🚀 go-pugleaf

From a practical point of view I think a far smaller run of data
should give you enough statisitical confidence that the 24"cables are
or aren't munging up your data. You might also think about how long
it's going to take even the fastest EIDE to complete your radom
read/write and seek test. After the drive and system buffers and
caches are all saturated the real world transfer rates of most drives
falls _way_ down no matter how fast the original platter to bus  spec
is.  Try transferring a gig worth of  10,000 files from EIDE drive to
EIDE drive on a fast PCI or VLB sytem with say 16 megs of RAM sometime
and see what happens to the transfer speed after about 3-8 minutes.
2-4 MPS platter to bus soon becomes a memory.   500K ps is more like
it depending on file size. At this rate your terabyte transfer is
going to take  around 2,000,000 seconds.  2 sec for a meg/ 2,000 secs
for a gig / 2,000,000 secs for a terabyte. 


Ian Stirling <root@mauve.demon.co.uk> wrote:

>I'm making an external drive box for my SCSI and IDE drives, 1\" thick wood,
>no airholes, fluid loop cooling (for QUIETNESS, I'm sleeping 6\" from it, and
>don't want it to intrude)
>The problem I have, is I'd like to extend the IDE cables a little beyond the
>18\" maximum, maybe out to 24\" or so. But, I don't want to lose data.
>Hence, I want to run a LARGE amount of data throgh it, about 1Tb or so
>(1000Gb) doing random read/write/seek.
>This should take the error uncertainty to below 1 in 10^12 or on the same
>order as the spec for soft errors.
>Is there one out there, or do I have to write my own?
>Ian Stirling.