avbp-06may_100iter_m36_ov1_gcc-o3-funroll | avbp-06may_100iter_m36_ov2 |
---|---|
[ 3 / 3 ] Host configuration allows retrieval of all necessary metrics. | [ 3 / 3 ] Host configuration allows retrieval of all necessary metrics. |
[ 0 / 0 ] Fastmath not used Consider to add ffast-math to compilation flags (or replace -O3 with -Ofast) to unlock potential extra speedup by relaxing floating-point computation consistency. Warning: floating-point accuracy may be reduced and the compliance to IEEE/ISO rules/specifications for math functions will be relaxed, typically 'errno' will no longer be set after calling some math functions. | Not available for this run |
[ 0 / 3 ] Compilation of some functions is not optimized for the target processor -march=x86-64 option is used but it is not specific enough to produce efficient code. Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ). | [ 0 / 3 ] Compilation of some functions is not optimized for the target processor Application run on the SKYLAKE micro-architecture while the code was specialized for CORE-AVX2. Architecture specific options are needed to produce efficient code for a specific processor ( -x(target) or -ax(target) ). |
[ 2.35 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information Functions without compilation information (typically not compiled with -g) cumulate 2.15% of the time spent in analyzed modules. Check that -g is present. Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case. | [ 2.06 / 3 ] Most of time spent in analyzed modules comes from functions without compilation information Functions without compilation information (typically not compiled with -g) cumulate 13.98% of the time spent in analyzed modules. Check that -g is present. Remark: if -g is indeed used, this can also be due to some compiler built-in functions (typically math) or statically linked libraries. This warning can be ignored in that case. |
[ 4 / 4 ] Application profile is long enough (355.64 s) To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds. | [ 4 / 4 ] Application profile is long enough (336.05 s) To have good quality measurements, it is advised that the application profiling time is greater than 10 seconds. |
[ 2 / 2 ] Application is correctly profiled ("Others" category represents 1.10 % 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 | [ 2 / 2 ] Application is correctly profiled ("Others" category represents 3.68 % 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 |
[ 3 / 3 ] Optimization level option is correctly used | [ 2.64 / 3 ] Optimization level option is correctly used |
[ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated. | [ 1 / 1 ] Lstopo present. The Topology lstopo report will be generated. |
avbp-06may_100iter_m36_ov1_gcc-o3-funroll | avbp-06may_100iter_m36_ov2 |
---|---|
[ 3 / 4 ] CPU activity is below 90% (87.64%) 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 ] CPU activity is good CPU cores are active 99.87% of time |
[ 3 / 4 ] Affinity stability is lower than 90% (88.31%) Threads are often migrating to other CPU cores/threads. For OpenMP, typically set (OMP_PLACES=cores OMP_PROC_BIND=close) or (OMP_PLACES=threads OMP_PROC_BIND=spread). With OpenMPI + OpenMP, use --bind-to core --map-by node:PE=$OMP_NUM_THREADS --report-bindings. With IntelMPI + OpenMP, set I_MPI_PIN_DOMAIN=omp:compact or I_MPI_PIN_DOMAIN=omp:scatter and use -print-rank-map. | [ 4 / 4 ] Affinity is good (100.00%) Threads are not migrating to CPU cores: probably successfully pinned |
[ 4 / 4 ] Enough time of the experiment time spent in analyzed loops (76.35%) If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances. | [ 4 / 4 ] Enough time of the experiment time spent in analyzed loops (72.88%) If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances. |
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (34.91%) lower than cumulative innermost loop coverage (41.44%) Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex | [ 3 / 3 ] Cumulative Outermost/In between loops coverage (23.44%) lower than cumulative innermost loop coverage (49.44%) Having cumulative Outermost/In between loops coverage greater than cumulative innermost loop coverage will make loop optimization more complex |
[ 3 / 4 ] A significant amount of threads are idle (12.38%) On average, more than 10% of observed threads are idle. Such threads are probably IO/sync waiting. Some hints: use faster filesystems to read/write data, improve parallel load balancing and/or scheduling. | [ 4 / 4 ] Threads activity is good On average, more than 99.88% of observed threads are actually active |
[ 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 BLAS2 operations BLAS2 calls usually could make a poor cache usage and could benefit from inlining. |
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (41.44%) 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 ] Enough time of the experiment time spent in analyzed innermost loops (49.44%) 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 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations It could be more efficient to inline by hand BLAS1 operations | [ 3 / 3 ] Less than 10% (0.00%) is spend in BLAS1 operations It could be more efficient to inline by hand BLAS1 operations |
[ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions) | [ 2 / 2 ] Less than 10% (0.00%) is spend in Libm/SVML (special functions) |
[ 4 / 4 ] Loop profile is not flat At least one loop coverage is greater than 4% (4.03%), representing an hotspot for the application | [ 4 / 4 ] Loop profile is not flat At least one loop coverage is greater than 4% (4.54%), representing an hotspot for the application |
Analysis | r0 | r1 | |
---|---|---|---|
Loop Computation Issues | Less than 10% of the FP ADD/SUB/MUL arithmetic operations are performed using FMA | 6 | 1 |
Presence of a large number of scalar integer instructions | 5 | 8 | |
Low iteration count | 0 | 6 | |
Control Flow Issues | Presence of more than 4 paths | 4 | 3 |
Non-innermost loop | 5 | 3 | |
Low iteration count | 0 | 6 | |
Data Access Issues | Presence of constant non-unit stride data access | 4 | 2 |
Presence of indirect access | 2 | 3 | |
More than 10% of the vector loads instructions are unaligned | 8 | 0 | |
Presence of special instructions executing on a single port | 6 | 4 | |
More than 20% of the loads are accessing the stack | 5 | 3 | |
Vectorization Roadblocks | Presence of more than 4 paths | 5 | 3 |
Non-innermost loop | 5 | 3 | |
Presence of constant non-unit stride data access | 4 | 2 | |
Presence of indirect access | 2 | 3 | |
Out of user code | 1 | 1 | |
Inefficient Vectorization | Presence of special instructions executing on a single port | 6 | 4 |