|
PureBytes Links
Trading Reference Links
|
"M. Simms" wrote:
> Totally agree.....that's why the DLL interface is there......
> and I agree that doubles will utilize the floating point processor of the
> Intel CPU...so the CPU hit won't be much at all.
>
> However, the big hit is the memory storage of all of the series-based
> variables......
> that could be significant....and cause much paging and swapping at the
> application-level.
<snip, snip>
Actually, if you will study the reference I gave, you will learn that
with most algorithms, only a few variables, and fewer arrays of data
or intermediate results, need to be stored in double precisions to
make huge differences in the numerical accuracy of the algorithm. For
example, to calculate the dot product of two arrays (vectors), only
the partial sum (one variable) needs to be retained in double precision,
not the data arrays. If this were coded in a tight loop in c with a
modern compiler, then the running partial sum would be retained in the
FPU registers, so that you get an 80 bit accummulator which is a lot
better than double precision.
Intel did not put 80 bit registers in their FPU's for nothing. There
is a reason, especially if you consider that they have been doing it
since the original 8087 in the early 80's.
Rod
|