AM> In a message to Paul Edwards about 8 meg disk cache, Paul Markham said:



 PM>> I don't know whether a RAM disk will get paged or not. When we

 PM>> were playing the other day, it seemed as though the disk cache

 PM>> got paged out, so I guess the RAM disk could as well. The

 PM>> question then is, which is more efficient - I/O caused by

 PM>> paging, or I/O caused by reading/writing files? I suspect paging

 PM>> would be better since it wouldn't be reading and writing the

 PM>> directory.



 AM> It doesn't really make sense to have a cache pageable!  I would be

 AM> very surprised if cache.exe data areas are paged!  The ramdisk otoh,

 AM> may or may not be.  One again, it doesn't make sense for them to be

 AM> paged, but you never know...



Anthony,



I agree that making a cache pagable doesn't make much sense. While Paul and
I were trying out a large cache on my machine (it only had 8 meg), I
defined a cache size of 6.5 meg. OS/2 still booted. I expected that it
would thrash itself to death paging code into the remaining 1.5 meg. But it
booted about the same speed it usually does (maybe a little slower),
without much (if any) increase in disk activity.



If I increased the cache passed that size, OS/2 complaining that it didn't
have enough memory to define the cache. Does this mean OS/2 really only
required 1.5 meg to boot to the work place shell (I doubt that)? Beats me.
Got any ideas?



Paul



--- GoldED 2.40