Help is available by moving the cursor above any 
symbol or by checking MAQAO website.
[ 4 / 4 ] Application profile is long enough (331.25 s)
To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.
[ 3.00 / 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.
[ 3 / 3 ] Optimization level option is correctly used
[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.
[ 3.00 / 3 ] Architecture specific option -x Host is used
[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.00 % 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 (100.00%)
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 99.17% of observed threads are actually active
[ 4 / 4 ] CPU activity is good
CPU cores are active 99.17% of time
[ 4 / 4 ] Loop profile is not flat
At least one loop coverage is greater than 4% (93.25%), representing an hotspot for the application
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (96.21%)
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% (72.08%)
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 (0.00%)
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (3.79%) lower than cumulative innermost loop coverage (96.21%)
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)
| Loop ID | Analysis | Penalty Score |
|---|---|---|
| ►Loop 28 - kmeans-icpx-Ofast | Execution Time: 93 % - Vectorization Ratio: 19.15 % - Vector Length Use: 16.49 % | |
| ►Data Access Issues | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Inefficient Vectorization | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Loop 26 - kmeans-icpx-Ofast | Execution Time: 3 % - Vectorization Ratio: 13.92 % - Vector Length Use: 12.75 % | |
| ►Loop Computation Issues | 2 | |
| ○ | [SA] Presence of a large number of scalar integer instructions - Simplify loop structure, perform loop splitting or perform unroll and jam. This issue costs 2 points. | 2 |
| ►Control Flow Issues | 6 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Data Access Issues | 9 | |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○ | [SA] More than 20% of the loads are accessing the stack - Perform loop splitting to decrease pressure on registers. This issue costs 2 points. | 2 |
| ►Vectorization Roadblocks | 12 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Inefficient Vectorization | 1 | |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○Loop 27 - kmeans-icpx-Ofast | Execution Time: 2 % - Vectorization Ratio: 0.00 % - Vector Length Use: 11.61 % |
[ 4 / 4 ] Application profile is long enough (165.71 s)
To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.
[ 3.00 / 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.
[ 3 / 3 ] Optimization level option is correctly used
[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.
[ 3.00 / 3 ] Architecture specific option -x Host is used
[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.00 % 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 (99.97%)
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 134.18% of observed threads are actually active
[ 3 / 4 ] CPU activity is below 90% (85.72%)
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.
[ 4 / 4 ] Loop profile is not flat
At least one loop coverage is greater than 4% (93.38%), representing an hotspot for the application
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (96.33%)
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% (72.06%)
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 (0.03%)
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (3.64%) lower than cumulative innermost loop coverage (96.33%)
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)
| Loop ID | Analysis | Penalty Score |
|---|---|---|
| ►Loop 28 - kmeans-icpx-Ofast | Execution Time: 93 % - Vectorization Ratio: 19.15 % - Vector Length Use: 16.49 % | |
| ►Data Access Issues | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Inefficient Vectorization | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Loop 26 - kmeans-icpx-Ofast | Execution Time: 3 % - Vectorization Ratio: 13.92 % - Vector Length Use: 12.75 % | |
| ►Loop Computation Issues | 2 | |
| ○ | [SA] Presence of a large number of scalar integer instructions - Simplify loop structure, perform loop splitting or perform unroll and jam. This issue costs 2 points. | 2 |
| ►Control Flow Issues | 6 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Data Access Issues | 9 | |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○ | [SA] More than 20% of the loads are accessing the stack - Perform loop splitting to decrease pressure on registers. This issue costs 2 points. | 2 |
| ►Vectorization Roadblocks | 12 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Inefficient Vectorization | 1 | |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○Loop 27 - kmeans-icpx-Ofast | Execution Time: 2 % - Vectorization Ratio: 0.00 % - Vector Length Use: 11.61 % |
[ 4 / 4 ] Application profile is long enough (82.93 s)
To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.
[ 3.00 / 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.
[ 3 / 3 ] Optimization level option is correctly used
[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.
[ 3.00 / 3 ] Architecture specific option -x Host is used
[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.00 % 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 (99.96%)
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 172.48% of observed threads are actually active
[ 3 / 4 ] CPU activity is below 90% (79.00%)
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% (93.34%), representing an hotspot for the application
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (96.32%)
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% (72.08%)
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 (0.00%)
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (3.64%) lower than cumulative innermost loop coverage (96.32%)
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)
| Loop ID | Analysis | Penalty Score |
|---|---|---|
| ►Loop 28 - kmeans-icpx-Ofast | Execution Time: 93 % - Vectorization Ratio: 19.15 % - Vector Length Use: 16.49 % | |
| ►Data Access Issues | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Inefficient Vectorization | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Loop 26 - kmeans-icpx-Ofast | Execution Time: 3 % - Vectorization Ratio: 13.92 % - Vector Length Use: 12.75 % | |
| ►Loop Computation Issues | 2 | |
| ○ | [SA] Presence of a large number of scalar integer instructions - Simplify loop structure, perform loop splitting or perform unroll and jam. This issue costs 2 points. | 2 |
| ►Control Flow Issues | 6 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Data Access Issues | 9 | |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○ | [SA] More than 20% of the loads are accessing the stack - Perform loop splitting to decrease pressure on registers. This issue costs 2 points. | 2 |
| ►Vectorization Roadblocks | 12 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Inefficient Vectorization | 1 | |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○Loop 27 - kmeans-icpx-Ofast | Execution Time: 2 % - Vectorization Ratio: 0.00 % - Vector Length Use: 11.61 % |
[ 4 / 4 ] Application profile is long enough (41.55 s)
To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.
[ 3.00 / 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.
[ 3 / 3 ] Optimization level option is correctly used
[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.
[ 3.00 / 3 ] Architecture specific option -x Host is used
[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.00 % 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 (99.92%)
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 206.21% of observed threads are actually active
[ 3 / 4 ] CPU activity is below 90% (75.69%)
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% (93.32%), representing an hotspot for the application
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (96.28%)
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% (72.11%)
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 (0.00%)
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (3.64%) lower than cumulative innermost loop coverage (96.28%)
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)
| Loop ID | Analysis | Penalty Score |
|---|---|---|
| ►Loop 28 - kmeans-icpx-Ofast | Execution Time: 93 % - Vectorization Ratio: 19.15 % - Vector Length Use: 16.49 % | |
| ►Data Access Issues | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Inefficient Vectorization | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Loop 26 - kmeans-icpx-Ofast | Execution Time: 3 % - Vectorization Ratio: 13.92 % - Vector Length Use: 12.75 % | |
| ►Loop Computation Issues | 2 | |
| ○ | [SA] Presence of a large number of scalar integer instructions - Simplify loop structure, perform loop splitting or perform unroll and jam. This issue costs 2 points. | 2 |
| ►Control Flow Issues | 6 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Data Access Issues | 9 | |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○ | [SA] More than 20% of the loads are accessing the stack - Perform loop splitting to decrease pressure on registers. This issue costs 2 points. | 2 |
| ►Vectorization Roadblocks | 12 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Inefficient Vectorization | 1 | |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○Loop 27 - kmeans-icpx-Ofast | Execution Time: 2 % - Vectorization Ratio: 0.00 % - Vector Length Use: 11.61 % |
[ 4 / 4 ] Application profile is long enough (20.81 s)
To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.
[ 2.98 / 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.
[ 3 / 3 ] Optimization level option is correctly used
[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.
[ 2.98 / 3 ] Architecture specific option -x Host is used
[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.00 % 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 (99.85%)
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 231.11% of observed threads are actually active
[ 3 / 4 ] CPU activity is below 90% (74.11%)
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% (92.76%), representing an hotspot for the application
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (96.19%)
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% (72.19%)
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 (0.01%)
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (3.66%) lower than cumulative innermost loop coverage (96.19%)
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)
| Loop ID | Analysis | Penalty Score |
|---|---|---|
| ►Loop 28 - kmeans-icpx-Ofast | Execution Time: 92 % - Vectorization Ratio: 19.15 % - Vector Length Use: 16.49 % | |
| ►Data Access Issues | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Inefficient Vectorization | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Loop 26 - kmeans-icpx-Ofast | Execution Time: 3 % - Vectorization Ratio: 13.92 % - Vector Length Use: 12.75 % | |
| ►Loop Computation Issues | 2 | |
| ○ | [SA] Presence of a large number of scalar integer instructions - Simplify loop structure, perform loop splitting or perform unroll and jam. This issue costs 2 points. | 2 |
| ►Control Flow Issues | 6 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Data Access Issues | 9 | |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○ | [SA] More than 20% of the loads are accessing the stack - Perform loop splitting to decrease pressure on registers. This issue costs 2 points. | 2 |
| ►Vectorization Roadblocks | 12 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Inefficient Vectorization | 1 | |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○Loop 27 - kmeans-icpx-Ofast | Execution Time: 2 % - Vectorization Ratio: 0.00 % - Vector Length Use: 11.61 % |
[ 4 / 4 ] Application profile is long enough (12.95 s)
To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.
[ 2.90 / 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.
[ 3 / 3 ] Optimization level option is correctly used
[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics.
[ 2.90 / 3 ] Architecture specific option -x Host is used
[ 2 / 2 ] Application is correctly profiled ("Others" category represents 0.00 % 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 (98.57%)
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 252.60% of observed threads are actually active
[ 3 / 4 ] CPU activity is below 90% (74.30%)
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% (89.01%), representing an hotspot for the application
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (94.91%)
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% (73.09%)
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 (0.02%)
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (3.67%) lower than cumulative innermost loop coverage (94.91%)
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)
| Loop ID | Analysis | Penalty Score |
|---|---|---|
| ►Loop 28 - kmeans-icpx-Ofast | Execution Time: 89 % - Vectorization Ratio: 19.15 % - Vector Length Use: 16.49 % | |
| ►Data Access Issues | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Inefficient Vectorization | 6 | |
| ○ | [SA] Presence of special instructions executing on a single port (INSERT/EXTRACT, BLEND/MERGE, SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 6 issues (= instructions) costing 1 point each. | 6 |
| ►Loop 26 - kmeans-icpx-Ofast | Execution Time: 3 % - Vectorization Ratio: 13.92 % - Vector Length Use: 12.75 % | |
| ►Loop Computation Issues | 2 | |
| ○ | [SA] Presence of a large number of scalar integer instructions - Simplify loop structure, perform loop splitting or perform unroll and jam. This issue costs 2 points. | 2 |
| ►Control Flow Issues | 6 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Data Access Issues | 9 | |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○ | [SA] More than 20% of the loads are accessing the stack - Perform loop splitting to decrease pressure on registers. This issue costs 2 points. | 2 |
| ►Vectorization Roadblocks | 12 | |
| ○ | [SA] Several paths (4 paths) - Simplify control structure or force the compiler to use masked instructions. There are 4 issues ( = paths) costing 1 point each. | 4 |
| ○ | [SA] Presence of constant non unit stride data access - Use array restructuring, perform loop interchange or use gather instructions to lower a bit the cost. There are 3 issues ( = data accesses) costing 2 point each. | 6 |
| ○ | [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points. | 2 |
| ►Inefficient Vectorization | 1 | |
| ○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 1 issues (= instructions) costing 1 point each. | 1 |
| ○Loop 27 - kmeans-icpx-Ofast | Execution Time: 2 % - Vectorization Ratio: 0.00 % - Vector Length Use: 11.61 % | |
| ►Loop 23 - kmeans-icpx-Ofast | Execution Time: 0 % - Vectorization Ratio: 100.00 % - Vector Length Use: 37.50 % | |
| ►Data Access Issues | 14 | |
| ○ | [SA] Presence of indirect accesses - Use array restructuring or gather instructions to lower the cost. There are 1 issues ( = indirect data accesses) costing 4 point each. | 4 |
| ○ | [SA] Presence of expensive instructions (GATHER/SCATTER) - Use array restructuring. There are 2 issues (= instructions) costing 4 points each. | 8 |
| ○ | [SA] Presence of special instructions executing on a single port (SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 2 issues (= instructions) costing 1 point each. | 2 |
| ►Vectorization Roadblocks | 4 | |
| ○ | [SA] Presence of indirect accesses - Use array restructuring or gather instructions to lower the cost. There are 1 issues ( = indirect data accesses) costing 4 point each. | 4 |
| ►Inefficient Vectorization | 10 | |
| ○ | [SA] Presence of expensive instructions (GATHER/SCATTER) - Use array restructuring. There are 2 issues (= instructions) costing 4 points each. | 8 |
| ○ | [SA] Presence of special instructions executing on a single port (SHUFFLE/PERM) - Simplify data access and try to get stride 1 access. There are 2 issues (= instructions) costing 1 point each. | 2 |