Help is available by moving the cursor above any symbol or by checking MAQAO website.
[ 0 / 9 ] Compilation options are not available
Compilation options are an important optimization leverage but ONE-View is not able to analyze them.
[ 4 / 4 ] Application profile is long enough (11.06 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 0 % 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
[ 4 / 4 ] Enough time of the experiment time spent in analyzed loops (100.01%)
If the time spent in analyzed loops is less than 30%, standard loop optimizations will have a limited impact on application performances.
[ 4 / 4 ] Loop profile is not flat
At least one loop coverage is greater than 4% (98.37%), representing an hotspot for the application
[ 4 / 4 ] Enough time of the experiment time spent in analyzed innermost loops (98.42%)
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%) is spend in BLAS1 operations
It could be more efficient to inline by hand BLAS1 operations
[ 3 / 3 ] Cumulative Outermost/In between loops coverage (1.59%) lower than cumulative innermost loop coverage (98.42%)
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%) is spend in Libm/SVML (special functions)
[ 2 / 2 ] Less than 10% (0%) is spend in BLAS2 operations
BLAS2 calls usually could make a poor cache usage and could benefit from inlining.
Loop ID | Module | Analysis | Penalty Score | Coverage (%) | Vectorization Ratio (%) | Vector Length Use (%) |
---|---|---|---|---|---|---|
►4 | convf32_avx512 | Partial or unexisting vectorization - Use pragma to force vectorization and check potential dependencies between array access. | 30 | 98.37 | 92.97 | 93.51 |
○ | [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. | 10 | ||||
○ | [SA] Presence of special instructions executing on a single port (BROADCAST) - Simplify data access and try to get stride 1 access. There are 7 issues (= instructions) costing 1 point each. | 7 | ||||
○ | [DA] Low iteration count (7 < 10) - Perform full unroll. Use compiler pragmas. Use PGO/FDO compiler options. Force compiler to use masked instructions. This issue costs 5 points. | 5 | ||||
○ | [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 a large number of scalar integer instructions - Simplify loop structure, perform loop splitting or perform unroll and jam. This issue costs 2 points. | 2 | ||||
○ | [SA] Inefficient vectorization: use of shorter than available vector length - Force compiler to use proper vector length. CAUTION: use of 512 bits vectors could be more expensive than 256 bits on some processors. Use intrinsics (costly and not portable). The issue costs 2 points. | 2 | ||||
○ | [DA] The ratio FP/LS (floating point / memory accesses) is smaller than 0.8 (0.76) - Focus on optimizing data accesses. | 0 | ||||
►3 | convf32_avx512 | Partial or unexisting vectorization - Use pragma to force vectorization and check potential dependencies between array access. | 12 | 1.36 | 34.78 | 40.17 |
○ | [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 4 issues ( = data accesses) costing 2 point each. | 8 | ||||
○ | [SA] Several paths (2 paths) - Simplify control structure or force the compiler to use masked instructions. There are 2 issues ( = paths) costing 1 point each. | 2 | ||||
○ | [SA] Non innermost loop (InBetween) - Collapse loop with innermost ones. This issue costs 2 points. | 2 | ||||
►2 | convf32_avx512 | Partial or unexisting vectorization - Use pragma to force vectorization and check potential dependencies between array access. | 18 | 0.23 | 0 | 7.81 |
○ | [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 7 issues ( = data accesses) costing 2 point each. | 14 | ||||
○ | [SA] Several paths (2 paths) - Simplify control structure or force the compiler to use masked instructions. There are 2 issues ( = paths) costing 1 point each. | 2 | ||||
○ | [SA] Non innermost loop (InBetween) - Collapse loop with innermost ones. This issue costs 2 points. | 2 | ||||
►6 | convf32_avx512 | Partial or unexisting vectorization - Use pragma to force vectorization and check potential dependencies between array access. | 1 | 0.05 | 0 | 10.94 |
○ | [SA] Presence of calls - Inline either by compiler or by hand and use SVML for libm calls. There are 1 issues (= calls) costing 1 point each. | 1 |