The change makes fastmath default and changes the previous env-var
trigger to GMX_OCL_DISABLE_FASTMATH to allow disabling it for testing.
It is known to be faster on all supported platforms (and we use it in
CUDA too) and it has been planned to be enabled for the r'16 release.
Change-Id: I8d196c9a14785ff43aae144ed95c3cb29241c83b
only the flavor required for the simulation is generated and
compiled.
-``GMX_OCL_FASTMATH``
- Adds the option ``cl-fast-relaxed-math`` to the compiler
- options (in the CUDA version this is enabled by default, it is likely that
- the same will happen with the OpenCL version soon)
+``GMX_OCL_DISABLE_FASTMATH``
+ Prevents the use of ``-cl-fast-relaxed-math`` compiler option.
``GMX_OCL_DUMP_LOG``
If defined, the OpenCL build log is always written to the
compilerOptions += " -cl-opt-disable";
}
- if (getenv("GMX_OCL_FASTMATH") )
+ /* Fastmath imprves performance on all supported arch */
+ if (getenv("GMX_OCL_DISABLE_FASTMATH") == NULL)
{
compilerOptions += " -cl-fast-relaxed-math";
}