Developer Guide

Contents

Reducing Floating-Point Rounding Operations (
-Xsffp-contract=fast
)

Include the
-Xsffp-contract=fast
option in your
dpcpp
command to direct the
Intel® oneAPI
DPC++/C++
Compiler
to remove intermediary floating-point rounding operations and conversions whenever possible and carry additional bits to maintain precision.
This flag turns on the same optimizations as the deprecated
-Xsfpc
flag for all instructions, unless denoted otherwise by the fp contract Pragma.
Implementing this optimization control also changes the rounding mode. It rounds towards zero only at the end of a chain of floating-point arithmetic operations (that is, multiplications, additions, and subtractions).
Example
To direct the
Intel® oneAPI
DPC++/C++
Compiler
to reduce the number of rounding operations, invoke the following command:
dpcpp -fintelfpga -Xshardware -Xsffp-contract=fast <source_file>.cpp

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.