RE: Translated w/ 2.4 CRASH
>
> Computer storage heirarchies (on-chip cache, secondary cache, main
> memory,
> swap area on disk) try to retain the most recently used
> information in fast memory, while pushing the rest out to slower memory.
> This tends to work fine for code, since programs spend a lot of time in
> small loops, but it often breaks down badly where
> a program is repeatedly sweeping through huge arrays of data
> from beginning to end. Instead of fetching most values
> from fast memory, you might end up fetching most from slow memory.
> This can be disasterous for performance.
>
> I've seen programs suddenly lose a few orders of magnitude of
> performance when their memory requirements exceeded some threshold.
>
>
That makes sense since the tree is partitioning up a big data set and
will need more or less random access to the set. However, the problem
still only occurs w/ 2.4, although the first iteration in 2.3 takes a
lot longer...
|
Not Categorized, Please Help
|
|
