bill@unixland.uucp (Bill Heiser) (11/04/90)
I ran the SCSICNTL.EXE program to examine my CDC Wren IV (hooked up to an Adaptec 1542b), and have some questions about the appropriate settings. The system is running Esix Rev D (SYSVR3 Unix). Parameters: Defaults Current Options Disable Correction enabled enabled Transfer on error enabled enabled post error disabled disabled early error correction disabled disabled read continuous disabled disabled transfer block disabled disabled auto read allocation disabled disabled auto write allocation disabled disabled retry count 27 27 1-255 buffer full ratio 4 4 0-255 buffer empty ratio 4 4 0-255 [a bunch of stuff about tracks/sectors per zone skipped for now] write index enable disabled disabled cache enable disabled disabled What are the best settings to have on these parameters in my application? Thanks in advance... Bill -- home: ...!{uunet,bloom-beacon,esegue}!world!unixland!bill bill@unixland.uucp, bill%unixland.uucp@world.std.com Public Access Unix - Esix SYSVR3 - (508) 655-3848 other: heiser@world.std.com Public Access Unix (617) 739-9753
larry@nstar.uucp (Larry Snyder) (11/04/90)
bill@unixland.uucp (Bill Heiser) writes: >I ran the SCSICNTL.EXE program to examine my CDC Wren IV (hooked up >to an Adaptec 1542b), and have some questions about the appropriate >settings. The system is running Esix Rev D (SYSVR3 Unix). >Parameters: > Defaults Current Options >Disable Correction enabled enabled >Transfer on error enabled enabled >post error disabled disabled >early error correction disabled disabled >read continuous disabled disabled >transfer block disabled disabled >auto read allocation disabled disabled >auto write allocation disabled disabled >retry count 27 27 1-255 >buffer full ratio 4 4 0-255 >buffer empty ratio 4 4 0-255 >[a bunch of stuff about tracks/sectors per zone skipped for now] >write index enable disabled disabled >cache enable disabled disabled >What are the best settings to have on these parameters in my application? what you need to do is to change one item at a time - then bench the system, change another - then bench, etc.. I have cache enabled (made a 10% difference) and changed the buffer ratios from 4 to 16 (made a + difference on large disk IO, and - on small; but since most of mine is large - I changed it).. >Thanks in advance... >Bill >-- >home: ...!{uunet,bloom-beacon,esegue}!world!unixland!bill > bill@unixland.uucp, bill%unixland.uucp@world.std.com > Public Access Unix - Esix SYSVR3 - (508) 655-3848 >other: heiser@world.std.com Public Access Unix (617) 739-9753 -- Larry Snyder, Northern Star Communications, Notre Dame, IN USA {larry@nstar, uunet!sco!romed!nstar!larry, nstar%larry@iuvax.cs.indiana.edu} backbone usenet newsfeeds available Public Access Unix Site (219) 289-0282 (5 high speed lines)
bill@unixland.uucp (Bill Heiser) (11/04/90)
In article <1990Nov04.004934.232@nstar.uucp> larry@nstar.uucp (Larry Snyder) writes: > >what you need to do is to change one item at a time - then >bench the system, change another - then bench, etc.. Thanks for the feedback Larry. What utility are you using to do your benchmarking? >I have cache enabled (made a 10% difference) and changed the >buffer ratios from 4 to 16 (made a + difference on large disk IO, >and - on small; but since most of mine is large - I changed it).. I'm a little uncertain about the "buffer ratios" -- I didn't see any explanation of it in the lenghy aha1542 manual -- do you have anything that describes them? My "largest" disk IO activity is news processing (Cnews) and compiling software -- does that qualify as "large", and would you expect it to benefit from the increase in buffer ratios? Thanks for the info. Bill -- home: ...!{uunet,bloom-beacon,esegue}!world!unixland!bill bill@unixland.uucp, bill%unixland.uucp@world.std.com Public Access Unix - Esix SYSVR3 - (508) 655-3848 other: heiser@world.std.com Public Access Unix (617) 739-9753
larry@nstar.uucp (Larry Snyder) (11/05/90)
bill@unixland.uucp (Bill Heiser) writes: >Thanks for the feedback Larry. What utility are you using to do >your benchmarking? We have a couple of benchmarking tools here on nstar - one of which is a copy of what was used in Byte a year or so ago.. >I'm a little uncertain about the "buffer ratios" -- I didn't see any >explanation of it in the lenghy aha1542 manual -- do you have anything >that describes them? My "largest" disk IO activity is news processing Nope - I just tweaked and benched, tweaked and benched, etc.. -- Larry Snyder, Northern Star Communications, Notre Dame, IN USA {larry@nstar, uunet!sco!romed!nstar!larry, nstar%larry@iuvax.cs.indiana.edu} backbone usenet newsfeeds available Public Access Unix Site (219) 289-0282 (5 high speed lines)
bill@unixland.uucp (Bill Heiser) (11/06/90)
In article <1990Nov04.222938.984@nstar.uucp> larry@nstar.uucp (Larry Snyder) writes: > >We have a couple of benchmarking tools here on nstar - one of >which is a copy of what was used in Byte a year or so ago.. Do you recommend either of them in particular for ease-of-use, portability (to sysvr3 (esix)), and accuracy? -- home: ...!{uunet,bloom-beacon,esegue}!world!unixland!bill bill@unixland.uucp, bill%unixland.uucp@world.std.com Public Access Unix - Esix SYSVR3 - (508) 655-3848 other: heiser@world.std.com Public Access Unix (617) 739-9753
larry@nstar.uucp (Larry Snyder) (11/06/90)
bill@unixland.uucp (Bill Heiser) writes: >Do you recommend either of them in particular for ease-of-use, portability >(to sysvr3 (esix)), and accuracy? No - it depends on what you are benching ... -- Larry Snyder, Northern Star Communications, Notre Dame, IN USA {larry@nstar, uunet!sco!romed!nstar!larry, nstar%larry@iuvax.cs.indiana.edu} backbone usenet newsfeeds available Public Access Unix Site (219) 289-0282 (5 high speed lines)
karl@robot.in-berlin.de (Karl-P. Huestegge) (11/07/90)
bill@unixland.uucp (Bill Heiser) writes: >larry@nstar.uucp (Larry Snyder) writes: >>what you need to do is to change one item at a time - then >>bench the system, change another - then bench, etc.. >Thanks for the feedback Larry. What utility are you using to do >your benchmarking? >>I have cache enabled (made a 10% difference) and changed the >>buffer ratios from 4 to 16 (made a + difference on large disk IO, >>and - on small; but since most of mine is large - I changed it).. >I'm a little uncertain about the "buffer ratios" -- I didn't see any >explanation of it in the lenghy aha1542 manual -- do you have anything >that describes them? My "largest" disk IO activity is news processing >(Cnews) and compiling software -- does that qualify as "large", and >would you expect it to benefit from the increase in buffer ratios? The Host Adapter has nothing to do with it - it's a SCSI-command. The Buffer-Full Ratio and Buffer-Empty Ratio are part of the Mode Select Command called the 'Dissconnect/Reconnect Control Page'. The Buffer-Full (Buffer-Empty) Ratio indicates on READ (WRITE) operations how full (empty) the buffer should be prior to attempting a reconnection. The Quantum Prodrive Specs say: Buffer-Full Ratio: During READ Operations, the drive disconnects when the buffer is empty. The Value of the buffer-full ratio represents the percentage of the buffer that must be filled, unless the buffer contains all the required data, prior to reconnection of the drive - that is, 255 (FFh) represents 100% full; 128 (80h), 50% full. When the value is zero, the drive determines when it will initiate reselection. The default value is 80h, indicating that the drive will attempt a reconnection whenever the buffer contains at least four kilobytes of data read from the disk. Buffer-Empty Ratio: When the drive is disconnected during WRITE operations, the value of the buffer-empty ratio represents the percentage of the buffer that must be empty to enable fetching data from the initiator, unless the buffer can hold all the required data - that is, 255 (FFh) represents 100% empty, 128 (80h) 50% empty. When the value is zero, the drive determines, when it will initiate reselection. The default value is 80h, indicating that the drive will attempt a reconnection whenever the buffer contains at least four kilobytes of data transferred from the initiator. -- Karl-Peter Huestegge karl@robot.in-berlin.de Berlin Friedenau ..unido!fub!geminix!robot!karl
norsk@sequent.UUCP (Doug Thompson) (11/08/90)
In article <1990Nov3.185524.468@unixland.uucp> bill@unixland.uucp (Bill Heiser) writes: >I ran the SCSICNTL.EXE program to examine my CDC Wren IV (hooked up >to an Adaptec 1542b), and have some questions about the appropriate deleted... sorry for a possibly repeated question, but can someone send a copy of SCSICNTL.exe to me or post a ftp site for it or something? thanks -- Douglas Thompson UUCP: ..{tektronix,ogicse,uunet}!sequent!norsk Internet: norsk@sequent.com "The scientist builds to learn; the engineer learns in order to build." Fred Brooks