solomon@chaos.utexas.edu (Thomas Solomon) (05/23/91)
I don't want to start any flame wars between ISC and SCO supporters but I need some information about performance of these unices on souped-up 486/33 MHz EISA machines. I've read quite a bit about pros and cons of SCO, but the pros and cons discussed deal mainly with features of the system, rather than shear horsepower. The Personal Workstation review of ISC and SCO has an intriguing section in it: "On the IOBench 2 disk, SCO Unix outperformed Interactive only on single-tasking reads. Interactive had an advantage of 25 to 30 percent on the impotant random read/write test, and of several hundred percent on sequential writes." SEVERAL HUNDRED PERCENT!?!?? Is this for real, or is this a typo? We are quite interested in sequential writes, because we will be setting up a partition on our disk for data-taking that will be wiped clean before every data run (to avoid forcing the head to jump around). Can anyone confirm or refute this "several hundred percent" thing? A few other questions: do either (or both) of these unices support 486-specific commands (presumably increasing efficiency)? How about features specific to the EISA bus, such as bus mastering (very important for high throughput in disk access) and DMA, and 33 MHz burst mode? Also, can either (or both) handle _synchronous_ SCSI transfers to disk? For the record, we are planning on purchasing an Austin Computer Systems 486/33 MHz EISA machine with an EISA, SCSI, non-caching disk controller (probably the UltraStor 24F), and a Seagate Elite (ST41600N) SCSI disk (rated for 3Mbytes/sec internal transfer rate). We want to get an operating system that will take advantage of all the performance of this machine. By the way, we are considering putting a DOS partition on the disk, and using DOS for real-time data-taking, then switch to unix for the analysis. Thanks. Tom Solomon solomon@chaos.utexas.edu
larry@nstar.rn.com (Larry Snyder) (05/24/91)
solomon@chaos.utexas.edu (Thomas Solomon) writes: >The Personal Workstation review of ISC and SCO has an intriguing section >in it: "On the IOBench 2 disk, SCO Unix outperformed Interactive only on >single-tasking reads. Interactive had an advantage of 25 to 30 percent >on the impotant random read/write test, and of several hundred percent on >sequential writes." SEVERAL HUNDRED PERCENT!?!?? Is this for real, or >is this a typo? We are quite interested in sequential writes, because Nope - the ISC SCSI FFS is much, much faster than SCO's Acer FFS. We started playing with the SVR4 FFS and it appears to be quite fast as well - benchmarks coming soon! -- Larry Snyder, NSTAR Public Access Unix 219-289-0287/317-251-7391 HST/PEP/V.32/v.32bis/v.42bis regional UUCP mapping coordinator {larry@nstar.rn.com, ..!uunet!nstar.rn.com!larry}
mikes@iuvax.cs.indiana.edu (Michael Squires) (05/24/91)
In article <49397@ut-emx.uucp> solomon@chaos.utexas.edu (Thomas Solomon) writes: > >The Personal Workstation review of ISC and SCO has an intriguing section My memory of this article is that it uses an obsolete version of SCO UNIX. I notice that in recent benchmarks in PW magazine they are still using SCO UNIX 3.2.0, not the current 3.2.2, and I understand that disk I/O is one area that was improved. My memory is that this article used 3.2.1, but that's still before the driver improvements went in. I am running SCO ODT 1.0 (upgrading to 1.1 soon) on a 486/25; the final release of 1.0 seems quite reliable. -- Mike Squires (mikes@iuvax.cs.indiana.edu) 812 855 3974 (w) 812 333 6564 (h) mikes@iuvax.cs.indiana.edu 546 N Park Ridge Rd., Bloomington, IN 47408 Under construction: mikes@sir-alan.cica.indiana.edu
rmk@rmkhome.UUCP (Rick Kelly) (05/25/91)
In article <49397@ut-emx.uucp> solomon@chaos.utexas.edu (Thomas Solomon) writes: [ ISC and SCO stuff deleted ] >A few other questions: do either (or both) of these unices support >486-specific commands (presumably increasing efficiency)? How about >features specific to the EISA bus, such as bus mastering (very important >for high throughput in disk access) and DMA, and 33 MHz burst mode? Also, >can either (or both) handle _synchronous_ SCSI transfers to disk? Bus mastering is not EISA specific. Micronics, Intel, and Mylex all build 386 33 mhz ISA motherboards that support bus masters, and will do synchronous transfer with Adaptec AHA-154xx SCSI controllers. If SCO and ISC don't have support in their drivers, they are just plain lazy. >For the record, we are planning on purchasing an Austin Computer Systems >486/33 MHz EISA machine with an EISA, SCSI, non-caching disk controller >(probably the UltraStor 24F), and a Seagate Elite (ST41600N) SCSI disk >(rated for 3Mbytes/sec internal transfer rate). We want to get an >operating system that will take advantage of all the performance of >this machine. By the way, we are considering putting a DOS partition >on the disk, and using DOS for real-time data-taking, then switch to >unix for the analysis. Or you could experiment with VPIX. Rick Kelly rmk@rmkhome.UUCP frog!rmkhome!rmk rmk@frog.UUCP
bill@unixland.uucp (Bill Heiser) (05/30/91)
In article <9105242244.15@rmkhome.UUCP> rmk@rmkhome.UUCP (Rick Kelly) writes: >In article <49397@ut-emx.uucp> solomon@chaos.utexas.edu (Thomas Solomon) writes: >>this machine. By the way, we are considering putting a DOS partition >>on the disk, and using DOS for real-time data-taking, then switch to >>unix for the analysis. > >Or you could experiment with VPIX. I wouldn't think anyone would want to use VP/ix for any kind of REAL TIME processing. It is a real performance dog, at least on a 25mhz machine. It definitely couldn't be counted on to perform real-time performance. Of course I may have mis-interpreted your need for "real time" here; maybe you mean just having someone type in numbers "as they're available" or something? -- bill@unixland.natick.ma.us The Think_Tank BBS & Public Access Unix ...!uunet!think!unixland!bill bill@unixland ..!{uunet,bloom-beacon,esegue}!world!unixland!bill 508-655-3848 (2400) 508-651-8723 (9600-HST) 508-651-8733 (9600-PEP-V32)