How to understand the speedup in optimization report from icc compiler?

Question

environment is:
icc version 19.0.0.117 (gcc version 5.4.0 compatibility)
Intel parallel studio XE cluster edition 2019
Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz
Ubuntu 16.04

compiler flags are:
-std=gnu11 -Wall -xHost -xCORE-AVX2 -O2 -fma -qopenmp -qopenmp-simd -qopt-report=5 -qopt-report-phase=all

I use OpenMP simd or intel parama to vectorize my loop to gain the speedup. In the optimization report generated by icc, I usually see the following result:

LOOP BEGIN at get_forces.c(3668,3)
   remark #15389: vectorization support: reference mon->fricforce[n1][d] has unaligned access   [ get_forces.c(3669,4) ]
   remark #15389: vectorization support: reference mon->vel[n1][d] has unaligned access   [ get_forces.c(3669,36) ]
   remark #15389: vectorization support: reference vel[n1][d] has unaligned access   [ get_forces.c(3669,51) ]
   remark #15389: vectorization support: reference mon->drag[n1][d] has unaligned access   [ get_forces.c(3671,4) ]
   remark #15389: vectorization support: reference mon->vel[n1][d] has unaligned access   [ get_forces.c(3671,40) ]
   remark #15389: vectorization support: reference vel[n1][d] has unaligned access   [ get_forces.c(3671,57) ]
   remark #15381: vectorization support: unaligned access used inside loop body
   remark #15305: vectorization support: vector length 2
   remark #15309: vectorization support: normalized vectorization overhead 0.773
   remark #15300: LOOP WAS VECTORIZED
   remark #15450: unmasked unaligned unit stride loads: 3 
   remark #15451: unmasked unaligned unit stride stores: 2 
   remark #15475: --- begin vector cost summary ---
   remark #15476: scalar cost: 21 
   remark #15477: vector cost: 11.000 
   remark #15478: estimated potential speedup: 1.050 
   remark #15488: --- end vector cost summary ---
   remark #25456: Number of Array Refs Scalar Replaced In Loop: 1
   remark #25015: Estimate of max trip count of loop=1
LOOP END

My question is: I do not understand how the speedup is calculated from

normalized vectorization overhead 0.773
scalar cost: 21 
vector cost: 11.000

Another more extreme and puzzled case could be

LOOP BEGIN at get_forces.c(2690,8)

   remark #15388: vectorization support: reference q12[j] has aligned access   [ get_forces.c(2694,19) ]
   remark #15388: vectorization support: reference q12[j] has aligned access   [ get_forces.c(2694,26) ]
   remark #15335: loop was not vectorized: vectorization possible but seems inefficient. Use vector always directive or -vec-threshold0 to override 
   remark #15305: vectorization support: vector length 2
   remark #15309: vectorization support: normalized vectorization overhead 1.857
   remark #15448: unmasked aligned unit stride loads: 1 
   remark #15475: --- begin vector cost summary ---
   remark #15476: scalar cost: 7 
   remark #15477: vector cost: 3.500 
   remark #15478: estimated potential speedup: 0.770 
   remark #15488: --- end vector cost summary ---
   remark #25436: completely unrolled by 3  
LOOP END

Now, 3.5+1.857=5.357 < 7
So, I could still simd this loop and gain a speedup or I should take the speedup number 0.770 in the report and not simd it?

How to understand the speedup in optimization report from icc compiler?

user_3557269 · Answer

The “scalar cost” means “cost of one iteration of scalar loop”.

The “vector cost” means “cost of one iteration of vectorized loop divided by vector_length*unroll_factor”, i.e. cost of somewhat equivalent to one scalar iteration.

The “vectorization overhead” shows normalized (by vector iteration cost) cost of vector initializations/finalizations before/after the loop.

The “estimated potential speedup” is calculated for the whole loop execution. It shows the normalized (by scalar iteration cost) potential gain of vectorized loop execution – including peel, remainder, and main loop for the estimated loop trip count. It can’t be derived explicitly from the scalar and vector cost shown above.

How to understand the speedup in optimization report from icc compiler?

1 Answers