Hi, everyone wants lean and fast code, so do I. :) Unfortunately the IXA SDK C-Compiler refuses to properly compile the ua_get/ua_set instructions when optimization ist turned on (-O switch set for example via workbench Build:Settings:C-Compiler:Optimize:Speed). The ua_get/ua_set instructions lead to compile time errors like this : "Unable to determine memory region for ua_set" This does not happen when the compiler is told not to do optimization, so I suspect it to be a bug. The unoptimized code is far bigger (tested with ua_get/ua_set commented out) and innapropriate for serious testing. As I suspect unaligned access to be a common thing here my question into the round : Anyone seen this, too ? And more important - howto overcome this Problem ? By the way - switching to assembly code is not an option to me. ;) Example follows : #include "ixp.h" void main (void) { __declspec(sdram) long long *ptr; ua_set_16(ptr,0,1234); } void exit (unsigned int arg) { } With best regards and thanks in advance. Yours, Robert Klima.