4 / 4 ] Loop profile is not flat

At least one loop coverage is greater than 4% (16.72%), representing an hotspot for the application

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (10.09%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 0 / 4 ] Affinity stability is lower than 90% (23.29%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 3 / 3 ] Functions mostly use all threads

Functions running on a reduced number of threads (typically sequential code) cover less than 10% of application walltime (4.06%)

[ 0 / 3 ] Cumulative Outermost/In between loops coverage (16.84%) greater than cumulative innermost loop coverage (10.09%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.00%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (254.60 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.78 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (14.97%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 4 / 4 ] Threads activity is good

On average, more than 269.09% of observed threads are actually active

[ 2 / 4 ] CPU activity is below 90% (67.33%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 4 / 4 ] Loop profile is not flat

At least one loop coverage is greater than 4% (7.37%), representing an hotspot for the application

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (7.48%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 1 / 4 ] Affinity stability is lower than 90% (33.78%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 3 / 3 ] Functions mostly use all threads

Functions running on a reduced number of threads (typically sequential code) cover less than 10% of application walltime (5.24%)

[ 0 / 3 ] Cumulative Outermost/In between loops coverage (7.49%) greater than cumulative innermost loop coverage (7.48%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.00%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (170.84 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.60 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (9.03%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 4 / 4 ] Threads activity is good

On average, more than 517.15% of observed threads are actually active

[ 2 / 4 ] CPU activity is below 90% (64.73%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 3.08%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (8.67%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (5.87%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 2 / 4 ] Affinity stability is lower than 90% (58.96%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 3 / 3 ] Functions mostly use all threads

Functions running on a reduced number of threads (typically sequential code) cover less than 10% of application walltime (7.64%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (3.17%) lower than cumulative innermost loop coverage (5.87%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.00%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (110.85 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.46 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (6.69%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 4 / 4 ] Threads activity is good

On average, more than 1036.50% of observed threads are actually active

[ 2 / 4 ] CPU activity is below 90% (64.91%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 1.53%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (6.40%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (5.07%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 3 / 4 ] Affinity stability is lower than 90% (86.31%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 0 / 3 ] Too many functions do not use all threads

Functions running on a reduced number of threads (typically sequential code) cover at least 10% of application walltime (12.01%). Check both "Max Inclusive Time Over Threads" and "Nb Threads" in Functions or Loops tabs and consider parallelizing sequential regions or improving parallelization of regions running on a reduced number of threads

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (1.62%) lower than cumulative innermost loop coverage (5.07%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.00%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (72.70 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.35 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.99%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 4 / 4 ] Threads activity is good

On average, more than 2188.61% of observed threads are actually active

[ 2 / 4 ] CPU activity is below 90% (68.72%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 1.49%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.76%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (4.34%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 3 / 4 ] Affinity stability is lower than 90% (80.72%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 0 / 3 ] Too many functions do not use all threads

Functions running on a reduced number of threads (typically sequential code) cover at least 10% of application walltime (19.29%). Check both "Max Inclusive Time Over Threads" and "Nb Threads" in Functions or Loops tabs and consider parallelizing sequential regions or improving parallelization of regions running on a reduced number of threads

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.65%) lower than cumulative innermost loop coverage (4.34%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.01%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (55.76 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.51 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.07%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 4 / 4 ] Threads activity is good

On average, more than 4117.98% of observed threads are actually active

[ 2 / 4 ] CPU activity is below 90% (64.97%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 1.85%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (3.90%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.88%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 2 / 4 ] Affinity stability is lower than 90% (69.16%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 3 / 3 ] Functions mostly use all threads

Functions running on a reduced number of threads (typically sequential code) cover less than 10% of application walltime (6.03%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.19%) lower than cumulative innermost loop coverage (3.88%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.00%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (52.45 s)

To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

[ 0 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 100.00% of the time spent in analyzed modules. Check that -g is present.
Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case.

[ 0 / 3 ] Some functions are compiled with a low optimization level (O0 or O1)

To have better performances, it is advised to help the compiler by using a proper optimization level (-O2 of higher). Warning, depending on compilers, faster optimization levels can decrease numeric accuracy.

[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.

[ 0 / 3 ] Compilation of some functions is not optimized for the target processor

Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ).

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.20 % of the execution time)

To have a representative profiling, it is advised that the category "Others" represents less than 20% of the execution time in order to analyze as much as possible of the user code

[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.06%)

If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.

[ 4 / 4 ] Threads activity is good

On average, more than 5301.33% of observed threads are actually active

[ 2 / 4 ] CPU activity is below 90% (62.48%)

CPU cores are idle more than 10% of time. Threads supposed to run on these cores are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling.

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 2.15%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (3.91%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.93%)

If the time spent in analyzed innermost loops is less than 15%, standard innermost loop optimizations such as vectorisation will have a limited impact on application performances.

[ 2 / 4 ] Affinity stability is lower than 90% (68.67%)

Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map.

[ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations

It could be more efficient to inline by hand BLAS1 operations

[ 0 / 3 ] Too many functions do not use all threads

Functions running on a reduced number of threads (typically sequential code) cover at least 10% of application walltime (27.48%). Check both "Max Inclusive Time Over Threads" and "Nb Threads" in Functions or Loops tabs and consider parallelizing sequential regions or improving parallelization of regions running on a reduced number of threads

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.12%) lower than cumulative innermost loop coverage (3.93%)

Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex

[ 2 / 2 ] Less than 10% (0.00%) is spend in BLAS2 operations

BLAS2 calls usually could make a poor cache usage and could benefit from inlining.

[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions)