Allow to fold vector load if there is more than one bitcast, so in the case:
[oota-llvm.git] / test / CodeGen / X86 / sse-intel-ocl.ll
2012-10-24 Elena DemikhovskySpecial calling conventions for Intel OpenCL built...