Rita Berkowitz <ritaberk2008@[EMAIL PROTECTED]
> wrote:
> Ron Hunter wrote:
>> Yes, it IS poor disk I/O, but even the BEST disk I/O is orders of
>> magnitude slower than even average RAM access.
> The bottom line here is the better the disk I/O the less RAM you need
> and the time it needs to be tied up storing information.
Duh. Obviously, all you need, is disk IO that's faster than RAM.
Dear Rita,
would you have some data on the average time to retrive 10MB of
semi-random data from the disk and the same from memory?
You may assume a single 1/3rd stroke of the disk head and 1/2
rotation of the platter before data comes in, if you have no copy
of the Art of Computer Programming to look up a more complete
model of disk behaviour.
Show the complete, relevant math and the URL for spec sheet of
the hard drives you look at at their manufacturers' websites.
Please show where, by your model, the USD is better spent at a
faster disk (which one?) than a USD-identical amount of RAM.
Additionally, you may try to argue about the access speed and
costs of using RAID0.
Additionally, you may try to argue what happens if other
IO-operations are happening.
If you decline to state provable facts --- as I know you will ---
you only prove, again and again, that you spout some rote-learned
truisms without even remote understanding of the necessary
preconditions under which they are true.
-Wolfgang
|
