🚀 go-pugleaf

fer@dxcern.cern.ch (Fer de Jong) writes:

>In article <408nqk$hjf@aimnet1.aimnet.com>,
>Yu-Ping Cheng <ycheng@advansys.com> wrote:
>>Most operating systems like UNIX will allocate all unused memory for cache.
>>Hence, sometimes the system will 4M+ of cache. Now, if you have 1M of cache
>>on your controller, you have ZERO chance of finding data in the controller
>>after missing the data in the OS.

>I wonder if this ZERO chance is actually true, as far as I understand it,
>sone of the caching controllers have pretty elaborate caching schemes
>implemented on an onboard CPU, I suppose you assume that they only
>implement a 'most recently accessed' caching scheme wich would justify
>your ZERO change statement.
>Any comments whether this is true, and perhaps any numbers to show how
>efficient a disk controller cache is during 'normal' linux work???

The point is if you're going to pay for memory, give the operating system
the chance to use it in a flexible way, depending on its workload.
Linux will use it intelligently.
People who buy cached disk controllers and get all excited about the
disk bench mark test don't seem to realise that all this exciting
I/O is pointless when the extra memory may have avoided the need for
an I/O in the first place, eg swapping................