RAM disk.

Gary Heston gary at sci34hub.UUCP
Sat Sep 15 01:24:12 AEST 1990


In article <6167 at titcce.cc.titech.ac.j`> mohta at necom830.cc.titech.ac.jp (Masataka Ohta) writes:
>In article <1990Sep13.002300.15266 at mlb.semi.harris.com>
> [ discussion of ramdisc performance deleted ]

>The reason is that memory disk can't do async write. Data is copied
>from user space to buffer cache and then to memory disk. With ordinary
>disk, data is only copied to buffer cache.

You're presuming that all hardware is incapable of unattended
memory-to-memory DMA.  I don't think this is the case. Further, a
well-integrated RAM disc should transfer directly from user space to
RAM disc in cases where the RAM disc is part of main memory. This is
also not necessarily the case. A RAM disc can be implemented separately
from main memory, or as a peripheral device. (Installable software
drivers, however, would only use main memory.)

>If you use elaborated and complicated memory disk, it can be only as
>slow as ordinary disk, but not faster.

I think you're forgetting drive latency, here. A RAM disc will respond
immediately, where a hard drive may take several ms just to find the
data. Somewhere around here, I have a data sheet for a RAM disc with
a SCSI interface. I think it'll show a lot of speed improvement over an
"ordinary disk". Granted, not everyone has one of these, but I'd
sure like one for my swap device.

>						Masataka Ohta


-- 
    Gary Heston     { uunet!sci34hub!gary  }    System Mismanager
   SCI Technology, Inc.  OEM Products Department  (i.e., computers)
"The esteemed gentlebeing says I called him a liar. It's true, and I
regret that." Retief, in "Retiefs' Ransom" by Keith Laumer.



More information about the Comp.unix.admin mailing list