[ 4 / 4 ] Application profile is long enough (89.17 s)

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

[ 3 / 3 ] Optimization level option is correctly used

[ 2.81 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.06 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (6.53%)

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 98.86% of observed threads are actually active

[ 4 / 4 ] CPU activity is good

CPU cores are active 98.86% of time

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 3.36%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (6.49%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (5.58%)

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.

[ 4 / 4 ] Affinity is good (99.96%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (0.00%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.94%) lower than cumulative innermost loop coverage (5.58%)

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% (0.20%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (44.30 s)

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

[ 3 / 3 ] Optimization level option is correctly used

[ 2.21 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 26.32% 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.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.07 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (5.97%)

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 189.55% of observed threads are actually active

[ 4 / 4 ] CPU activity is good

CPU cores are active 95.68% of time

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 3.47%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (5.91%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (5.12%)

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.

[ 4 / 4 ] Affinity is good (99.59%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (0.87%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.85%) lower than cumulative innermost loop coverage (5.12%)

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% (0.18%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (22.50 s)

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

[ 3 / 3 ] Optimization level option is correctly used

[ 2.65 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.06 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (5.81%)

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 357.96% of observed threads are actually active

[ 4 / 4 ] CPU activity is good

CPU cores are active 91.92% of time

[ 0 / 4 ] Loop profile is flat

No hotspot found in the application (greatest loop coverage is 3.39%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (5.75%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (4.94%)

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.

[ 4 / 4 ] Affinity is good (98.92%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (4.03%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.86%) lower than cumulative innermost loop coverage (4.94%)

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% (0.20%) is spend in Libm/SVML (special functions)

[ 4 / 4 ] Application profile is long enough (12.43 s)

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

[ 3 / 3 ] Optimization level option is correctly used

[ 2.53 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.05 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.91%)

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 642.20% of observed threads are actually active

[ 3 / 4 ] CPU activity is below 90% (84.52%)

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 3.04%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.84%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (4.08%)

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.

[ 4 / 4 ] Affinity is good (97.95%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (3.71%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.83%) lower than cumulative innermost loop coverage (4.08%)

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% (0.23%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (7.66 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.63 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.04 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.52%)

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 1072.59% of observed threads are actually active

[ 3 / 4 ] CPU activity is below 90% (72.26%)

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 2.51%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.46%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.84%)

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.

[ 4 / 4 ] Affinity is good (96.92%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (9.07%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.68%) lower than cumulative innermost loop coverage (3.84%)

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% (0.19%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (5.54 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.84 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.04 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.30%)

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 1321.85% of observed threads are actually active

[ 2 / 4 ] CPU activity is below 90% (60.23%)

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 2.72%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.23%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.69%)

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.

[ 4 / 4 ] Affinity is good (96.35%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (6.85%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.61%) lower than cumulative innermost loop coverage (3.69%)

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% (0.23%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (4.41 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.32 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information

Functions without compilation information (typically not compiled with -g) cumulate 22.73% 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.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.14%)

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 1456.49% of observed threads are actually active

[ 1 / 4 ] CPU activity is below 90% (49.98%)

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 2.38%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.11%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.61%)

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.

[ 4 / 4 ] Affinity is good (96.08%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (6.93%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.53%) lower than cumulative innermost loop coverage (3.61%)

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% (0.20%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (3.71 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.61 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.24%)

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 1521.54% of observed threads are actually active

[ 1 / 4 ] CPU activity is below 90% (41.84%)

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 2.66%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.18%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.77%)

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.

[ 4 / 4 ] Affinity is good (96.06%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (6.55%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.47%) lower than cumulative innermost loop coverage (3.77%)

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% (0.19%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (3.26 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.57 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.26%)

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 1525.66% of observed threads are actually active

[ 1 / 4 ] CPU activity is below 90% (34.81%)

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 2.73%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.22%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.81%)

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.

[ 4 / 4 ] Affinity is good (96.15%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (7.47%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.45%) lower than cumulative innermost loop coverage (3.81%)

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% (0.22%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (2.76 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.55 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.43%)

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 1518.43% of observed threads are actually active

[ 0 / 4 ] CPU activity is below 90% (29.71%)

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 2.78%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.37%)

[ 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.

[ 4 / 4 ] Affinity is good (96.03%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (6.35%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.50%) 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% (0.22%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (2.46 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.85 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.56%)

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 1487.74% of observed threads are actually active

[ 0 / 4 ] CPU activity is below 90% (25.39%)

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 2.65%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.50%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (4.01%)

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.

[ 4 / 4 ] Affinity is good (96.12%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (6.24%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.54%) lower than cumulative innermost loop coverage (4.01%)

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% (0.23%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (2.29 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.71 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.97%)

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 1442.79% of observed threads are actually active

[ 0 / 4 ] CPU activity is below 90% (21.76%)

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 1.90%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.91%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (4.51%)

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.

[ 4 / 4 ] Affinity is good (96.44%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (6.23%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.46%) lower than cumulative innermost loop coverage (4.51%)

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% (0.19%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (2.07 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.61 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.32%)

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 1436.63% of observed threads are actually active

[ 0 / 4 ] CPU activity is below 90% (19.47%)

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 2.32%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.26%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (3.84%)

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.

[ 4 / 4 ] Affinity is good (96.49%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (5.91%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.48%) lower than cumulative innermost loop coverage (3.84%)

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% (0.24%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (1.92 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.80 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (4.94%)

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 1384.42% of observed threads are actually active

[ 0 / 4 ] CPU activity is below 90% (16.97%)

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 1.74%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (4.89%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (4.51%)

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.

[ 4 / 4 ] Affinity is good (96.61%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (5.58%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.43%) lower than cumulative innermost loop coverage (4.51%)

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% (0.26%) is spend in Libm/SVML (special functions)

[ 0 / 4 ] Application profile is too short (1.83 s)

If the overall application profiling time is less than 10 seconds, many of the measurements at function or loop level will very likely be under the measurement quality threshold (0,1 seconds). Rerun to increase runtime duration: for example use a larger dataset or include a repetition loop.

[ 3 / 3 ] Optimization level option is correctly used

[ 2.55 / 3 ] Most of time spent in analyzed modules comes from functions compiled with -g and -fno-omit-frame-pointer

-g option gives access to debugging informations, such are source locations. -fno-omit-frame-pointer improve the accuracy of callchains found during the application profiling.

[ 2 / 3 ] Security settings from the host restrict profiling. Some metrics will be missing or incomplete.

Current value for kernel.perf_event_paranoid is 2. If possible, set it to 1 or check with your system administrator which flag can be used to achieve this.

[ 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 ( -mcpu=native ). Application run on the ARM_NEOVERSE_V2 micro-architecture while the code was specialized for armv8.2-a+crypto.

[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.03 % 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 / 0 ] Fastmath not used

Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions.

[ 0 / 4 ] Too little time of the experiment time spent in analyzed loops (5.17%)

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 1325.86% of observed threads are actually active

[ 0 / 4 ] CPU activity is below 90% (14.83%)

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 3.07%), and the twenty hottest loops cumulated coverage is lower than 20% of the application profiled time (5.10%)

[ 0 / 4 ] Too little time of the experiment time spent in analyzed innermost loops (4.74%)

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.

[ 4 / 4 ] Affinity is good (96.88%)

Threads are not migrating to CPU cores: probably successfully pinned

[ 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 (5.39%)

[ 3 / 3 ] Cumulative Outermost/In between loops coverage (0.44%) lower than cumulative innermost loop coverage (4.74%)

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% (0.20%) is spend in Libm/SVML (special functions)