|
PureBytes Links
Trading Reference Links
|
>I don't understand why the amount of RAM would affect CPU
>usage, but it does.
The lest ram you've got the more the system (O.S.) is forced to swap and move
memory blocks for the same work/load. (Swapping/Paging concepts)
With more ram, the ratio "user cpu"/"system cpu" is increasing,
CPU should follow (approximatively) a 1/x fonction.
ie.
The more ram you add the less it has effect.
128 to 256 is more important than from 256 to 384.
256 to 384 is more important than from 384 to 512.
etc...
Windows(3.x, 9X) got troubles to manage more than 512Mo.
NT/W2k manages it without trouble.
Bug or feature ? :))
my 0.02
At 2001-05-15 06:42 -0700, ribau@xxxxxxxxxxx wrote:
>I noticed a substantial drop in CPU usage when I installed the second 256Mb
>RDRAM in this system. CPU usage dropped from average 58% at the open to
>average 21%, and the average during the day dropped from 26% to less than
>10% average. I don't understand why the amount of RAM would affect CPU
>usage, but it does.
>
> >>A 2-CPU machine will achieve the same result with much slower processors.
> >>Unfortunately I'm not in a position to do the test, as I don't
> currently use that
> >>type of real-time datafeed, but this has been my experience in similar
> situations.
> >>What's happening at the open is that the one CPU is completely busy
> processing
> >>incoming data, and hence unavailable to do anything else like move a
> trendline (and
> >>Omega wrote the process in such a way as to completely monopolize the
> processor).
> >>On a 2-CPU machine there's always another processor available, so you
> rarely see a
> >>delay - the keyboard never goes "dead". The advantage is not so much that
> >>individual operations are faster as that everything is *smoother*. For
> example you
> >>can get a lot more done comfortably in other programs while running a long
> >>optimization.
> >>
|