NERSCPowering Scientific Discovery Since 1974

Intel Compiler Peformance on Edison

 

These are the Intel optimization options we compared.  The quotations are from the Intel compiler on-line man pages.

Intel Compiler Options Tested
-fast This "maximizes speed across the entire program".  It is a very high level of optimization and includes interprocedural optimization across different source files.  It increases compilation time significantly, and occasionally compiles will fail with this option which succeed with the other options, probably due to the greater processor and memory requirements.
-fast -no-ipo This includes all of the optimizations of the -fast option except for interprocedural optimization.
default No optimization flags.  By default the Intel compiler has a high level of optimization.  It is comparable to the -O2 optimization level which "enables  optimizations  for speed", and is the recommended option for codes in the online man page.
-O3 This performs all of the -O2 options as well as additional more aggressive loop transformations.  It is recommended for "applications  that have loops that heavily use floating-point calculations and process large data sets."
-O3 -unroll-aggressive -opt-prefetch This was recommended to us by benchmarkers as being a good supplement to the -O3 optimizations.
-O The same as -O2.

Edison Intel Compiler Performance

 

The optimizations associated with -fast produce faster code on the average than other options.  Adding -no-ipo to -fast significantly reduces the compilation time, but still produces very well optimized code.