gfredin@enlog.Wichita.NCR.COM (10/06/89)
This is probably an ignorant question, but here it is: I am not that familiar with Cray systems, but my understanding has always been that they transfer large amounts of data when an I/O request is made. This seems to contradict my perception of UNIX using relatively small block sizes spread randomly across the disk. Does UNIX on a Cray use a modified file system, raw I/O, or some other method I am not familiar with? If it uses a modified file system, what does an I/O access look like in terms of block size and randomness? If you think the response to this is below most on the net, please email responses. Thanks in advance! Standard disclaimers apply. ********************************************************************** * Jerry Fredin NCR Corporation * * Sr. Consulting Analyst Peripheral Product Division * * I/O Architecture 3718 North Rock Road * * Wichita, KS 67226-1397 * * * * uunet!ncrlnk!ncrwic!enlog!gfredin Jerry.Fredin@Wichita.NCR.COM * **********************************************************************
mike@BRL.MIL (Mike Muuss) (10/07/89)
Cray uses a special file system of their own. They are extent based. Note that the XMP and Cray-2 have *different* filesystems, an unfortunate and grauitous difference. Best, -Mike