eddjp@edi386.UUCP ( Dewey Paciaffi ) (02/08/91)
While I have access to a loaner Sparc2, I've been trying to bench it against my RS/6000 Model 320. I've been finding that the 6000 has about twice the I/O throughput of the Sparc (1.5MB vs .75 MB ) using real simple read and write system calls on large files (25MB). Other I/O related tasks like multiple execl calls run about 2 times as fast on the 6000. Both machines are using 3-4 MB/sec SCSI drives and both machines are running with only a single user during the benchmarks. Both have 16 MB memory. Is this consistent with what others are finding, or is there something wrong with this particular Sparc box? -- Dewey Paciaffi ...!uunet!edi386!eddjp
root@sun1.ruf.uni-freiburg.de (Martin Walter) (02/12/91)
eddjp@edi386.UUCP ( Dewey Paciaffi ) writes: >While I have access to a loaner Sparc2, I've been trying to bench it against >my RS/6000 Model 320. I've been finding that the 6000 has about twice the >I/O throughput of the Sparc (1.5MB vs .75 MB ) using real simple read and >write system calls on large files (25MB). Other I/O related tasks >like multiple execl calls run about 2 times as fast on the 6000. >Both machines are using 3-4 MB/sec SCSI drives and both machines are running >with only a single user during the benchmarks. Both have 16 MB memory. >Is this consistent with what others are finding, or is there something >wrong with this particular Sparc box? >-- >Dewey Paciaffi ...!uunet!edi386!eddjp I just tried at a SS2: # /bin/time dd if=/dev/zero of=x bs=1024k count=25 25+0 records in 25+0 records out 16.4 real 0.0 user 5.8 sys # /bin/time dd of=/dev/null if=x bs=1024k count=25 25+0 records in 25+0 records out 17.5 real 0.0 user 4.6 sys ==> 1.52 MB/sec write and 1.42 MB/sec read speed. Martin -- Martin Walter | Mail: mawa@sun1.ruf.uni-freiburg.de Rechenzentrum der Universitaet | University Computing Center Hermann-Herder-Str.10 | Phone: +49 761 203 4532 D-7800 Freiburg i.Br / Germany | FAX: +49 761 203 4122