🚀 go-pugleaf

In article <5rPFb5roUjB@irata.Jena.Thur.De>
Marc Mueller <MaMue@irata.Jena.Thur.De> wrote:
> ycheng@advansys.com (Yu-Ping Cheng)  wrote :
> > Use the regular PCI SCSI controller with the SMART drive from your
> > Windows. Don't waste any money on the cacheing controller, which is only
> > good for benchmark tests.
> 
> The best thing I ever bought is my PCI Cache Controller. SMARTDRIVE is the  
> *absolut* *slowest* cacheing software (several programs are faster, e.g.  
> HyperDisk).

I think we need some intro how about cache controller works.

Most operating systems, including DOS and Windows, already cashe the hard
disk data.  Depending on the cache size when the requested data is not
in cache, the operating system needs to go to a controller, which in turn
goes to hard disk, to get the data. The purpose of cache is to avoid the
hard disk access time -- normally around 15 msec.  Cache can be accessed
in sub-milliesecond.

Say, if the operating system keeps 1M of cache and your controller has 2M 
of cache, you have some chance finding the data in the controller cache 
after missing it in the operating system cache.  But, if we reverse this,
2M in OS and 1M in controller, you have ZERO chance finding the data in the
controller cache after missing the data in the OS cache.

SmartDrive is a form of operating system cache supplimenting the OS cache.
When data are transferred from SmartDrive, it is done by PIO, about 24 CPU
clocks every four bytes.  When data are transferred from the controller
cache, it is done by bus master, four bytes burst every bus clock.  This
is why your SmartDrive will always be slower than your controller in your
benchmark tests.  This is what I mean that the cache controller is great
for benchmarks.

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.  SmartDrive is a very effective way of
reducing the 15 msec disk access time to sub-milliesecond memory access
time for DOS and Windows users. 

-- 
Y.P. Cheng,   ycheng@advansys.com
"We will be successful, because we won't accept failure."