	*************************************************
	*                                               *
	*          ONE-View report generation           *
	*                                               *
	*************************************************

[MAQAO] Info: Experiment configuration summary is available adding -dbg=1 in command line

* [MAQAO] Warning: Experiment directory /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/run/oneview_runs/compilers/armclang_9/oneview_results_1781762087 already exists and is reused.
           It can be replaced using --replace in the command line.
[MAQAO] Info: 
[MAQAO] Info: START THE APPLICATION PROFILING
[MAQAO] Info: -> RUNNING THE PROFILER...
[MAQAO] Info:   LPROF has already been run
[MAQAO] Info: STOP THE APPLICATION PROFILING
[MAQAO] Info: 
[MAQAO] Info: START FUNCTIONS AND LOOPS ANALYSIS ...
[MAQAO] Info: STOP FUNCTIONS AND LOOPS ANALYSIS ...
[MAQAO] Info: 
[MAQAO] Info: START THE REPORT GENERATION
[MAQAO] Info: -> ONE-VIEW EXPERIMENT DIRECTORY: /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/run/oneview_runs/compilers/armclang_9/oneview_results_1781762087


+===================================================================================================================+
+                                                   1  -  GLOBAL                                                    +
+===================================================================================================================+


+-------------------------------------------------------------------------------------------------------------------+
+                                            1.1  -  Experiment Summary                                             +
+-------------------------------------------------------------------------------------------------------------------+

  Application:			/home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/run/binaries/armclang_9/exec
  Timestamp:			2026-06-18 05:54:47
  Universal Timestamp:		1781762087
  Experiment Type:		MPI; OpenMP; Throughput; 
  Machine:			ip-172-31-38-240.ec2.internal
  Architecture:			aarch64
  Micro Architecture:		ARM_NEOVERSE_V1
  OS Version:			Linux 6.1.170-213.321.amzn2023.aarch64 #1 SMP Thu May 14 12:18:13 UTC 2026
  Compilation Options:		
		exec: Arm Toolchain for Linux 22.1.0 clang version 22.1.0 (https://github.com/arm/arm-toolchain.git c95792353373404441df364b5a762338e5642230) /opt/arm/arm-toolchain-for-linux/bin/clang-22 -frtlib-add-rpath -fveclib=ArmPL -mllvm -gvn-add-phi-translation=1 -mllvm -store-to-load-forwarding-conflict-detection=0 -I /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp -I /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/armclang_9 -D DO_MPI -O2 -mcpu=neoverse-v1+nosve+nosve2 -I /opt/arm/arm-performance-libraries/lib/pkgconfig/../../include -L/opt/arm/arm-performance-libraries/lib/pkgconfig/../../lib -larmpl -lastring -lamath -lm -ffast-math -g -fno-omit-frame-pointer -fcf-protection=none -no-pie -grecord-command-line -fopenmp=libomp -MD -MT CMakeFiles/CoMD-openmp-mpi.dir/CoMD/src-openmp/ljForce.c.o -MF CMakeFiles/CoMD-openmp-mpi.dir/CoMD/src-openmp/ljForce.c.o.d -o CMakeFiles/CoMD-openmp-mpi.dir/CoMD/src-openmp/ljForce.c.o -c /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c -I /home/eoseret/tools/mpi/openmpi-armclang-22.1/include 
  Number of processes observed:	1
  Number of threads observed:	64
  MAQAO version:		2026.0.0
  MAQAO build:			Build information not available




+-------------------------------------------------------------------------------------------------------------------+
+                                              1.2  -  Global Metrics                                               +
+-------------------------------------------------------------------------------------------------------------------+

  Total Time:				19.98 s
  Max (Thread Active Time):		18.62 s
  Average Active Time:			18.08 s
  Activity Ratio:			90.7 %
  Average number of active threads:	57.933
  Affinity Stability:			99.9 %
  Time spent in analyzed loops:		61.2 %
  Time spent in analyzed innermost loops: 4.66 %
  Time spent in user code:		61.5 %
  Compilation Options Score:		100
  Array Access Efficiency:		87.8 %

   Potential Speedups
  ----------------------------------------------------
  Perfect Flow Complexity:		1.00
  Perfect OpenMP/MPI/Pthread/TBB:	1.24
  Perfect OpenMP/MPI/Pthread/TBB + Load Distribution:	1.66
  If No Scalar Integer:
      Potential Speedup:		1.17
      Nb Loops to get 80%:		1
  If FP Vectorized:
      Potential Speedup:		1.14
      Nb Loops to get 80%:		1
  If Fully Vectorized:
      Potential Speedup:		1.75
      Nb Loops to get 80%:		1
  If Only FP Arithmetic:
      Potential Speedup:		1.19
      Nb Loops to get 80%:		2




+-------------------------------------------------------------------------------------------------------------------+
+                                            1.3  -  Potential Speedups                                             +
+-------------------------------------------------------------------------------------------------------------------+

  If No Scalar Integer:
      Number of loops   | 1      | 4      | 9      | 12     | 17     | 
      Cumulated Speedup | 1.1688 | 1.1741 | 1.1744 | 1.1744 | 1.1744 | 
  Top 5 loops:
    exec - 100:	1.1688
    exec - 62:	1.1725
    exec - 91:	1.1739
    exec - 45:	1.1741
    exec - 107:	1.1743

  If FP Vectorized:
      Number of loops   | 1      | 4      | 9      | 12     | 17     | 
      Cumulated Speedup | 1.1324 | 1.1403 | 1.1406 | 1.1406 | 1.1406 | 
  Top 5 loops:
    exec - 100:	1.1324
    exec - 109:	1.1393
    exec - 111:	1.14
    exec - 114:	1.1403
    exec - 91:	1.1404

  If Fully Vectorized:
      Number of loops   | 1      | 4      | 9      | 12     | 17     | 
      Cumulated Speedup | 1.6447 | 1.7091 | 1.7416 | 1.7440 | 1.7456 | 
  Top 5 loops:
    exec - 100:	1.6447
    exec - 95:	1.6693
    exec - 62:	1.6931
    exec - 99:	1.7091
    exec - 111:	1.7233

  If Only FP Arithmetic:
      Number of loops   | 1      | 4      | 9      | 12     | 17     | 
      Cumulated Speedup | 1.1413 | 1.1702 | 1.1844 | 1.1857 | 1.1860 | 
  Top 5 loops:
    exec - 100:	1.1413
    exec - 62:	1.1529
    exec - 95:	1.1619
    exec - 109:	1.1702
    exec - 99:	1.1779



+===================================================================================================================+
+                                                   2  -  SUMMARY                                                   +
+===================================================================================================================+


+-------------------------------------------------------------------------------------------------------------------+
+                                            2.1  -  EXPERIMENT QUALITY                                             +
+-------------------------------------------------------------------------------------------------------------------+

  [4 / 4] Application profile is long enough (18.62 s)
To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds.

  [3 / 3] Most of time spent in analyzed modules comes from functions with source/debug info
-g option gives access to debugging informations, such are source locations.

  [2.9990523855176 / 3] Architecture specific option -mcpu is used


  [3 / 3] Most of time spent in analyzed modules comes from functions with compilation options informations and
-fno-omit-frame-pointer is present
-fno-omit-frame-pointer improves the accuracy of callchains found during the application profiling.

  [3 / 3] Optimization level option is correctly used


  [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.

  [2 / 2] Application is correctly profiled ("Others" category represents 0.27 % 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.



+-------------------------------------------------------------------------------------------------------------------+
+                                               2.2  -  CODE QUALITY                                                +
+-------------------------------------------------------------------------------------------------------------------+

  [4 / 4] Enough time of the experiment time spent in analyzed loops (61.22%)
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 90.52% of observed threads are actually active 

  [4 / 4] CPU activity is good
CPU cores are active 90.69% of time

  [4 / 4] Loop profile is not flat
At least one loop coverage is greater than 4% (55.70%), representing an hotspot for the application

  [0 / 4] Too little time of the experiment time spent in analyzed innermost loops (4.66%)
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 (99.90%)
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

  [0 / 3] Too many functions do not use all threads
Functions running on a reduced number of threads (typically sequential code) cover at least 10% of application
walltime (26.40%). Check both "Max Inclusive Time Over Threads" and "Nb Threads" in Functions or Loops tabs and
consider parallelizing sequential regions or improving parallelization of regions running on a reduced number of
threads

  [0 / 3] Cumulative Outermost/In between loops coverage (56.56%) greater than cumulative innermost loop coverage (4.66%)
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)



+-------------------------------------------------------------------------------------------------------------------+
+                                              2.3  -  LOOPS OVERVIEW                                               +
+-------------------------------------------------------------------------------------------------------------------+

  Top 5 loops:
   + exec - 100 :
     analysis: Execution Time: 55 % - Vectorization Ratio: 17.07 % - Vector Length Use: 28.96 %
     Loop Computation Issues: 6
        [4] [SA] Presence of expensive FP instructions - Perform hoisting, change algorithm, use SVML or proper
            numerical library or perform value profiling (count the number of distinct input values). There are 1
            issues (= instructions) costing 4 points each.
        [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.
     Control Flow Issues: 2
        [2] [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points.
     Vectorization Roadblocks: 1002
        [1000] [SA] Too many paths (at least 1000 paths) - Simplify control structure. There are at least 1000 issues (
            = paths) costing 1 point.
        [2] [SA] Non innermost loop (Outermost) - Collapse loop with innermost ones. This issue costs 2 points.

   + exec - 109 :
     analysis: Execution Time: 1 % - Vectorization Ratio: 80.00 % - Vector Length Use: 60.00 %

   + exec - 95  :
     analysis: Execution Time: 1 % - Vectorization Ratio: 0.00 % - Vector Length Use: 27.50 %
     Data Access Issues: 10
        [10] [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 5 issues ( = data accesses) costing 2
            point each.
     Vectorization Roadblocks: 10
        [10] [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 5 issues ( = data accesses) costing 2
            point each.

   + exec - 62  :
     analysis: Execution Time: 1 % - Vectorization Ratio: 0.00 % - Vector Length Use: 22.06 %
     Loop Computation Issues: 2
        [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.

   + exec - 111 :
     analysis: Execution Time: 0 % - Vectorization Ratio: 5.00 % - Vector Length Use: 25.63 %
     Loop Computation Issues: 4
        [4] [SA] Presence of expensive FP instructions - Perform hoisting, change algorithm, use SVML or proper
            numerical library or perform value profiling (count the number of distinct input values). There are 1
            issues (= instructions) costing 4 points each.
     Data Access Issues: 2
        [2] [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 1 issues ( = data accesses) costing 2
            point each.
     Vectorization Roadblocks: 2
        [2] [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 1 issues ( = data accesses) costing 2
            point each.



+===================================================================================================================+
+                                                 3  -  APPLICATION                                                 +
+===================================================================================================================+


+-------------------------------------------------------------------------------------------------------------------+
+                                              3.1  -  Categorization                                               +
+-------------------------------------------------------------------------------------------------------------------+

   Category | IO     | Exe    | Others  | TBB   | String | Pthread | MPI   | OMP   | System | Memory | Math  |
  ----------+--------+--------+---------+-------+--------+---------+-------+-------+--------+--------+-------+
   Time (%) | 0.00   | 61.55  | 0.27    | 0.00  | 0.00   | 0.00    | 0.01  | 38.13 | 0.00   | 0.03   | 0.00  |




+-------------------------------------------------------------------------------------------------------------------+
+                                         3.2  -  Function Based Profiling                                          +
+-------------------------------------------------------------------------------------------------------------------+

   Buckets                   | Nb Functions              | Coverage                  | Cumulated Coverage        |
  ---------------------------+---------------------------+---------------------------+---------------------------+
   > 8%                      | 2                         | 89.59                     | 89.59                     |
   4% to 8%                  | 0                         | 0.00                      | 89.59                     |
   2% to 4%                  | 1                         | 3.46                      | 93.05                     |
   1% to 2%                  | 4                         | 4.77                      | 97.82                     |
   0.5% to 1%                | 1                         | 0.79                      | 98.61                     |
   0.25% to 0.5%             | 1                         | 0.27                      | 98.88                     |
   0.125% to 0.25%           | 3                         | 0.52                      | 99.39                     |
   < 0.125%                  | 41                        | 0.61                      | 100.00                    |




+-------------------------------------------------------------------------------------------------------------------+
+                                           3.3  -  Loop Based Profiling                                            +
+-------------------------------------------------------------------------------------------------------------------+

   Buckets                   | Nb Loops                  | Coverage                  | Cumulated Coverage        |
  ---------------------------+---------------------------+---------------------------+---------------------------+
   > 8%                      | 0                         | 0.00                      | 0.00                      |
   4% to 8%                  | 0                         | 0.00                      | 0.00                      |
   2% to 4%                  | 0                         | 0.00                      | 0.00                      |
   1% to 2%                  | 3                         | 3.43                      | 3.43                      |
   0.5% to 1%                | 1                         | 0.79                      | 4.22                      |
   0.25% to 0.5%             | 0                         | 0.00                      | 4.22                      |
   0.125% to 0.25%           | 1                         | 0.17                      | 4.38                      |
   < 0.125%                  | 15                        | 0.28                      | 4.66                      |


+===================================================================================================================+
+                                                  4  -  FUNCTIONS                                                  +
+===================================================================================================================+


+-------------------------------------------------------------------------------------------------------------------+
+                                             4.1  -  Top 10 Functions                                              +
+-------------------------------------------------------------------------------------------------------------------+

   Function                                               | Module               | Coverage (%)  | Time (s)      |
  --------------------------------------------------------+----------------------+---------------+---------------+
   ljForce.omp_outlined.2                                 | exec                 | 56.54         | 10.23         |
   kmp_flag_64<false, true>::wait(kmp_info*, int, void*)  | libomp.so            | 33.04         | 5.98          |
   kmp_flag_native<unsigned long long, (flag_type)1, t... | libomp.so            | 3.46          | 0.63          |
   advanceVelocity.omp_outlined                           | exec                 | 1.30          | 0.24          |
   sortAtomsInCell                                        | exec                 | 1.18          | 0.21          |
   __GI___sched_yield                                     | libc.so.6            | 1.17          | 0.21          |
   ljForce.omp_outlined                                   | exec                 | 1.12          | 0.20          |
   advancePosition.omp_outlined                           | exec                 | 0.79          | 0.14          |
   unknown_function                                       | [vdso]               | 0.27          | 0.05          |
   msort_with_tmp.part.0                                  | libc.so.6            | 0.21          | 0.04          |


+===================================================================================================================+
+                                                    5  -  LOOPS                                                    +
+===================================================================================================================+


+-------------------------------------------------------------------------------------------------------------------+
+                                               5.1  -  Top 10 Loops                                                +
+-------------------------------------------------------------------------------------------------------------------+

   Loop Id        | Module               | Source Location                                       | Coverage (%)  |
  ----------------+----------------------+-------------------------------------------------------+---------------+
   100            | exec                 | ljForce.c:191-191,ljForce.c:197-198,ljForce.c:201-... | 55.70         |
   109            | exec                 | timestep.c:74-78                                      | 1.22          |
   95             | exec                 | mytype.h:23-23,ljForce.c:158-161                      | 1.12          |
   62             | exec                 | haloExchange.c:621-630                                | 1.09          |
   111            | exec                 | timestep.c:88-94                                      | 0.79          |
   99             | exec                 | ljForce.c:184-184,ljForce.c:187-187,ljForce.c:191-191 | 0.69          |
   91             | exec                 | linkCells.c:295-301,linkCells.c:352-365,linkCells.... | 0.17          |
   98             | exec                 | ljForce.c:175-175,ljForce.c:178-182,ljForce.c:187-187 | 0.15          |
   108            | exec                 | timestep.c:74-78                                      | 0.07          |
   45             | exec                 | haloExchange.c:380-389                                | 0.05          |





+===================================================================================================================+
+                                                     6  -  CQA                                                     +
+===================================================================================================================+


+-------------------------------------------------------------------------------------------------------------------+
+                                                   6.1  -  Loops                                                   +
+-------------------------------------------------------------------------------------------------------------------+





      6.1.1  -  Loop 100 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:173-191,197-216.

Analyzed code is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:191,197-198,201,206-210,213,216.

Warnings:
 - Non-innermost loop: analyzing only self part (ignoring child loops).
 - Ignoring paths for analysis
 - Failed to get the number of paths
 - RecMII not computed since number of paths is unknown or > max_paths
 - Streams not analyzed since number of paths is unknown or > max_paths


      6.1.1.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

13% of peak computational performance is used (4.44 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.1.1.1  -  Code clean check
  ---------------------------------------------------------------------------------------------------------

Detected a slowdown caused by scalar integer instructions (typically used for address computation).
By removing them, you can lower the cost of an iteration from 6.75 to 5.00 cycles (1.35x speedup).

Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.1.1.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 28% of vector register length is used (average across all VPU instructions).
By fully vectorizing your loop, you can lower the cost of an iteration from 6.75 to 2.00 cycles (3.38x speedup).

Details
17% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 27% of VPU loads are used in vector version.
 - 25% of VPU stores are used in vector version.
 - 14% of VPU addition or subtraction instructions are used in vector version.
 - 11% of VPU multiply instructions are used in vector version.
 - 20% of VPU fused multiply-add instructions are used in vector version.
 - 0% of VPU divide and square root instructions are used in vector version.
 - 0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.
Since your execution units are vector units, only a fully vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)





      6.1.1.1.3  -  FMA
  ---------------------------------------------------------------------------------------------------------

Detected 6 FMA (fused multiply-add) operations.
Presence of both ADD/SUB and MUL operations.

Workaround
Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible.
For instance a + b*c is a valid FMA (MUL then ADD).
However (a+b)* c cannot be translated into an FMA (ADD then MUL).




      6.1.1.1.4  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 30 FP arithmetical operations:
 - 13: addition or subtraction (6 inside FMA instructions)
 - 16: multiply (6 inside FMA instructions)
 - 1: divide
The binary loop does not load or store any data.







      6.1.2  -  Loop 109 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:74-78.

The related source loop is unrolled by 2 (including vectorization).

      6.1.2.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

18% of peak computational performance is used (6.00 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.2.1.1  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is partially vectorized.
Only 60% of vector register length is used (average across all VPU instructions).
By fully vectorizing your loop, you can lower the cost of an iteration from 2.00 to 1.50 cycles (1.33x speedup).

Details
80% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 33% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.
Since your execution units are vector units, only a fully vectorized loop can use their full power.





      6.1.2.1.2  -  FMA
  ---------------------------------------------------------------------------------------------------------

Detected 6 FMA (fused multiply-add) operations.




      6.1.2.1.3  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 12 FP arithmetical operations:
 - 6: addition or subtraction (all inside FMA instructions)
 - 6: multiply (all inside FMA instructions)
The binary loop does not load or store any data.







      6.1.3  -  Loop 95 from exec
  =========================================================================================================

The loop is defined in:
 - /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/mytype.h:23
 - /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:158-161


The related source loop is not unrolled or unrolled with no peel/tail loop.

      6.1.3.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

0% of peak computational performance is used (0.00 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.3.1.1  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 27% of vector register length is used (average across all VPU instructions).
By vectorizing your loop, you can lower the cost of an iteration from 2.50 to 0.50 cycles (5.00x speedup).

Details
All VPU instructions are used in scalar version (process only one data element in vector registers).
Since your execution units are vector units, only a vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)




No data for this section



      6.1.3.1.2  -  Slow data structures access
  ---------------------------------------------------------------------------------------------------------

Detected data structures (typically arrays) that cannot be efficiently read/written

Details
 - Constant unknown stride: 4 occurrence(s)
 - Constant non-unit stride: 1 occurrence(s)
Non-unit stride (uncontiguous) accesses are not efficiently using data caches


Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.3.1.3  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop does not contain any FP arithmetical operations.
The binary loop does not load or store any data.







      6.1.4  -  Loop 62 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/haloExchange.c:621-630.

The related source loop is not unrolled or unrolled with no peel/tail loop.

      6.1.4.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

0% of peak computational performance is used (0.00 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.4.1.1  -  Code clean check
  ---------------------------------------------------------------------------------------------------------

Detected a slowdown caused by scalar integer instructions (typically used for address computation).
By removing them, you can lower the cost of an iteration from 5.33 to 4.00 cycles (1.33x speedup).

Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.4.1.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 22% of vector register length is used (average across all VPU instructions).
By vectorizing your loop, you can lower the cost of an iteration from 5.33 to 1.21 cycles (4.41x speedup).

Details
All VPU instructions are used in scalar version (process only one data element in vector registers).
Since your execution units are vector units, only a vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)




No data for this section



      6.1.4.1.3  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop does not contain any FP arithmetical operations.
The binary loop is loading 6 bytes.
The binary loop is storing 6 bytes.







      6.1.5  -  Loop 111 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:88-94.

The related source loop is not unrolled or unrolled with no peel/tail loop.

      6.1.5.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

6% of peak computational performance is used (2.22 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.5.1.1  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 25% of vector register length is used (average across all VPU instructions).
By fully vectorizing your loop, you can lower the cost of an iteration from 4.50 to 1.75 cycles (2.57x speedup).

Details
5% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 11% of VPU loads are used in vector version.
 - 0% of VPU stores are used in vector version.
 - 0% of VPU addition or subtraction instructions are used in vector version.
 - 0% of VPU multiply instructions are used in vector version.
 - 0% of VPU fused multiply-add instructions are used in vector version.
 - 0% of VPU divide and square root instructions are used in vector version.
Since your execution units are vector units, only a fully vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)





      6.1.5.1.2  -  FMA
  ---------------------------------------------------------------------------------------------------------

Detected 3 FMA (fused multiply-add) operations.




      6.1.5.1.3  -  Slow data structures access
  ---------------------------------------------------------------------------------------------------------

Detected data structures (typically arrays) that cannot be efficiently read/written

Details
 - Constant unknown stride: 1 occurrence(s)
Non-unit stride (uncontiguous) accesses are not efficiently using data caches


Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.5.1.4  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 10 FP arithmetical operations:
 - 3: addition or subtraction (all inside FMA instructions)
 - 6: multiply (3 inside FMA instructions)
 - 1: divide
The binary loop is loading 3 bytes.


      6.1.5.1.5  -  Arithmetic intensity
  ---------------------------------------------------------------------------------------------------------

Arithmetic intensity is 3.33 FP operations per loaded or stored byte.







      6.1.6  -  Loop 99 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:173-191.

Analyzed code is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:184,187,191.

Warnings:
 - Non-innermost loop: analyzing only self part (ignoring child loops).
 - Ignoring paths for analysis
 - Failed to get the number of paths
 - RecMII not computed since number of paths is unknown or > max_paths
 - Streams not analyzed since number of paths is unknown or > max_paths


      6.1.6.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

0% of peak computational performance is used (0.00 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.6.1.1  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 22% of vector register length is used (average across all VPU instructions).
By vectorizing your loop, you can lower the cost of an iteration from 1.75 to 0.34 cycles (5.09x speedup).

Details
All VPU instructions are used in scalar version (process only one data element in vector registers).
Since your execution units are vector units, only a vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)




No data for this section



      6.1.6.1.2  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop does not contain any FP arithmetical operations.
The binary loop is loading 3 bytes.







      6.1.7  -  Loop 91 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/linkCells.c:295-301,352-365,371,378.

The related source loop is not unrolled or unrolled with no peel/tail loop.
This loop has 4 execution paths.

The presence of multiple execution paths is typically the main/first bottleneck.
Try to simplify control inside loop: ideally, try to remove all conditional expressions, for example by (if applicable):
 - hoisting them (moving them outside the loop)
 - turning them into conditional moves, MIN or MAX


Ex: if (x<0) x=0 => x = (x<0 ? 0 : x) (or MAX(0,x) after defining the corresponding macro)


      6.1.7.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

Warnings:
Detected a function call instruction: ignoring called function instructions.
Rerun with --follow-calls=append to include them to analysis  or with --follow-calls=inline to simulate inlining.
2% of peak computational performance is used (0.91 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.7.1.1  -  Code clean check
  ---------------------------------------------------------------------------------------------------------

Detected a slowdown caused by scalar integer instructions (typically used for address computation).
By removing them, you can lower the cost of an iteration from 6.63 to 2.67 cycles (2.48x speedup).

Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.1.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 23% of vector register length is used (average across all VPU instructions).
By fully vectorizing your loop, you can lower the cost of an iteration from 6.63 to 1.50 cycles (4.42x speedup).

Details
11% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 44% of VPU loads are used in vector version.
 - 0% of VPU addition or subtraction instructions are used in vector version.
 - 0% of VPU multiply instructions are used in vector version.
 - 0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.
Since your execution units are vector units, only a fully vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)





      6.1.7.1.3  -  FMA
  ---------------------------------------------------------------------------------------------------------

Presence of both ADD/SUB and MUL operations.

Workaround
Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible.
For instance a + b*c is a valid FMA (MUL then ADD).
However (a+b)* c cannot be translated into an FMA (ADD then MUL).




      6.1.7.1.4  -  CALL instructions
  ---------------------------------------------------------------------------------------------------------

Detected function call instructions.


Details
Calling (and then returning from) a function prevents many compiler optimizations (like vectorization), breaks control flow (which reduces pipeline performance) and executes extra instructions to save/restore the registers used inside it, which is very expensive (dozens of cycles). Consider to inline small functions.
 - getBoxFromTuple: 1 occurrences<<list_path_1_call_1>>
 - moveAtom: 1 occurrences<<list_path_1_call_2>>



      6.1.7.1.5  -  Slow data structures access
  ---------------------------------------------------------------------------------------------------------

Detected data structures (typically arrays) that cannot be efficiently read/written

Details
 - Constant unknown stride: 2 occurrence(s)
Non-unit stride (uncontiguous) accesses are not efficiently using data caches


Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.1.6  -  Conversion instructions
  ---------------------------------------------------------------------------------------------------------

Detected expensive conversion instructions.

Details
 - FCVTMS: 3 occurrences<<list_path_1_cvt_1>>


Workaround
Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.


      6.1.7.1.7  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 6 FP arithmetical operations:
 - 3: addition or subtraction
 - 3: multiply
The binary loop is loading 6 bytes.


      6.1.7.1.8  -  Arithmetic intensity
  ---------------------------------------------------------------------------------------------------------

Arithmetic intensity is 1.00 FP operations per loaded or stored byte.




      6.1.7.2  -  Path 2
  ---------------------------------------------------------------------------------------------------------

Warnings:
Detected a function call instruction: ignoring called function instructions.
Rerun with --follow-calls=append to include them to analysis  or with --follow-calls=inline to simulate inlining.
3% of peak computational performance is used (1.00 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.7.2.1  -  Code clean check
  ---------------------------------------------------------------------------------------------------------

Detected a slowdown caused by scalar integer instructions (typically used for address computation).
By removing them, you can lower the cost of an iteration from 6.00 to 2.67 cycles (2.25x speedup).

Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.2.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 23% of vector register length is used (average across all VPU instructions).
By fully vectorizing your loop, you can lower the cost of an iteration from 6.00 to 1.50 cycles (4.00x speedup).

Details
12% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 44% of VPU loads are used in vector version.
 - 0% of VPU addition or subtraction instructions are used in vector version.
 - 0% of VPU multiply instructions are used in vector version.
 - 0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.
Since your execution units are vector units, only a fully vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)





      6.1.7.2.3  -  FMA
  ---------------------------------------------------------------------------------------------------------

Presence of both ADD/SUB and MUL operations.

Workaround
Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible.
For instance a + b*c is a valid FMA (MUL then ADD).
However (a+b)* c cannot be translated into an FMA (ADD then MUL).




      6.1.7.2.4  -  CALL instructions
  ---------------------------------------------------------------------------------------------------------

Detected function call instructions.


Details
Calling (and then returning from) a function prevents many compiler optimizations (like vectorization), breaks control flow (which reduces pipeline performance) and executes extra instructions to save/restore the registers used inside it, which is very expensive (dozens of cycles). Consider to inline small functions.
 - getBoxFromTuple: 1 occurrences<<list_path_2_call_1>>



      6.1.7.2.5  -  Slow data structures access
  ---------------------------------------------------------------------------------------------------------

Detected data structures (typically arrays) that cannot be efficiently read/written

Details
 - Constant unknown stride: 2 occurrence(s)
Non-unit stride (uncontiguous) accesses are not efficiently using data caches


Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.2.6  -  Conversion instructions
  ---------------------------------------------------------------------------------------------------------

Detected expensive conversion instructions.

Details
 - FCVTMS: 3 occurrences<<list_path_2_cvt_1>>


Workaround
Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.


      6.1.7.2.7  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 6 FP arithmetical operations:
 - 3: addition or subtraction
 - 3: multiply
The binary loop is loading 6 bytes.


      6.1.7.2.8  -  Arithmetic intensity
  ---------------------------------------------------------------------------------------------------------

Arithmetic intensity is 1.00 FP operations per loaded or stored byte.




      6.1.7.3  -  Path 3
  ---------------------------------------------------------------------------------------------------------

Warnings:
Detected a function call instruction: ignoring called function instructions.
Rerun with --follow-calls=append to include them to analysis  or with --follow-calls=inline to simulate inlining.
2% of peak computational performance is used (0.73 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.7.3.1  -  Code clean check
  ---------------------------------------------------------------------------------------------------------

Detected a slowdown caused by scalar integer instructions (typically used for address computation).
By removing them, you can lower the cost of an iteration from 5.50 to 2.00 cycles (2.75x speedup).

Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.3.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 24% of vector register length is used (average across all VPU instructions).
By fully vectorizing your loop, you can lower the cost of an iteration from 5.50 to 1.25 cycles (4.40x speedup).

Details
15% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 57% of VPU loads are used in vector version.
 - 0% of VPU addition or subtraction instructions are used in vector version.
 - 0% of VPU multiply instructions are used in vector version.
 - 0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.
Since your execution units are vector units, only a fully vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)





      6.1.7.3.3  -  FMA
  ---------------------------------------------------------------------------------------------------------

Presence of both ADD/SUB and MUL operations.

Workaround
Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible.
For instance a + b*c is a valid FMA (MUL then ADD).
However (a+b)* c cannot be translated into an FMA (ADD then MUL).




      6.1.7.3.4  -  CALL instructions
  ---------------------------------------------------------------------------------------------------------

Detected function call instructions.


Details
Calling (and then returning from) a function prevents many compiler optimizations (like vectorization), breaks control flow (which reduces pipeline performance) and executes extra instructions to save/restore the registers used inside it, which is very expensive (dozens of cycles). Consider to inline small functions.
 - getBoxFromTuple: 1 occurrences<<list_path_3_call_1>>
 - moveAtom: 1 occurrences<<list_path_3_call_2>>



      6.1.7.3.5  -  Slow data structures access
  ---------------------------------------------------------------------------------------------------------

Detected data structures (typically arrays) that cannot be efficiently read/written

Details
 - Constant unknown stride: 2 occurrence(s)
Non-unit stride (uncontiguous) accesses are not efficiently using data caches


Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.3.6  -  Conversion instructions
  ---------------------------------------------------------------------------------------------------------

Detected expensive conversion instructions.

Details
 - FCVTMS: 2 occurrences<<list_path_3_cvt_1>>


Workaround
Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.


      6.1.7.3.7  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 4 FP arithmetical operations:
 - 2: addition or subtraction
 - 2: multiply
The binary loop is loading 6 bytes.


      6.1.7.3.8  -  Arithmetic intensity
  ---------------------------------------------------------------------------------------------------------

Arithmetic intensity is 0.67 FP operations per loaded or stored byte.




      6.1.7.4  -  Path 4
  ---------------------------------------------------------------------------------------------------------

Warnings:
Detected a function call instruction: ignoring called function instructions.
Rerun with --follow-calls=append to include them to analysis  or with --follow-calls=inline to simulate inlining.
2% of peak computational performance is used (0.80 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.7.4.1  -  Code clean check
  ---------------------------------------------------------------------------------------------------------

Detected a slowdown caused by scalar integer instructions (typically used for address computation).
By removing them, you can lower the cost of an iteration from 5.00 to 2.00 cycles (2.50x speedup).

Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.4.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 25% of vector register length is used (average across all VPU instructions).
By fully vectorizing your loop, you can lower the cost of an iteration from 5.00 to 1.25 cycles (4.00x speedup).

Details
17% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 57% of VPU loads are used in vector version.
 - 0% of VPU addition or subtraction instructions are used in vector version.
 - 0% of VPU multiply instructions are used in vector version.
 - 0% of VPU instructions that are not load, store, addition, subtraction nor multiply instructions are used in vector version.
Since your execution units are vector units, only a fully vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)





      6.1.7.4.3  -  FMA
  ---------------------------------------------------------------------------------------------------------

Presence of both ADD/SUB and MUL operations.

Workaround
Try to change order in which elements are evaluated (using parentheses) in arithmetic expressions containing both ADD/SUB and MUL operations to enable your compiler to generate FMA instructions wherever possible.
For instance a + b*c is a valid FMA (MUL then ADD).
However (a+b)* c cannot be translated into an FMA (ADD then MUL).




      6.1.7.4.4  -  CALL instructions
  ---------------------------------------------------------------------------------------------------------

Detected function call instructions.


Details
Calling (and then returning from) a function prevents many compiler optimizations (like vectorization), breaks control flow (which reduces pipeline performance) and executes extra instructions to save/restore the registers used inside it, which is very expensive (dozens of cycles). Consider to inline small functions.
 - getBoxFromTuple: 1 occurrences<<list_path_4_call_1>>



      6.1.7.4.5  -  Slow data structures access
  ---------------------------------------------------------------------------------------------------------

Detected data structures (typically arrays) that cannot be efficiently read/written

Details
 - Constant unknown stride: 2 occurrence(s)
Non-unit stride (uncontiguous) accesses are not efficiently using data caches


Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.7.4.6  -  Conversion instructions
  ---------------------------------------------------------------------------------------------------------

Detected expensive conversion instructions.

Details
 - FCVTMS: 2 occurrences<<list_path_4_cvt_1>>


Workaround
Avoid mixing data with different types. In particular, check if the type of constants is the same as array elements.


      6.1.7.4.7  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 4 FP arithmetical operations:
 - 2: addition or subtraction
 - 2: multiply
The binary loop is loading 6 bytes.


      6.1.7.4.8  -  Arithmetic intensity
  ---------------------------------------------------------------------------------------------------------

Arithmetic intensity is 0.67 FP operations per loaded or stored byte.







      6.1.8  -  Loop 98 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:173-187.

Analyzed code is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/ljForce.c:175,178-182,187.

Warnings:
 - Non-innermost loop: analyzing only self part (ignoring child loops).
 - Ignoring paths for analysis
 - Failed to get the number of paths
 - RecMII not computed since number of paths is unknown or > max_paths
 - Streams not analyzed since number of paths is unknown or > max_paths


      6.1.8.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

0% of peak computational performance is used (0.00 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.8.1.1  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 20% of vector register length is used (average across all VPU instructions).
By vectorizing your loop, you can lower the cost of an iteration from 2.00 to 0.25 cycles (8.00x speedup).

Details
All VPU instructions are used in scalar version (process only one data element in vector registers).
Since your execution units are vector units, only a vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)




No data for this section



      6.1.8.1.2  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop does not contain any FP arithmetical operations.
The binary loop is loading 6 bytes.







      6.1.9  -  Loop 108 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/timestep.c:74-78.

The related source loop is unrolled by 2 (including vectorization).

      6.1.9.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

4% of peak computational performance is used (1.57 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.9.1.1  -  Unrolling/vectorization cost
  ---------------------------------------------------------------------------------------------------------

This loop is peel/tail of a unrolled/vectorized loop. If its cost is not negligible compared to the main (unrolled/vectorized) loop, unrolling/vectorization is counterproductive due to low trip count.

Details
The more iterations the main loop is processing, the higher the trip count must be to amortize peel/tail overhead.

Workaround
 - recompile with -fprofile-instr-generate, execute ,merge raw profiles with 'llvm-profdata merge -o default.profdata default.profraw' and recompile with -fprofile-instr-use (profile-guided optimization)
 - hardcode most frequent values of loop bounds by adding specialized paths.:
  *  For instance, replace for (i=0; i<n; i++) foo(i) with:
switch (n) {
  case (4): for (i=0; i<4; i++) foo(i); break;
  case (6): for (i=0; i<6; i++) foo(i); break;
  default : for (i=0; i<n; i++) foo(i); break;
}



      6.1.9.1.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 26% of vector register length is used (average across all VPU instructions).


Details
7% of VPU instructions are used in vector version (process two or more data elements in vector registers):
 - 12% of VPU loads are used in vector version.
 - 0% of VPU stores are used in vector version.
 - 0% of VPU fused multiply-add instructions are used in vector version.





      6.1.9.1.3  -  FMA
  ---------------------------------------------------------------------------------------------------------

Detected 3 FMA (fused multiply-add) operations.




      6.1.9.1.4  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 6 FP arithmetical operations:
 - 3: addition or subtraction (all inside FMA instructions)
 - 3: multiply (all inside FMA instructions)
The binary loop does not load or store any data.







      6.1.10  -  Loop 45 from exec
  =========================================================================================================

The loop is defined in /home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/build/CoMD/CoMD/src-openmp/haloExchange.c:380-389.

The related source loop is not unrolled or unrolled with no peel/tail loop.

      6.1.10.1  -  Path 1
  ---------------------------------------------------------------------------------------------------------

1% of peak computational performance is used (0.53 out of 32.00 FLOP per cycle (GFLOPS @ 1GHz))

      6.1.10.1.1  -  Code clean check
  ---------------------------------------------------------------------------------------------------------

Detected a slowdown caused by scalar integer instructions (typically used for address computation).
By removing them, you can lower the cost of an iteration from 5.67 to 4.00 cycles (1.42x speedup).

Workaround
 - Try to reorganize arrays of structures to structures of arrays
 - Consider to permute loops (see vectorization gain report)



      6.1.10.1.2  -  Vectorization
  ---------------------------------------------------------------------------------------------------------

Your loop is not vectorized.
Only 22% of vector register length is used (average across all VPU instructions).
By vectorizing your loop, you can lower the cost of an iteration from 5.67 to 1.21 cycles (4.69x speedup).

Details
All VPU instructions are used in scalar version (process only one data element in vector registers).
Since your execution units are vector units, only a vectorized loop can use their full power.


Workaround
 - Try another compiler or update/tune your current one
 - Remove inter-iterations dependences from your loop and make it unit-stride:
  * If your arrays have 2 or more dimensions, check whether elements are accessed contiguously and, otherwise, try to permute loops accordingly:
C storage order is row-major: for(i) for(j) a[j][i] = b[j][i]; (slow, non stride 1) => for(i) for(j) a[i][j] = b[i][j]; (fast, stride 1)<<image_row_maj>>
  * If your loop streams arrays of structures (AoS), try to use structures of arrays instead (SoA):
for(i) a[i].x = b[i].x; (slow, non stride 1) => for(i) a.x[i] = b.x[i]; (fast, stride 1)




No data for this section



      6.1.10.1.3  -  Matching between your loop (in the source code) and the binary loop
  ---------------------------------------------------------------------------------------------------------

The binary loop is composed of 3 FP arithmetical operations:
 - 3: addition or subtraction
The binary loop is loading 9 bytes.
The binary loop is storing 6 bytes.


      6.1.10.1.4  -  Arithmetic intensity
  ---------------------------------------------------------------------------------------------------------

Arithmetic intensity is 0.20 FP operations per loaded or stored byte.





[MAQAO] Info: STOP THE REPORT GENERATION
[MAQAO] Info: 
[MAQAO] Info: If your application produces files, they can be found in directory "/home/eoseret/qaas/qaas_runs/178-176-0594/intel/CoMD/run/oneview_runs/compilers/armclang_9/oneview_run_1781762087"
[MAQAO] Info: 
