Hello all,

We are using the IAR toolset (Embedded Workbench 7.20H) to build applications for our 8051 based microcontrollers.

All of our controllers have a second data pointer so we have been compiling with support for the second data pointer assuming that it would result in smaller and more efficient code.

However in debugging a recent defect, we ran some builds with the second data pointer support turned off and much to our surprise, the code built smaller (1k on our 32k project up to 4k on our 128k project) and appeared to run faster (we don't have exact numbers on performance).

Has anyone else experienced this type of difference with IAR? Does it seem unreasonable to expect smaller/faster code when the compiler is aware of and using the second data pointer?

Thanks,
Keith Richeson