Developer Reference

Intel® oneAPI Math Kernel Library Vector Mathematics Performance and Accuracy Data

ID 772989
Date 12/04/2020
Public
Document Table of Contents

Powr

Computes a to the power b for elements of two vectors, where the elements of vector argument a are all non-negative.

Performance:

  Intel® Xeon® Processor E5-2680 Intel® Xeon® Processor E5-2699 v3 Intel® Xeon® Processor E5-2699 v4 Intel® Xeon® Gold 6148 Processor
single HA 10.07 10.19 8.54 2.82
LA 10.07 4.51 3.71 2.61
EP 10.07 3.17 2.74 1.69
double HA 18.78 19.79 16.61 6.47
LA 18.19 11.40 9.43 5.42
EP 10.78 10.38 8.61 3.38
Notes:
  1. Units - CPE (Clocks per Element)
  2. Performance was measured on "working" interval (double precision: [ 5.0, 100.0 ] for both arguments; single precision: [ 0.1, 10.0 ] for both arguments)

  3. Data - vectors of 1000 elements with random generated numbers.

  4. All data was collected with 64-bit libraries.

Accuracy:

  Intel® Xeon® Processor E5-2680 Intel® Xeon® Processor E5-2699 v3 Intel® Xeon® Processor E5-2699 v4 Intel® Xeon® Gold 6148 Processor
single HA 0.78 0.90 0.90 0.97
LA 1.73 1.64 1.64 1.05
EP 1.87E+03 1.87E+03 1.87E+03 277.80
double HA 0.51 0.51 0.51 0.74
LA 1.02 1.02 1.02 1.06
EP 6.45E+07 6.45E+07 6.45E+07 2.02E+07
NOTE:
Units - ULP (Units in the Last Place).

Performance Graphs: