Intel® Advisor Help
This column enables you to view the type of the loop/function. Provides the following data when expanded:
Peeled/Remainder metric for loops that have child loops. The metric appears only when scalar peel and/or remainder executed.
Threadedmetric for loops that have child loops. The metric appears when some parallel framework (Intel® Cilk Plus, OpenMP, or automatically by Intel® Compiler) is used in the loop.
Vectorized (<loop part(s)>) metric for vectorized parent and child loops. The metric appears when a parent loop has any of the following parts executed: peeled, body, and/or remainder.
Also appears for child loops that have one of the loop parts executed (peeled, body, and remainder).
Peeled metric appears for small, (usually) compiler-generated loops created to align the memory accesses inside the loop body and maximize its efficiency.
Body metric appears for vectorized loops (compiler-generated from a source loop). Most loop iterations should execute in body, as body normally processes more data than peel or remainder. Vector length in body is usually bigger than in peeled and/or remainder, which means body is the most efficient place performance-wise.
Remainder metric appears for (usually) compiler-generated loops created to clean up any remaining iterations that do not fit within the scope of the loop body.
[Not Executed] mark appears next to any other loop metric when a loop wasn't executed.
Scalar metric appears when non-vector loops were executed.
Completely Unrolled metric appears when the loop body was copied several times (equal to trip counts value) by the compiler.
Inside vectorized metric appears when the inner loop was vectorized along with the outer loop.
Inlined Function metric appears when the function body was inlined into the loop/function body.
Vector Function metric appears when SIMD-enabled version of function was executed. (See Intel® Compiler documentation for details).
Function metric appears when scalar version of the function was executed.