CH>I have read where mitsumi has developed an IDE
CH>writable CD-Rom. I assumed
CH>it was on the streets, but may not. There is no good
CH>technical reason that
CH>a writeable CD can't use an EIDE buss.
JG> Sure there is - IDE transfers require lots of CPU time,
JG> which may cause the
JG> system to be too busy to keep the writer's buffer full.
JG> *The* major problem
That is probably not as much as a problem as it used to be Jeff. Now I
beleive all newer Intel-Triton chipset EIDE are busmastering. Intel has
busmastering drivers for NT, W95 and DOS. Also newer computers have quite a
bit faster EIDE hosts and CPU's, so even without busmastering it isn't the
CPU load on a 166 or P200 that it would be on a slower pentium or 486.
Been running one with a 5gig EIDE Maxtor on one of my computers. Quite an
improvement over older EIDE. Haven't had a chance to really wring it out, but
it sure seems good so far. Probably not like my AHA2940W, but good.
Anyway, I'm not too concerned about it being EIDE or SCSI if it is cheap
enought and works in NT.
IDE CD-ROM is more CPU over head than SCSI, but I have never bought off on
the ragazine stories of 80%. That really depends on the Computer system it is
running on. If you benchmarked it on an old 386 or 486, then certainly the
CPU over head would be way high, perhaps approaching the 80% mark, but if you
test it on a 166mhz or 200 mhz CPU and the 5 fold faster EIDE controllers
that are standard on a decent system, it -couldn't- be all that much CPU
overhead. Maybe 20 or 30% as compared to 10% with SCSI. With EIDE
bussmastering drivers, perhaps as low as SCSI. In anycase I really doubt the
equipement I use, EIDE or SCSI is going to have any difficulty with buffer
underuns. A 2x (300kb/sec max) or a 4x (600kb/sec max) writeable CD just
can't be that much overhead. It just isn't that much throughput. (300kb/sec
is about the same throughput as a 2:1 interleaved MFM hard drive (250kbsc).
4x is about the same through put as the old AT 286 1:1 interleave MFM
controller (550kb/sec). Considering that, it would seem to me that the
ragazine writers that are the self proclaimed "experts" on the subject, might
be grossly over-exagerating the performance differences between EIDE and SCSI
when they are making up thier stories about CD-ROM and IDE vs SCSI.
--- Maximus/NT 3.01b1
---------------
* Origin: 33,600bps Windows NT _Powered_! (1:303/1)
|