it is true that if less libraries are compiled, then things will be faster. This will be difficult though, as in 75 most things, including the NAG library, are compiled into origin.h to speed up compiling and loading of other files.
Do you make calls to LT from your OC code? If that is the case, it is possible to speed things up a bit if you know the LT code does not then call OC functions. As when LT interprets, it will need to search for all the words to check and see if it is an OC function. The @OC system variable can turn this OC search off, as shown below,
bool LT_fast(LPCSTR lpcszLT)
{
double dVal;
LT_get_var("@OC", &dVal);
LT_set_var("@OC", 0); // turn off all OC searching for subsequent LT
bool bRet = LT_execute(lpcszLT);
LT_set_var("@OC", dVal);
return bRet;
}
void test()
{
LT_execute("@OC=;sec;type from regular LT_execute;for(i=1;i<10000;i++) {type -q $(sin(i))};watch");
LT_fast("@OC=;sec;type from LT_fast;for(i=1;i<10000;i++) {type -q $(sin(i))};watch");
}
Now this LT string only has one sin(i) function, so the difference is small, but if the LT code involves a lot more LT functions and other terms, then difference might be bigger.
CP
Performance: V7.0 vs V7.5
Posted - 01/10/2005 : 12:15:08 PM
