home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.ibm.pc.hardware
- Path: sparky!uunet!ferkel.ucsb.edu!taco!gatech!swrinde!cs.utexas.edu!sdd.hp.com!hp-col!fc.hp.com!chuckc
- From: chuckc@fc.hp.com (Chuck Cairns)
- Subject: Re: Is 5 MB/sec to disk possible?
- Sender: news@fc.hp.com (news daemon)
- Message-ID: <BxH4rH.BB5@fc.hp.com>
- Date: Tue, 10 Nov 1992 00:16:29 GMT
- References: <45537@ogicse.ogi.edu>
- Organization: Hewlett-Packard Fort Collins Site
- X-Newsreader: TIN [version 1.1.6 PL6]
- Lines: 67
-
- John Hunt (johnhunt@apee.ogi.edu) wrote:
- : We are studing turbulence in the atmosphere and would like
- : to record bursts of 20 Megabytes or so of data to disk
- : as quickly as possible.
- :
- : In the ideal case we would like to achieve a sustained
- : transfer rate of 5 MB/second, but we could also make use
- : of 1 MB/second rates.
- :
- : Our platform is a Gateway 486-33 (currently on order).
- : Once we have the data on disk we will transfer it to an
- : Exabyte tape, then take another burst of data.
- :
- : 1. Can anyone suggest a disk for this application?
- : I assume I will need a separate disk for data collection,
- : probably a 3.5", 5400 RPM type. Am I better off with
- : a separate SCSI controller or with the local bus IDE
- : controller supplied with the Gateway system?
- :
- : 2. Would I significantly gain speed if I bypass the DOS
- : file system and perform raw writes to the disk? Can someone
- : who has succeeded with this technique comment on it?
- :
- :
- : John Hunt
- :
- : Senior Research Engineer
- : Dept. of Electrical Engineering and Applied Physics
- : Oregon Graduate Institute
- : Beaverton, Or. 97006
- : (503) 690-1143
- : (503) 690-1406 fax
- :
- : johnhunt@apee.ogi.edu
-
- Some thoughts,
-
- I have heard of a true ~4 megabytes/sec to the new Maxtor 6400 rpm drives.
- They don't seem to be really available quite yet. That would put a good
- 5400 RPM FAST-SCSI-II drive in the (very approx) 3.2 to 3.4 megabytes/sec
- range. Standard is ~2 Megabutes/sec for SCSI. Noe of this "to the buffer
- stuff" that the magazines keep feeding, this is "bits on the platter"!
-
- So you could consider a RAID array (maybe talk to Dell) or you could
- go to two FAST-SCSI-II high-RPM disks and ping-pong them. Ultrastor
- has a VESA local-bus FAST-SCSI-II controller called the 34F. I
- SERIOUSLY doubt that IDE will do it. Moreover, IDE can't talk to 2
- drives simultaneously like SCSI can. I haven't seen anything even close to
- the SCSI-II rate above noted for IDE, even on local bus. As for thru DOS or
- not, I'd guess that you will need to do raw. One good option is to go
- UN*X, or better yet, one of the Real-time Un*x lookalikes. Don't know
- if the US-34F would be supported there or not. Your data-acquisition
- card will need enough buffer to survive the burp on the bus to handle
- the disks. VL-bus can bus-master but (as I understand it) that freezes
- the processor as currently implemented. ( EISA would be better in that
- regard.) Also are you going to use a home-brew VL-bus card to grab the
- data ? (Can you get 5-Megabytes/sec thru an AT card at the standard
- 8.33MHz clock rate? You actually need ~ >10 Megabytes/sec peak throughput
- so you can ping-pong down to 5 Megabytes/sec.)
-
- Very best of luck,
- Chuck
-
- chuckc@fc.hp.com
-
- PS: If you get stuck, email me, and I'll put on my "sales" hat and see if
- I can find something "RISC-Y" which will do this around here. Probably can.
-