Questions on coding my FFT&LLR implementation

[retina]

Originally Posted by nuggetprime

The above was not stated precisely, I think.
I want to know if in C, when i define a ,say, unsigned short int, how can I access a certain bit of it,in reasonable CPU time?
Thanks,
--nuggetprime

I would have thought a simple bitwise and mask would suffice. (1 << bit_num) && the_int

[jrk]

You can use the bitwise operator 'and' (&) and a bitmask to test a specific bit.

x & 0x0001
x & 0x0002
x & 0x0004
x & 0x0008
x & 0x0010
etc

Or if you want to use a variable to specify the bit, you can use a shift operator (>>).

x >> y & 0x0001

[nuggetprime]

It works. Thanks a lot retina & jrk.
If you read this, Xyzzy,can you please rename this thread to "Questions on coding my FFT&LLR implementation"?

Thanks,
--nuggetprime

[nuggetprime]

If I compile the following C code and run the program, it doesn't print anything!

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
unsigned short apl_uiswap(unsigned short nm)
{
unsigned short buf;
int le;
for (le=0;le>16;le++)
{
printf("HI\n");
}
return 3;
}
int main(int argc, char* argv[])
{
apl_uiswap(43);
}

Is that the same for your system?
What the hell is going on here??

[retina]

One presumes you mean: for (le=0;le<16;le++)

[nuggetprime]

Thanks retina.
well, it now prints
But if I print le before the loop, it always prints a somewhat random integer:

#include <stdio.h>
#include <stdlib.h>
#include <math.h>
unsigned short apl_uiswap(unsigned short nm)
{
unsigned short buf;
unsigned short le;
printf("%d\n",le);
for (le=0;le<16;le++)
{
printf("HI\n");
}
return 3;
}
int main(int argc, char* argv[])
{
apl_uiswap(43);
}
#include <stdio.h>
#include <stdlib.h>
#include <math.h>
unsigned short apl_uiswap(unsigned short nm)
{
unsigned short buf;
unsigned short le;
printf("%d\n",le);
for (le=0;le<16;le++)
{
printf("HI\n");
}
return 3;
}
int main(int argc, char* argv[])
{
apl_uiswap(43);
}

is that normal?

thanks,
nuggetprime

[retina]

Originally Posted by nuggetprime

But if I print le before the loop, it always prints a somewhat random integer:
is that normal?

Sure it is. You're printing the value of an uninitialised variable, it could be anything. Not until the "le=0" is encountered does le become a known value.

[nuggetprime]

If I write

void aplint_init(apl_int pt)
{
pt.dg=(unsigned short*) calloc(1,2);
}

i get a segfault.
but if I write

#define aplint_init(x) x.dg=(unsigned short *)calloc(1,2)

everything is ok.

What am i doing wrong?

thanks,
nugget

[nuggetprime]

Solved above thing.
The next thing I think i'm not sure about is speed related.If you have a large set of numbers and have to add them,
is it then faster or equally fast to directly compute x1+x2+x3... instead of adding a2 to a1, then adding a3 to the result and so on...

Nice day,
nuggetprime

[retina]

Originally Posted by nuggetprime

The next thing I think i'm not sure about is speed related.If you have a large set of numbers and have to add them,
is it then faster or equally fast to directly compute x1+x2+x3... instead of adding a2 to a1, then adding a3 to the result and so on...

Short answer: It depends upon many many factors.

Long answer: Just code up the two methods you propose and see which is faster on your hardware with your data with your set sizes.

[nuggetprime]

Thanks again retina.
One question more, if I may do some more "spamming" here:
I'm not able to get long double precision. If I run:

#include <stdio.h>
int main()
{
long double c;
c=1/7.0;
printf("%3.30Lf\n",c);
return 0;
}

it prints:

0.142857142857142849212692681249

As seen, only the first 16 digits of the mantissa are true, which is equal to normal double. What's wrong?

Thanks,
nuggetprime