1 INFO-VAX	Tue, 12 Sep 2000	Volume 2000 : Issue 510       Contents:! Add SCSI port in existing cluster   Re: apache cluster configuration  Re: apache cluster configuration1 AS4100 w/VMS 7.2-1 console support for DE602-AA?? 5 Re: AS4100 w/VMS 7.2-1 console support for DE602-AA?? 5 Re: AS4100 w/VMS 7.2-1 console support for DE602-AA??  BACKUP priviledges Re: BACKUP priviledges BACKUP priviledges. Re: Barrel rolls, was Re: aircraft are not Sun Re: Big AlphaServer Sale& Re: Booting OpenVMS 7.2 w/o DecWindows& Re: Booting OpenVMS 7.2 w/o DecWindows& Re: Booting OpenVMS 7.2 w/o DecWindows Re: Clustering & multiple sites  CMD Disk Controller  Re: CMD DISK Controller  Re: Export DECterm File Format from SET HOST/LOG ! Re: File Format from SET HOST/LOG # RE: Is there any new Alpha CPU out? + Re: Jupiter (was Re: Q: Why not (2^n)-bit?) + Re: Jupiter (was Re: Q: Why not (2^n)-bit?) + Re: Jupiter (was Re: Q: Why not (2^n)-bit?)  Killfile on Deja Re: LAT printer servers? Re: LAT printer servers?
 MAIL and SMTP " MLM History!  Zig Ziglar Goes MLM!% Re: Off-Topic: DS10 Hardware question % Re: Off-Topic: DS10 Hardware question % Re: Off-Topic: DS10 Hardware question   Re: Safearray in COM for openvms! Re: Sun Hardware problems persist ! Re: Sun Hardware problems persist ! Re: Sun Hardware problems persist ! Re: Sun Hardware problems persist ! Re: Sun Hardware problems persist ! Re: Sun Hardware problems persist ! RE: Sun Hardware problems persist P Re: visual interpretation of algorithms. -> Sound interpretation of    algorythm  F ----------------------------------------------------------------------  % Date: Mon, 11 Sep 2000 20:33:26 +0000 - From: ezzaoudi med <m.ezzaoudi@digitem.co.ma> * Subject: Add SCSI port in existing cluster- Message-ID: <39BD4196.36FE12CE@digitem.co.ma>    Hi  . I have an operationnal two nodes VMS cluster ., The two nodes share the same external disks., With this config the SCSI alloclass is used.  A I want to add external disks drived by two SCSI boards , each one % will be installed in a single server.   H By adding this SCSI bus in cluster , do  I have to use CLUSTERCONFIG.COM  H to  enable the SCSI alloclass port , or have I to edit some config files ?    Thanks.    ------------------------------  % Date: Mon, 11 Sep 2000 14:18:38 -0400 3 From: "Gaitan D'Antoni" <gaitan.dantoni@compaq.com> ) Subject: Re: apache cluster configuration 6 Message-ID: <8pj7nb$ccq$1@mailint03.im.hou.compaq.com>  : There is no need to modify the logical name definitions inF APACHE$STARTUP.COM and there is no need to move APACHE$CONFIG.DAT intoK sys$common. Please undo any changes you may have made to APACHE$STARTUP.COM L and move APACHE$CONFIG.DAT back where it was and delete the copy you made inH sys$common. Once you do this you will need to undefine any logical namesI that may have been created with your changes. This should get you back to I the point where Apache had been configured correctly on SX9071. Now go to K SX9072 and execute APACHE$CONFIG.COM. This should create the necessary data J files and logical names that will allow Apache to operate properly in your cluster.  & Jean-Franois Marchal wrote in message) <8pg539$hog$2@s1.read.news.oleane.net>...  >Bonjour  tous  > : >New apache user, I've just added a new node to my cluster6 >When I look at the definition of apache$root, I get : > F >on SX9071 (first node of the cluster on which I had Apache running) : > F >   "APACHE$ROOT"   = "DSA1:[SYS71.SYSCOMMON.APACHE.SPECIFIC.SX9071.]"& >                   = "APACHE$COMMON:"6 >1  "APACHE$COMMON" = "DSA1:[SYS71.SYSCOMMON.APACHE.]" > 2 >on SX9072 (new node, without any configuration) : > , >   "APACHE$ROOT"   = "DSA1:[SYS72.SYSEXE.]" > B >Theese definitions seem rather strange ... What did I miss duringB >the initial configuration on SX9071, and what should I do, unlessE >I manually modify the logical names' definitions in apache$startup ?  > 
 >Cordialement  >  >Jean-Franois Marchal >X9000 - Lyon (FR) >  >  >  >  >  >    ------------------------------  % Date: Tue, 12 Sep 2000 07:33:21 +0200 > From: "Jean-Franois Marchal" <jean-francois.marchal@x9000.fr>) Subject: Re: apache cluster configuration 2 Message-ID: <8pkfnb$ma5$1@s1.read.news.oleane.net>  ) Done (I had only move apache$config.dat).  Going to check !   Merci 
 Jean-Franois     > "Gaitan D'Antoni" <gaitan.dantoni@compaq.com> wrote in message0 news:8pj7nb$ccq$1@mailint03.im.hou.compaq.com...< > There is no need to modify the logical name definitions inH > APACHE$STARTUP.COM and there is no need to move APACHE$CONFIG.DAT into: > sys$common. Please undo any changes you may have made to APACHE$STARTUP.COMK > and move APACHE$CONFIG.DAT back where it was and delete the copy you made  inJ > sys$common. Once you do this you will need to undefine any logical namesK > that may have been created with your changes. This should get you back to K > the point where Apache had been configured correctly on SX9071. Now go to H > SX9072 and execute APACHE$CONFIG.COM. This should create the necessary dataL > files and logical names that will allow Apache to operate properly in your
 > cluster.   ------------------------------   Date: 11 Sep 2000 16:08:18 PDTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515): Subject: AS4100 w/VMS 7.2-1 console support for DE602-AA??3 Message-ID: <VB87bnLftWIU@mccdev.slac.stanford.edu>   H         We have a pair of Alphaserver  4100 5/400's whose PCI busses areH     "full".   For that and a few other reasons, I'm replacing the  DE500H     single-port 10/100 Ethernet adapter in each with a DE602-AA, gettingH     us a pair of 10/100 ports in a single PCI slot.  So far so good, the1     parts are due on site in a couple of weeks...   H         We're running VMS/Alpha 7.2-1  with  all applicable VMS721 ECO'sH     so  the  DE602's are supported by VMS.  However, the VMS  7.2-1  New1     Features and Release Notes says of the DE602,   C         "OpenVMS displays these devices as EIx0 devices, where x is C         the controller's unit letter.  The device drivers for these A         devices are SYS$EIDRIVER.EXE for the run-time environment /         and SYS$EIBTDRIVER.EXE for LAN booting.   D         The SRM console provides an environment variable for settingB         the mode of operation.  The variable is EIx0_MODE, where xD         is the controller's unit letter.  Only newer Alpha platforms5         will provide console support for these NICs."   H     We have Console Firmware  5.5-5  loaded  on  the AS4100's and a SHOWH     command  lists  only  EWx0_MODE variables, no  EIx0_MODE  variables.H     I've been using the EWA0_MODE set to FastFD  for  the  DE500's.   It.     looks like I can't do that with the DE602.           So three questions:   @          1) What  constitutes  a  "new  Alpha",  that  is, which@             Alphas   support  the  EIx0_MODE  variables  in  the             console?  @          2) Will  a  subsequent  console  firmware  release, say@             coinciding  with  the VMS 7.3 release,  support  the              DE602 on the AS4100?  @          3) Am I correct that I'll need to set the EIA0 and EIB0@             (or whatever letters they  come  up  with) to FastFD@             using LANCP?  Are there any gotcha's in setting this             up?     H         I'd like to hear from anyone  who is successfully using DE602's,     especially in an AS4100.           Thanks, Ken  --  M  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edu :  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------   Date: 11 Sep 2000 23:14:40 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)> Subject: Re: AS4100 w/VMS 7.2-1 console support for DE602-AA??6 Message-ID: <8pjp10$fvu$1@mailint03.im.hou.compaq.com>   In article <VB87bnLftWIU@mccdev.slac.stanford.edu>, Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515) writes:I :        We have a pair of Alphaserver  4100 5/400's whose PCI busses are I :    "full".   For that and a few other reasons, I'm replacing the  DE500 C :    single-port 10/100 Ethernet adapter in each with a DE602-AA...      Hang on, lemme check...   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  # Date: Tue, 12 Sep 2000 01:14:45 GMT = From: system@SendSpamHere.ORG (Brian Schenkenberger, VAXman-) > Subject: Re: AS4100 w/VMS 7.2-1 console support for DE602-AA??0 Message-ID: <009EFFBD.C6203A84@SendSpamHere.ORG>   In article <VB87bnLftWIU@mccdev.slac.stanford.edu>, Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515) writes:I >        We have a pair of Alphaserver  4100 5/400's whose PCI busses are I >    "full".   For that and a few other reasons, I'm replacing the  DE500 I >    single-port 10/100 Ethernet adapter in each with a DE602-AA, getting I >    us a pair of 10/100 ports in a single PCI slot.  So far so good, the 2 >    parts are due on site in a couple of weeks... > I >        We're running VMS/Alpha 7.2-1  with  all applicable VMS721 ECO's I >    so  the  DE602's are supported by VMS.  However, the VMS  7.2-1  New 2 >    Features and Release Notes says of the DE602, > D >        "OpenVMS displays these devices as EIx0 devices, where x isD >        the controller's unit letter.  The device drivers for theseB >        devices are SYS$EIDRIVER.EXE for the run-time environment0 >        and SYS$EIBTDRIVER.EXE for LAN booting. > E >        The SRM console provides an environment variable for setting C >        the mode of operation.  The variable is EIx0_MODE, where x E >        is the controller's unit letter.  Only newer Alpha platforms 6 >        will provide console support for these NICs." > I >    We have Console Firmware  5.5-5  loaded  on  the AS4100's and a SHOW I >    command  lists  only  EWx0_MODE variables, no  EIx0_MODE  variables. I >    I've been using the EWA0_MODE set to FastFD  for  the  DE500's.   It / >    looks like I can't do that with the DE602.  >  >        So three questions: > A >         1) What  constitutes  a  "new  Alpha",  that  is, which A >            Alphas   support  the  EIx0_MODE  variables  in  the  >            console?  > A >         2) Will  a  subsequent  console  firmware  release, say A >            coinciding  with  the VMS 7.3 release,  support  the ! >            DE602 on the AS4100?  > A >         3) Am I correct that I'll need to set the EIA0 and EIB0 A >            (or whatever letters they  come  up  with) to FastFD A >            using LANCP?  Are there any gotcha's in setting this  >            up? >  > I >        I'd like to hear from anyone  who is successfully using DE602's,  >    especially in an AS4100.  >  >        Thanks, Ken >-- N > Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edu; > SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924 ; > Menlo Park, CA  94025           |  FAX:      650-926-3515 O > ----------------------------------------------------------------------------- C > These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   G The newest Compaq NICs NI31** card were filed tested with 7.1-2 and are H supported under V7.2+.  Finding any information about them on the Compaq web site... well, good luck.  $ I'm using the NI3121 in my Alphas.      --O VAXman- OpenVMS APE certification number: AAA-0001     VAXman(at)TMESIS(dot)COM              O city, n., 1. a place where trees are cut down and streets are named after them.    ------------------------------   Date: 11 Sep 2000 21:29:31 GMT# From: system@niuhep.physics.niu.edu  Subject: BACKUP priviledges + Message-ID: <8pjirr$4b5$1@husk.cso.niu.edu>   @ Trying to set up a dedicated backup account with minimal priv.s,* the following is the output of my attempt:   $ set proc/priv=readall * $ backup/image/igno=(inter,label) user1: -/ mkb500:user1_8sep00.bck/sav/ver/log/block=16384 " %MOUNT-I-OPRQST, medium is offlineE %MOUNT-I-OPRQST, Please mount volume USER1_ in device _NIUHEP$MKB500: J BACKUP requests: Saveset USER1_8SEP00.BCK, Volume number 01, write ENABLED3 %MOUNT-I-MOUNTED, BACK01 mounted on _NIUHEP$MKB500: I %MOUNT-I-RQSTDON, operator request canceled; mount completed successfully A %BACKUP-F-PROCINDEX, error processing index file on USER1:, RVN 1 G -SYSTEM-F-NOPRIV, insufficient privilege or object protection violation 8   BACKUP       job terminated at  8-SEP-2000 14:00:19.42   Running VMS 7.2,    " $ dir/sec user1:[000000]indexf.sys   Directory USER1:[000000]  E INDEXF.SYS;1         [SYSTEM]                         (RWED,RWED,RE,)   C I punted and gave teh account all priv.s and it worked, but I would # rather not do that in the long run.    Robert Morphis   ------------------------------   Date: 11 Sep 2000 22:00:59 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman) Subject: Re: BACKUP priviledges 6 Message-ID: <8pjkmr$f5c$2@mailint03.im.hou.compaq.com>  Q In article <8pjirr$4b5$1@husk.cso.niu.edu>, system@niuhep.physics.niu.edu writes: A :Trying to set up a dedicated backup account with minimal priv.s,  .. :$ set proc/priv=readall+ :$ backup/image/igno=(inter,label) user1: - 0 :mkb500:user1_8sep00.bck/sav/ver/log/block=16384 ..B :%BACKUP-F-PROCINDEX, error processing index file on USER1:, RVN 1H :-SYSTEM-F-NOPRIV, insufficient privilege or object protection violation ..     From the BACKUP help text:  A      INTERLOCK        Processes files that otherwise could not be A                       processed due to file access conflicts. Use F                       this option to save or copy files currently openC                       for writing. You must have the user privilege F                       SYSPRV, a system UIC, or ownership of the volume)                       to use this option.    	...  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 11 Sep 2000 20:59:47 -0400 2 From: "Richard B. Gilbert" <DRAGON@compuserve.com> Subject: BACKUP priviledges 7 Message-ID: <200009112100_MC2-B2F2-812D@compuserve.com>   H         You need read access to each file you expect to back up, in thisI case all  of them.  You need control access if you are doing a /RECORD. =   E READALL takes care of this.   As Hoff points out, /IGNORE=3DINTERLOCK J requires SYSPRV (etc.).  Depending of just what else you are doing in you= r J backup command files, you might need OPER and/or VOLPRO.  OPER allows use=  G of REPLY and VOLPRO allows you to mount or  initialize volumes (disk or  tape) that don't belong to you.   F         I generally set up a BACKUP account with (READALL,SYSPRV,OPER,J VOLPRO) as authorized privs and turn them on as needed (and off again whe= n > done).  If you find you need another priv, add it to the list.  > Message text written by INTERNET:system@niuhep.physics.niu.eduA >Trying to set up a dedicated backup account with minimal priv.s, * the following is the output of my attempt:   $ set proc/priv=3Dreadall , $ backup/image/igno=3D(inter,label) user1: -1 mkb500:user1_8sep00.bck/sav/ver/log/block=3D16384 " %MOUNT-I-OPRQST, medium is offlineE %MOUNT-I-OPRQST, Please mount volume USER1_ in device _NIUHEP$MKB500: J BACKUP requests: Saveset USER1_8SEP00.BCK, Volume number 01, write ENABLE= D 3 %MOUNT-I-MOUNTED, BACK01 mounted on _NIUHEP$MKB500: J %MOUNT-I-RQSTDON, operator request canceled; mount completed successfully=  A %BACKUP-F-PROCINDEX, error processing index file on USER1:, RVN 1 G -SYSTEM-F-NOPRIV, insufficient privilege or object protection violation 8   BACKUP       job terminated at  8-SEP-2000 14:00:19.42   Running VMS 7.2, =    " $ dir/sec user1:[000000]indexf.sys   Directory USER1:[000000]  E INDEXF.SYS;1         [SYSTEM]                         (RWED,RWED,RE,)   C I punted and gave teh account all priv.s and it worked, but I would # rather not do that in the long run.2 <	   ------------------------------  % Date: Mon, 11 Sep 2000 13:55:24 -0400 - From: "Peter Weaver" <peter.weaver@stelco.ca>c7 Subject: Re: Barrel rolls, was Re: aircraft are not Sunf. Message-ID: <srq71oqrct628@corp.supernews.com>  6 "Jerry Leslie" <leslie@clio.rice.edu> wrote in message! news:8pgnhj$be9$1@joe.rice.edu...e >...D >  The Barrel Roll is a not competition maneuver.  I The barrel roll is aF >  combination between a loop and a roll.  You complete one loop whileB >  completing one roll at the same time.  The flight path during a barrel@ >  roll has the shape of a horizontal cork screw.  Imagine a big barrel,MF >  with the airplanes wheels rolling along the inside of the barrel in a corkB >  screw path.  During a barrel roll, the pilot experiences always positiveC >  G's.  The maximum is about 2.5 to 3 G, the minimum about 0.5 G."n >...  F Right, I've only done one barrel roll (actually now that I think aboutB it, the instructor demonstrated one and then I did one) since theyF were not allowed in competition I concentrated on the slow rolls. Slow@ rolls are what you see at airshows and usually they happen quiteB quickly and really get the pilot "dancing" with both feet and yourD control hand having to keep busy. IIRC (its been a while since I didA the barrel roll) you just push the stick over to the side and theeE plane does the rest. You may have to add some forward pressure as youeF round the top of the roll, but not as much as in a slow roll. I do notF remember the rudders being used at all. The barrel roll is really veryA forgiving and you do not need to be as careful as in a slow roll.0   -- $! Peter Weaverd8 $ input = "0756475627E277561667562704374756C636F6E23616" $ i = 0u $loop:, $ my_address2[i*4,4]=%X'F$EXTRACT(i,1,input) $ i = i + 1e* $ if i .lt. f$length(input) then goto loop1 $ write sys$output "My address is ''my_address2'"i   ------------------------------    Date: 11 Sep 2000 14:40:25 -0400/ From: jordan@lisa.gemair.com (Jordan Henderson)t! Subject: Re: Big AlphaServer Saled* Message-ID: <8pj8up$fil$1@lisa.gemair.com>  : Yet another installment in the watch Andrew spin, lie and ' misdirect when he's called on his bull.   : To those interested, there were several very good points I; made in the original article, have made repeatedly and willt; continue to make that Andrew saw fit to snip below (without 9 the common courtesy of any indication of a deletion).  Goh9 read my original article that Andrew cut up and wonder atP! what Andrew REFUSES to address...e  ; This is why I must believe that Andrew is in marketing.  Heh# commits lies of omission so deftly.R  * In article <39BD1010.5E6ACB7D@uk.sun.com>,D Andrew Harrison SUNUK Consultancy  <andrew.nospam@uk.sun.com> wrote: >Jordan Henderson wrote: >y- >> In article <39BCFCE6.B8BFC494@uk.sun.com>, G >> Andrew Harrison SUNUK Consultancy  <andrew.nospam@uk.sun.com> wrote:e >> >jlsue wrote: >> >K >> >> On Fri, 08 Sep 2000 12:00:55 +0100, Andrew Harrison SUNUK Consultancye' >> >> <andrew.nospam@uk.sun.com> wrote:a >> >>tQ >> >> >I really hope youe arn't a Compaq stock holder because I doubt that Compaqr2 >> >> >can afford too many of these kinds of deal. >> >> > [snip... more drivel]N >> >>hJ >> >> Look Andrew, you've been called to the floor many times for speaking5 >> >> about that which you know nothing.  Get a clue!e >> >>  >> >, >> >Have I really how very very interesting. >> > >>D >> Sure you have, but you ignore those posts and change the subject. >>A >> Here's a short list of incidents where you've been shown to beA% >> lying or, charitably, misinformed:h >>? >>         * You stated flatly that the Sun E10K Cache problemsX> >>           were repaired MONTHS ago.  Rob was able to dig up( >>           your own words on this one. >> >1B >Nice try but what I said at the time was correct, Sun changed the( >part and used another e-cache supplier. >i  A You said clearly in that post that the problem had been isolated, @ the part replaced and that these reports of ecache problems were@ 'old news'.  But, low and behold, it appears that they are STILL
 happening.  < So, either you were lying, or you have to admit that Sun has9 had very similar problems over a period of 18 months (seem8 referenced article) with two different 400 Mhz/4MB Cache controllers.  = Remember, when you said that this problem was 'old news' back : in December, you were referring to a Gartner advisory from< Nov. 1999.  Gartner is now updating that advisory because of# a continuation of the SAME PROBLEM.v  ? That was just spin back in Dec. 1999, you've been caught making ; misleading statements about the problems being resolved andG) you're trying to spin your way out of it.y   >>? >>         * You have been clearly suggesting that the problemsa= >>           with the E10K Cache are due to static discharge.s@ >>           You should tell Sun Executive VP Shoemaker this, so; >>           he doesn't get caught making such embarrassingu< >>           statements as (from the article that you allude7 >>           to but never give the URL because it's toot" >>           embarrassing to Sun): >> >-< >No I have not, I have said that the vast majority of ecache9 >failures are caused by static. One reason why we changed9> >the part was to introduce one that was less static sensitive. >   : hmmm... that must be why VP Shoemaker lists "environmental= factors" as last below, after all the embarrassing admissions $ of shoddy Engineering on Sun's part.  9 Gosh.  You'd think if the "vast majority" of the failurese: were environmental in nature, then VP Shoemaker would have4 said this and avoided bearing their corporate dirty  laudry in public.t  1 Give us one reason to believe that these failures : are caused by static in the 'vast majority' of cases, just7 one reason...  And remember, an independent source, theM3 Gartner Group says that you should 'challenge Sun'sl! claims of environmental factors'.    >>; >>             According to Shoemaker, Sun hasn't been able : >>             to narrow the problem   to any one specific= >>             cause. Sun believes the problems may have been < >>             caused by a combination of factors, includingB >>             defective components   from one of Sun's suppliers,A >>             poor packaging of the memory chips on   the systeme1 >>             boards, and environmental factors.y >> >AF >Correct but the vast majority of failures have been caused by static,> >changing the environment for example has a huge effect on the= >failure rate and this was reported in the origional article.E >_  7 Baloney, you show me where in the article that it says F: that changing the environment has had a huge effect on the; failure rate.  All they say in the article is that this ONE : customer claims to have seen an improvement after changing the environment.  7 I guess a customer who has good things to say about Suno6 can feel free to speak around the NDA/gag that Sun put on them.  7 What the article actually says is that Sun comes in and,9 changes any number of things, without explaining why, and 9 that problems persist.  Doesn't sound like a problem that 0 can be fixed just by turning up the thermostats.   >> >>tL >> >> You can't read media accounts of things and understand all there is toI >> >> know.  Trust that Compaq is making money, and if not, then you haven2 >> >> something hard and fast to prove your point. >> >>  >> >: >> >So is the announcement on Compaqs web site incorrect ? >> >; >> >I know you can't necessarely trust news reports but you:: >> >would have thought that you could trust one of the two >> >people who signed the deal.- >> >0 >> >The Compaq WEB site states that the deal for1 >> >375 GS320's with 12000 CPU's in total runningo/ >> >at 1250 Mhz +600 TB of storage is worth 200- >> >million dollars. >> >1 >> >I very much doubt that Compaq is lying do youH >> >or do you dispute this.4 >> >2 >> >Someone has allready posted a list pricing for. >> >a 4 CPU GS320 of ~900,000 dollars and this2 >> >would mean that Compaq were discounting by 38% >> >of the current list price. >>2 >> 38% off of list?  That's not at all unusual for >> such volume purchases.A >> >v5 >Do you have some really major comprehension problem.F  6 I do not.  You didn't state your updated 80% discounts in the above quoted article. n  7 >The person who quoted a price for a GS320 did it for ac0 >4 CPU machine and Compaq would have been giving3 >the DOE 38% discount if all 375 machines containede1 >only 4 CPU's. But each machine contains 32 CPU'sa8 >so the actual discount is well over 80%. 375 * 4 = 15002 >which is only 10500 CPU's short of the 12000 that >the DOE system requires.e  5 I don't know what goes into such purchases, really I p. don't.  Companies often sell early machines at4 substantial discounts to get them fielded.  This may0 be what's at work here.  Are you saying that the1 first few E10K customers paid anywhere near list?w  0 Why don't you go look up those numbers, Andrew, 1 for comparison?  I bet it would be interesting...t  / Uhhh... Unless those numbers aren't made public0, like the steep discounts that Sun must have & made to eBay to keep them in the fold.   >a: >And TPC-C is a very good place to get list prices because& >they have to be audited and accurate. >o >Regards >Andrew Harrison >Enterprise IT Architect >i >  >    -Jordan Henderson' jordan@greenapple.com'   ------------------------------  % Date: Mon, 11 Sep 2000 13:10:15 -0700e0 From: Mark Berryman <Mark.Berryman@Mvb.Saic.Com>/ Subject: Re: Booting OpenVMS 7.2 w/o DecWindows , Message-ID: <39BCD9B7.40D9A7DF@Mvb.Saic.Com>   Hoff Hoffman wrote:t > r > In article <39b934af.524144494f47414741@radiogaga.harz.de>, martin@radiogaga.harz.de (Martin Vorlaender) writes:5 > :Vance Haemmerle (vance@toyvax.Tucson.AZ.US) wrote:tI > ::   I just tried that today on a VAX with VMS 7.1 (the first command).xJ > :: DecWindows still started up.  There was a BADVALUE error and it reset > :: WINDOW_SYSTEM to 1. > :vM > :Does anyone besides me sees this as one of the worst jokes VMS engineering  > :has ever played on us?d > :OF > :User: I don't want DECwindows to start: SYSGEN> SET WINDOW_SYSTEM 0 > : 7 > :DECwindows: What's this? No me? That can't be right.t) > :            Better correct that one...  > ( >   User: Why isn't DECwindows starting? > # >   CSC: Did you set WINDOW_SYSTEM?t > " >   Ya just can't win sometimes...   Sure ya can...  1 WINDOW_SYSTEM = 0     ;VMS will decide what to do B WINDOW_SYSTEM = 1     ;I really know what I want, start DECwindows
 regardless> WINDOW_SYSTEM = 2     ;I really know what I want, do NOT start
 DECwindows  G In other words, simply add a state where VMS *knows* the system managerhC has chosen not to start DECwindows and doesn't have to assume (s)het simply forgot.  
 Mark BerrymanF Mark.Berryman@Mvb.Saic.Com   ------------------------------   Date: 11 Sep 2000 22:03:12 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)/ Subject: Re: Booting OpenVMS 7.2 w/o DecWindowsS6 Message-ID: <8pjkr0$f5c$3@mailint03.im.hou.compaq.com>  _ In article <39BCD9B7.40D9A7DF@Mvb.Saic.Com>, Mark Berryman <Mark.Berryman@Mvb.Saic.Com> writes:  :Hoff Hoffman wrote: ..# :>   Ya just can't win sometimes...F :  :Sure ya can...  : 2 :WINDOW_SYSTEM = 0     ;VMS will decide what to doC :WINDOW_SYSTEM = 1     ;I really know what I want, start DECwindowsa :regardlesst? :WINDOW_SYSTEM = 2     ;I really know what I want, do NOT startt :DECwindows  : H :In other words, simply add a state where VMS *knows* the system managerD :has chosen not to start DECwindows and doesn't have to assume (s)he :simply forgot.m  I   While I certainly understand the intent, you just started up VWS... :-)   N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 11 Sep 2000 16:39:40 -0700s0 From: Mark Berryman <Mark.Berryman@Mvb.Saic.Com>/ Subject: Re: Booting OpenVMS 7.2 w/o DecWindowsl, Message-ID: <39BD0ACC.153C16F5@Mvb.Saic.Com>   Hoff Hoffman wrote:i > a > In article <39BCD9B7.40D9A7DF@Mvb.Saic.Com>, Mark Berryman <Mark.Berryman@Mvb.Saic.Com> writes:  > :Hoff Hoffman wrote: > ..% > :>   Ya just can't win sometimes...o > :U > :Sure ya can...U > :K4 > :WINDOW_SYSTEM = 0     ;VMS will decide what to doE > :WINDOW_SYSTEM = 1     ;I really know what I want, start DECwindowsE
 > :regardlessUA > :WINDOW_SYSTEM = 2     ;I really know what I want, do NOT startl
 > :DECwindowsR > :EJ > :In other words, simply add a state where VMS *knows* the system managerF > :has chosen not to start DECwindows and doesn't have to assume (s)he > :simply forgot.0 > K >   While I certainly understand the intent, you just started up VWS... :-)t  G VWS on my Alpha?  Ooh, I like that.  So, in the next release of the VMS  documentation we will read:a  E     Leaving WINDOW_SYSTEM set to its default value of zero will cause  VMSiC     to execute the DWIM algorithm and startup VWS or DECwindows, as  appropriate.  # Hot Dog!  DEClander here I come....a  D (tongue firmly in cheek in case there are any too-literal minded out there)  
 Mark Berrymann Mark.Berryman@Mvb.Saic.Com   ------------------------------  # Date: Mon, 11 Sep 2000 20:26:33 GMTr From: kparris@my-deja.com ( Subject: Re: Clustering & multiple sites) Message-ID: <8pjf4t$87a$1@nnrp1.deja.com>u  E If you have GIGAswitches, you definitely want the ClearViSN software.gD It only costs a few hundred bucks, and is a great help in setting upF the switches and for things like easily viewing the bandwidth utilizedG during a full-copy in the form of a bar-graph.  Then you can always use 4 the MIBs with OpenView to monitor for the long-term.  F I'm in the process of setting up a new disaster-tolerant cluster rightD now.  We're using Cisco Catalyst 6500 switching bridges with GigabitF Ethernet (GbE) on one side, with ol' reliable GIGAswitches and FDDI asF the other LAN rail.  When your target is very-high availability, usingF all-new equipment choices for a new cluster would be a very high risk.F In addition to the LAN / cluster interconnect, we're being cautious inE other areas as well: shadowing between ol' reliable HSJ50 controllers-> and new HSJ80 controllers, and having ol' reliable GS140 boxesD alongside our new GS160 Wildfire boxes.  Once the newer products are/ proven, we'll upgrade from the older equipment.@  C I've been told by people whom I have a lot of respect for that whent> Ciscos get overloaded, they respond by dropping small packets.E PEDRIVER Hello messages happen to be small packets.  So I'll be usingfD the SYS$EXAMPLES:LAVC$FAILURE_ANALYSIS program and monitoring to seeD what happens when the going gets tough.  I feel a lot better knowing( FDDI is there to fall back on if needed.  C In this configuration, PEDRIVER does indeed see GbE as having lowermF latency than FDDI, so, all else being equal, it prefers GbE over FDDI.E Since there is not yet support for GbE's Jumbo Packets under VMS, you G must use standard Ethernet-sized packets -- i.e. set NISCS_MAX_PKTSZ tofG 1498 -- because if you set NISCS_MAX_PKTSZ to 4468 to allow FDDI to use C larger packets, then PEDRIVER will prefer FDDI over GbE.  Note that E this also implies that you will have higher host-CPU overhead for SCS C traffic, particularly MSCP-serving, with GbE (or Fast Ethernet, forlF that matter) than you would with FDDI, since with the smaller packets,D there will be more packets for the hosts to process to move the same amount of block-transfer data.  G I used the programs LOCKTIME_1 and LOCKTIME_3 (which measure round-trip=F lock request latency) written by Roy.G.Davis@Compaq.com (of VAXcluster> Principles fame) and got the following results for simple lock. conversions (which take exactly 1 round trip):9    GS160:  GbE: 240 microseconds   FDDI: 360 microsecondse9    GS140:  GbE: 230 microseconds   FDDI: 270 microsecondsi@ (This is with zero inter-site distance, as both "sites" for this8 cluster are presently staged in the same computer room.)  E With regard to shadow copy times, there are various factors involved. D Bandwidth is obviously a factor, but due to the sequential nature ofH VMS Volume Shadowing full-copy operations (which carefully avoid double-@ buffering and sequentially copy 127-block segments one at a timeE starting at the beginning and continuing to the end of the disk), the E size of the shadowed volumes can affect the shadow-copy time.  I onceLG cut shadow-copy time in half by eliminating controller-based stripesetseD and instead presenting the individual disks to VMS, then using host-C based RAID software to stripe across shadowsets of these individualcB disks to form RAID 0+1 arrays.  Shadow-copying these smaller unitsC (even though there were many more of them) in parallel proved to be D twice as fast in elapsed time.  So think carefully about whether you; really want to use those new 36 GB drives (or else consider.D partitioning them into smaller pieces; in a recent test I found thatC dividing a drive into 4 partitions at the controller level and then F shadow-copying all 4 in parallel cut the shadow-copy time to less thanC half that of the time required to shadow-copy the disk as a whole).   B To minimize shadow-copy times, you'll want write-back cache in theE controllers to minimize the latency of each of those 127-block shadowcE server writes on the target side.  Read-ahead cache in the controller D (a la HSJ80) helps on the source side (although most SCSI disks haveG track read-ahead buffers, which should help to some extent, and for thee? same reasons, if your controllers can't do read-ahead caching).h  F The MSCP Shadowing Assists (particularly Write-History Logging) on theF HSJ series of controllers allow you to avoid the shadowing full-mergesD that occur with HSZ and HSG controllers whenever a VMS node crashes.E Full merges present roughly the same bandwidth load on the inter-site E link(s) as a full-copy.  You can't predict or control when a VMS node D will crash.  And full-merges typically take longer than a full-copy,D and in my experience, can actually have a more-detrimental effect onE application I/O performance than a full-copy, should your applicatione be sensitive to that.g  ? If you have redundant links, be sure they either are completelyeG separate LANs (NOT bridged together), or else that you've carefully sethC up the bridges' Root Priority values correctly so that the SpanningeG Tree algorithm doesn't turn off any of your inter-site links.  No sensewF paying big bucks for a high-bandwidth inter-site link, only to have it+ sit idly by waiting for a failure to occur.o  C Shadow copies go faster if you perform them on a node on the targethE side.  In some versions of VMS, the copy tends to be performed by therC node which does the Mount command that initiates the copy; in otheraD versions, the node selection seems arbitrary and you may have to setG the dynamic SYSGEN parameter SHADOW_MAX_COPY to zero temporarily on theIC nodes where you do NOT want the copy to be performed, do the mount,CG wait a moment for one of the preferred nodes to pick up the shadow-copy>F thread from the queue, and then set SHADOW_MAX_COPY back to its normal value everywhere.o  > Here are a couple of real-world examples of shadow-copy times:  > Dual 45-megabit DS-3 links between GIGAswitches, 130-mile siteB separation distance, 252-mile circuit path length (4-5 millisecond< round-trip time), about 400 GB shadowed between sites, HSJ50H controllers with write-back cache: 4-5 hour shadowing full-copy time; 2- 3 second mini-merge time.e  G Dual 100-megabit ATM links between GIGAswitches (ATM does 155 Mbits butg@ the GIGAswitch line card can only handle 100 Mbits), 6-mile siteE separation distance (500 microsecond round-trip time), roughly 120 GB,D shadowed, HSZ80 controllers with WBC: about 1.5 hours full-copy time( and 1.5 hours (no-load) full-merge time.G ----------------------------------------------------------------------- G Keith Parris|Integrity Computing,Inc.|parris@decuserve.decus.org-nospamgF VMS Consulting: Clusters, Perf., Alpha porting, Storage&I/O, Internals    & Sent via Deja.com http://www.deja.com/ Before you buy.r   ------------------------------  % Date: Tue, 12 Sep 2000 12:09:36 +1200o. From: Nivlesh Chandra <NChandra001@itc.gov.fj> Subject: CMD Disk ControllerO Message-ID: <791C2856E8FDD211BAFB0008C759919501A628D8@exchange01.govnet.gov.fj>o   > Hello. > I > We have CMD disks that are served off a controller. One of the disks is.E > unmountable and it comes up with fatal drive error. I would like toD" > connect to the controller using G > set host but do not know the task-name that I should connect to sinceaB > there is no manuals for this here. Can someone please help me... >  >  > Niv9   ------------------------------  % Date: Tue, 12 Sep 2000 12:18:38 +1200g. From: Nivlesh Chandra <NChandra001@itc.gov.fj>  Subject: Re: CMD DISK ControllerO Message-ID: <791C2856E8FDD211BAFB0008C759919501A628D9@exchange01.govnet.gov.fj>e   Hello once again...h  E I guess you all must have already read the previous mail that I had =  sent to C this group just a few wee minutes back.. well this was my prob..=20eF we have a CMD DIsk set which houses 4 disks... now what had happened = waseE that one of the disks became unmountable .. coming up with mounVerify  timeout... thuseI I was trying to set host to the controller for this so that I could see =  ifG there is anything I might be able to do it get it up... but the thing =e is? that there is no manuals for this...that is to connect to the =e
 controller...a   I tried using the followinga  @ $ SYSMAN IO CONNECT  FYA0: /NOADAPTR /DRIVER_NAME=3DSYS$FYDRIVERA $ SET HOST /DUP/SERVER=3DMSCP$DUP /TASK=3Dtask-name   server-namee    but I did not know the task-name  H But guess what .. I have found out how to get alist of all the utility = progs F that one can connect to on the controller... well I guess all of you = mustB have already found this .. but for a newbie like me .. I guess the% enthusiasm is too much to contain ...   D just use DIRECT as the task-name and it lists all the progs that are
 running...    6 $ SET HOST /DUP/SERVER=3DMSCP$DUP /TASK=3DDIRECT  CMD3   and the output was? %HSCPAD-I-LOCPROGEXE, Local program executing - type ^\ to exite =20s' Copyright =A9 1994 CMD TECHNOLOGY, INC.           CDI-4000 Serial No: 4368!         Firmware Rev. B4   (X47 )c =20c$ DIRECT V1.0  D  Aug 17 1995 14:32:12$ PARAMS V1.0  D  Aug 17 1995 14:32:12$ UTILIT V1.0  D  Aug 17 1995 14:32:12 =20  End of directory =20nB %HSCPAD-S-REMPGMEND, Remote program terminated - message number 1.. %HSCPAD-S-END, Control returned to node ITCS10     Enjoy!!x   Niva   ------------------------------   Date: 11 Sep 2000 14:19:07 PDTT From: Fairfield@SLAC.Stanford.EDU (Ken Fairfield; SLAC: 650-926-2924; FAX: 926-3515) Subject: Re: Export DECterma3 Message-ID: <WtoCUJ8Hkypz@mccdev.slac.stanford.edu>a  0 In article <39BD0EAB.4A06AE85@cyberway.com.sg>, )     laukk <laukk@cyberway.com.sg> writes:  > Hi,iG >  I need to launch a DECterm from a VMS system onto a VXT terminal.  > TheeG > DECterm is required to run some DCL commands automatically once it isaC > launched onto the VXT terminal and return to VMS prompt after the > > commands are executed . I have tried the following command : > 6 >  $set display/create/node=xx.xx.xx.xx/trans=tcpip) >  $create/term/detach/input=input.come > G > The DECterm exit immediately after those DCL commands contains in thet > input.com file was executed. > D > What can be done to make the DECterm to return to VMS prompt after  > execution of the DCL commands?  H         There are probably several ways,  some easier than others.  NoteH     that  one of the drawbacks of using an /INPUT file on CREATE/TERM isH     that the command file runs in a subprocess, so  there  is  potentialH     for   insufficient  resources  available.   For  simple  things,  it     probably doesn't matter.  H         One possibility is to  put  some  code  in the target username'sH     LOGIN.COM  that executes the desired commands when the  DECW$DISPLAYH     points to that particular VXT.   However,  this  doesn't  give  muchH     chance  for  executing  _various_  different  commands  on different     logins.t  H         Another possibility, one which  we've  used  _extensively_ in myH     cluster,  is to use a shared logical name table which stores variousH     bits of information and which is  queried  during  our  SYS$SYLOGIN.H     That  could be LNM$GROUP, but we used a dedicated logical name tableH     which is created during system startup.   Once you've got the sharedH     name  table, when you want to create the DECterm, you first define aH     logical name that contains the command you want executed,  then  youH     create  the DECterm with /LOGICAL which will _return_ the FTA deviceH     created, then you use  that  FTA  device  returned to define anotherH     logical  in  the shared name table which is translated  during  yourH     SYS$SYLOGIN (or the user's LOGIN)  and  points  the  the  name  thatH     contains the command to execute.  If there's no matching FTA logicalH     name, you do a normal login.  If there is a match, you translate theH     appropriate  logicals  and execute the command.  At that point, it'sH     up to  your  SYS$SYLOGIN  whether  you  LOGOUT  after  executing the0     commands, or just exit the procedure to DCL.  H         [The last paragraph is from memory...I haven't looked at this inH     detail for a couple of  years.   If anyone wants the details, e-mailG     me and I'll review what we do and provide a more complete picture.]o           -Ken -- 1M  Kenneth H. Fairfield            |  Internet: Fairfield@SLC.Slac.Stanford.Edug:  SLAC, 2575 Sand Hill Rd, MS 46  |  Voice:    650-926-2924:  Menlo Park, CA  94025           |  FAX:      650-926-3515N  -----------------------------------------------------------------------------B  These opinions are mine, not SLAC's, Stanford's, nor the DOE's...   ------------------------------  % Date: Mon, 11 Sep 2000 21:50:37 +0100m  From: steven.reece@quintiles.com& Subject: File Format from SET HOST/LOGH Message-ID: <OFB04F2F8E.CB849BE8-ON80256957.0071F13F@qedi.quintiles.com>  K Having logged a VMS install earlier today using SET HOST/LOG 0 and then SETOI HOST/DTE to go from a VAXstation to the serial port on an Alpha using the I TTA2: port on the VAXstation, I've now got a file with lots of CRLF pairs G and with lines that were all nice and neat on the original screen split?$ over multiple lines in the log file. Like, for example :    >>>i so hi o.   d  ee v  <CR> <LF> dW ka0.0.0.2000.0 DKA0w       Seagate ST32550N    = What I'd like to do is to tidy the file up so that it gives :    >>> sho devD) dka0.0.0.2000.0     DKA0 Seagate ST32550Nt  F so that I do not have to edit the individual lines before printing the
 stuff out.   What should I be doing?n   Steve.   ------------------------------   Date: 11 Sep 2000 21:57:34 GMT2 From: hoffman@xdelta.zko.dec.nospam (Hoff Hoffman)* Subject: Re: File Format from SET HOST/LOG6 Message-ID: <8pjkge$f5c$1@mailint03.im.hou.compaq.com>  k In article <OFB04F2F8E.CB849BE8-ON80256957.0071F13F@qedi.quintiles.com>, steven.reece@quintiles.com writes:.< :What I'd like to do is to tidy the [SETHOST.LOG] file up...  H   One of the locals has a tool that cleans up the SETHOST.LOG file, I'm G   trying to convince him to contribute it to the next OpenVMS Freeware.=  H   I will assume you know that you can redirect the SRM console output toF   a FAT floppy.  (If not, please see the OpenVMS FAQ section ALPHA20.)  N  --------------------------- pure personal opinion ---------------------------L    Hoff (Stephen) Hoffman   OpenVMS Engineering   hoffman#xdelta.zko.dec.com   ------------------------------  % Date: Mon, 11 Sep 2000 23:41:51 -0400i+ From: "Main, Kerry" <Kerry.Main@compaq.com>p, Subject: RE: Is there any new Alpha CPU out?J Message-ID: <910612C07BCAD1119AF40000F86AF0D8052847A4@kaoexc4.kao.dec.com>   Andrew,n   You are a great straight man ..c  I >> You may well have info that Oracle are optimising their DBMS for Tru64sI and OpenVMS and for all I know it's true however you have an awfully longeK way to go. On the basis of the current Oracle benchmark results this tuningfI needs to improve the throughput on the Alpha systems by a factor of 2.5 xe? just to get it into a possition of parity with Sun and IBM. <<<t  I Checkout this press release using TPC-C - GS Series is now single fastest # Oracle server based on TPC numbers.l8 http://www.compaq.com/newsroom/pr/2000/pr2000091103.html; http://www.ideasinternational.com/benchmark/tpc/tpccnc.htmlY  F [how long will it last? Who knows, but so much for your tuning fud ..]  J >>> Hey Kerry do you know the difference between a commercial DBMS app andL the kind of application that DOE want to run. The DOE apps are almost all FPK and many of them fall into the embarrasingly parallel category where thingsrI like MPI scale really really well they could not be more different from ar DBMS base OLTP app.>>>  H Andrew, Andrew ... do you think the Compaq stockholders or analysts care< what the application is that this Customer is going to run?   5 Or do you think they are looking at the revenue $'s?    J Or perhaps at the press of Alpha being chosen to be the base of one of theL premier technical computing platforms to be developed in the next few years.  L [now this is where you drag out your 80% discount fantasy stories - btw, theJ basis of this pure speculation ie. using a single system list pricing from8 TPC to guess what discounts were provided had me rotfl.]     Regards,  
 Kerry Main Senior Consultant,
 Compaq Canadat Professional Servicesa Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.comi       -----Original Message-----' From: Andrew Harrison SUNUK Consultancy.! [mailto:andrew.nospam@uk.sun.com]s* Sent: Tuesday, September 05, 2000 10:54 AM To: Info-VAX@Mvb.Saic.Come, Subject: Re: Is there any new Alpha CPU out?     "Main, Kerry" wrote:   > Andrew, Andrew ... >dF > Note - you seem really high on using Oracle benchmarks. Thats great, becauseFG > based on some recent tidbits I heard, there is some really good stuffuL > happening with Oracle being much more optimized for the Alpha (OpenVMS and > Tru64) architecture..c >t  B No Kerry I used Oracle Applications as an example because that was? the benchmark used by Compaq (you know the people you work for)lJ to justify the claim that the GS320 was the fastest RISC server available.E The GS320 result was the one used in the origional launch performancer white paper for the GS systems.   F You may well have info that Oracle are optimising their DBMS for Tru64D and OpenVMS and for all I know it's true however you have an awfullyI long way to go. On the basis of the current Oracle benchmark results thisNG tuning needs to improve the throughput on the Alpha systems by a factoruF of 2.5 x just to get it into a possition of parity with Sun and IBM. I assumeJ that this tuning will be to make Oracle kinder to NUMA systems and if thisE is the case then it is almost certainly targeted at SuperDome and thehC next generation of IBM servers both of which are also NUMA boxes asuB well as the obvious impact it would have for existing NUMA systems% like the Origin 3000 and the Sequent.T  C BTW I take your comment as being an admission that Oracle currentlyeF underperforms on the WildFires and that to get any sort of performanceE equivalent to the Compaq hype customers are going to need to wait forh a new version of Oracle.   >hE > Perhaps this is because of all the attention Alpha has been gettingh recentlyL > ? Perhaps its because the load on many Customers is going through the roof5 > and they absolutely need very high end performance?M >  > :-)m  H Perhaps this is the case for FP based systems, there is however preciousJ little informations about really big commercial wins for the GS320, I meanF wins not upgrades BTW because Alpha Servers cannot survive by churning the installed base.3   >c >eH > re: Oracle pricing .. Andrew, you stated that Oracle charges less than Alpha I > systems for the lower speed Sun and HP boxes. Thats very true. However,t anyiG > comments on what Oracle is planning to charge for new SPARC III based 
 > systems? >   H Well of course Kerry the UltraIII will cost will be based on the current Oracle  I pricing model which is clock rate * 1.5 (RISC factor) * Number of CPU's *r unit price.H I don't see how this can be of any comfort for you. Remeber it isn't the fact thatI the Compaq CPU's have a high clock rate thats an issue, its the fact that  theyL don't deliver the expected performance at that clock rate thats the problem.  A > re: your comments about Wildfire systems being slow and late ..  >nJ > Late, yes. Slow, nope. Nice try. The TPC benchmark you keep referring to wascL > an initial one that was put up for the announcement. As I am sure you knowF > Andrew, it takes time to optimize the compilers to the fullest. Stay tuned. > :-)t  H Sure Kerry so how come people on this group and very knowledgable peopleJ at that were predicting that the TPC-C results for the GS160 were going to beJ significantly higher than they are. The particular person (I am not naming himdK to spare his blushes) was specific about a publication date and an expectednK performance. He was pretty much correct about the date and very wrong aboutmI the result. Now where did he get the number, I respect his postings and I  wouldm  J be very suprised if they were made up, so they must have come from Compaq.  J So what happened, did you find that NUMA boxes particularly ones that haveK the kind of local/remote latency as the WildFire are much more difficult toh tune than you thought.  L I take this as an admission that the WildFire is not the fastest RISC serverJ available and won't be untill some more tuning is done. I hope this tuningL does not fall into the Sequent category BTW, OPS in a box with the customersI paying to tune the system to get to a level of performance somewhat belownK what they had been let to expect from the brochures and the "white papers".   H If for example Compaq publish a TPC-C result based on the Sequent OPS in: a box scheme then you will look as ridiculous as they did.  , So late yes, slow yes by your own admission.  L > It is also why I do not expect SPARC III TPC numbers for a good time after > it is introduced either. >vL > I think the first url above for a DOE order for 375 GS320 Wildfire systemsE > (approximately 12,000 Alpha cpus's) and 600Tb of storage speaks fort itself.  >u > Slow and late? >t  F Hey Kerry do you know the difference between a commercial DBMS app andF  the kind of application that DOE want to run. The DOE apps are almostK all FP and many of them fall into the embarrasingly parallel category whereoI things like MPI scale really really well they could not be more differentn from a DBMS base OLTP app.  F Nor do I think that oracle licensing costs were uppermost in the DOE'sC mind though I can tell you they are in the minds of your commercialu
 customers.   Regardsy Andrew Harrisonm Enterprise IT Architecte   ------------------------------    Date: 11 Sep 2000 15:51:43 -07003 From: Eric Smith <eric-no-spam-for-me@brouhaha.com>p4 Subject: Re: Jupiter (was Re: Q: Why not (2^n)-bit?)0 Message-ID: <qh7l8ik0e8.fsf@ruckus.brouhaha.com>  ' hg/jb <shsrms@bellatlantic.net> writes:a2 > Address path was 30bits - I won't get into that.  @ If it supported 30-bit physical addresses, I'd very much like to: know the details of the changes to the page table entries.   Cheers,n Eric   ------------------------------  # Date: Mon, 11 Sep 2000 23:32:05 GMTe% From: hg/jb <shsrms@bellatlantic.net>p4 Subject: Re: Jupiter (was Re: Q: Why not (2^n)-bit?)0 Message-ID: <39BD6B9A.22B21A35@bellatlantic.net>   Eric Smith wrote:w > ) > hg/jb <shsrms@bellatlantic.net> writes:l4 > > Address path was 30bits - I won't get into that. > B > If it supported 30-bit physical addresses, I'd very much like to< > know the details of the changes to the page table entries.B As Barb has said the software lagged the design, probably rather aE good thing too.  I do not recall the impact, I know we were focussingo on Fortran first.s bobg   > 	 > Cheers,, > Eric   ------------------------------  # Date: Mon, 11 Sep 2000 23:37:47 GMT/! From: Mark nospam@garetech.com.au-4 Subject: Re: Jupiter (was Re: Q: Why not (2^n)-bit?)0 Message-ID: <B5E3A85D.8BAF%gcs@s054.aone.net.au>  < in article qh7l8ik0e8.fsf@ruckus.brouhaha.com, Eric Smith at; eric-no-spam-for-me@brouhaha.com wrote on 12/09/2000 08:51:.  ) > hg/jb <shsrms@bellatlantic.net> writes: 3 >> Address path was 30bits - I won't get into that.  >cB > If it supported 30-bit physical addresses, I'd very much like to< > know the details of the changes to the page table entries. > 	 > Cheers,e > Eric   Hi Eric,L    vir address 30 bits is not a problem all the  pointers in memory had roomK for 30 bits go look. At the phy lvl the core page pointer only has room forrF 27 bit phy with out modifying the old format bits 18-22 in KL MBZ. TheH problem is how TOPS now deals with 4096 section pointers in UPT/EPT. XKLK went for some method using "Super Section" pointers in UPT/EPT and a seconde. lvl of indirection ( I think not an XKL guru).    
     Cheers         Mark :)a   ------------------------------  # Date: Tue, 12 Sep 2000 01:02:07 GMTi From: jgessling@yahoo.comd Subject: Killfile on Dejai) Message-ID: <8pjva3$rhg$1@nnrp1.deja.com>u  ? Anybody know how to set up a kill file on Deja?  I'm reduced tolD using that service after remarq closed down and not sure where to go# somewhere else but I would like to.n  D In the meantime, this you said/he said stuff has gone on long enoughG for me.  If your equipment is so crappy that you have to argue over the C environmental manual then I don't need to hear about it.  Vendors I3D deal with support their kit, without question, wherever it is.  I'veD seen plenty of kit operated out of spec and just fine too, and fixed4 without any question if it failed in that situation.   Again, killfile?   Jim     & Sent via Deja.com http://www.deja.com/ Before you buy.i   ------------------------------  % Date: Mon, 11 Sep 2000 22:20:51 +0100e5 From: "Adrian Lumsden" <A.Lumsden@spamtrap.xdt.co.uk>t! Subject: Re: LAT printer servers?U/ Message-ID: <8pjioa$9qd$1@newsg2.svr.pol.co.uk>h  E I use a couple of Lantronix LPS1-T Micro Print Servers. They have LATe	 build in.   > They have 10baseT and a parallel port. Seem to work OK for me.  H I've often wondered if one were to put an RS232 to Parallel converter onA the parallel port, would that make a single port terminal server?    regards,   Adrian   --( Adrian Lumsden, XDT Computer Systems, UK" A dot Lumsden at xdt dot co dot uk      . Jim Agnew <agnew@hsc.vcu.edu> wrote in message% news:39BD0086.EBA1B1C5@hsc.vcu.edu...-F > we're running a Lantronix that has lat, we don't use the lat, but it doesF > have it, the model eps2 printserver.. very nice, even uses decserver 500 commands.. >r> > only a very satisfied user.. it even does vms and windows nt simultanelously also.8 >y > jimc >s > Paul Repacholi wrote:b > >.9 > > Are there any LAT capable servers available out theref: > > now days? H-Ps support when asked about Jetdirects was1 > > "What's LAT?" From their networking expert...c > >i9 > > Oh, I do know about the terminal servers. Par port isp< > > whats needed. .ps bitmaps are not serial port fiendly at > > any speed! > >  > > --@ > > Paul Repacholi                               1 Crescent Rd.,; > > +61 (08) 9257-1001                           Kalamunda.tD > >                                              West Australia 60762 > > Raw, Cooked or Well-done, it's all half baked.   ------------------------------  # Date: Tue, 12 Sep 2000 03:11:49 GMT + From: rjordan@mars.mcs.net (Richard Jordan)o! Subject: Re: LAT printer servers?d2 Message-ID: <V9hv5.384$z77.49519@news.goodnet.com>  H We have an Emulex NetJet MIO card in a LaserJet 5 and an external NetQUEH driving a LaserJet IID (its a tank!) via parallel.  They support Telnet,F LAT, Novell, Netbios, and supposedly LPR/LPD (never tried it).  I alsoH have a NetQUE at home driving a DEC inkjet printer via parallel, sharingF access between my home cluster, a Mac, and a skanky NT box (that at isB at least not a skanky intel NT box).  Works great.  But I've never( done anything withPostscript printing...   Rich Jordan  rjordan@mcs.nets   ------------------------------  # Date: Tue, 12 Sep 2000 03:48:44 GMT  From: deja@chovav.comm Subject: MAIL and SMTP) Message-ID: <8pk92l$74p$1@nnrp1.deja.com>   E I have a VAXstation on our internal network. Normally we use ExchangeuE to send/receive mail for the PC. I want to send mail dirctly from thesA VAX. How can I tell my VMS that the SMTP gateway is the real one?t   VMS 7.2t    & Sent via Deja.com http://www.deja.com/ Before you buy.s   ------------------------------  + Date: Mon, 11 Sep 2000 15:59:15 -0700 (PDT)  From: opp2000@usenm.netT+ Subject: MLM History!  Zig Ziglar Goes MLM!A; Message-ID: <200009112259.e8BMxFi21656@smtp1.avistacom.net>a  ^ |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |( THE SINGLE BIGGEST EVENT IN MLM HISTORY!^ |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |  |    MLM Veterans - DO NOT MISS THIS!   Zig Ziglar is going MLM. d  ' Zig Ziglar has pre-launched his own new 5 network Company - Official Launch date: January 2001.e  + We are recruiting for leaders who instantlye7 recognize the GLOBAL potential of Zig Ziglar going MLM.   D This is going to be the biggest thing in Network Marketing History.   * THOUSANDS are Joining Right Now.... HURRY!  3 For information and consideration Send an Email to:   % mailto:autoreply@ethos.st?subject=ZZNh  # Note:  Please make sure the lettersa     -    ZZN    -  f   are in the Subject Field.   ; Click here and send:  mailto:autoreply@ethos.st?subject=ZZN    ---------------------------------------------------------------------------------------------------------------------------------r     Please note:8 PS -  THIS IS NOT A SPAM.  You have received this email C because you have either sent us your offer or requested informationqJ from us in the past thereby adding yourself on our in-house mailing list. # We do not sell our lists to anyone.u  F Removal Instructions:  Reply to this Email and type  "Remove"   in the8 Subject Field.   mailto:opp2000@usenm.net?subject=Remove   ---------------------------------------------------------------------------------------------------------------------------------    ------------------------------  % Date: Mon, 11 Sep 2000 17:18:42 -0400n* From: "Paul A. Jacobi" <nospan@nospam.com>. Subject: Re: Off-Topic: DS10 Hardware question, Message-ID: <8pjia7$2gqg$1@lead.zk3.dec.com>  B "David J. Dachtera" <djesys.nospam@earthlink.net> wrote in message' news:39B9B5EE.683BBD0C@earthlink.net...e  H > The problem is, of course, that many of the newest "commodity" devicesJ > are coming out in USB-only form. Even Compaq acknowledges this with it's= > "ipaq" product line. For more examples, see your choice of:a  I OpenVMS also acknowledges the eventual transition of commodity devices toT USB.I We expect that a some point Alpha system will transition to "legacy free"FL designs, i.e. PS/2 keyboard and mouse port will be removed and replaced withG USB ports.  When this happens, the requirements for USB support becomes.J obvious.   That's why we are continuing to enhance the USB code in OpenVMSJ so we will be ready for the new designs.  Unfortuantly, exact date for the1 transistion to USB-only Alpha systems is unclear.e  K It also unclear when the transistion to USB-only systems will happen in thecK general PC industry.  The "ipaq" designs certainly signals the start of the-F transistion, but it remains a special purpose configuration.  Window98I cannot support USB-only systems.  Only Windows 2000 and Windows ME (to be L released this month) can support USB-only systems.  Intel and Microsoft haveL consistantly mis-judged that time required to transistion the PC industry toH USB.  For example, it is still possible to find PC motherboards with theE older style non-PS/2 keyboard connector.  Even the transition to PS/2r  keyboards is not fully complete!     Paul A. Jacobi Compaq Computer Corporationl! OpenVMS Systems Group, ZKO3-4/U14f 110 Spitbrook Road Nashua, NH 03062-2698t Email: Paul.Jacobi@compaq.comr   ------------------------------  % Date: Mon, 11 Sep 2000 17:33:41 -0400o* From: "Paul A. Jacobi" <nospan@nospam.com>. Subject: Re: Off-Topic: DS10 Hardware question, Message-ID: <8pjj6a$2h4j$1@lead.zk3.dec.com>  ? "Robert Deininger" <rdeininger@mindspring.com> wrote in message1 news:rdeininger-Buti  8 > I'd like the DMA to the IDE CD-ROM to get fixed first.  K I've made the latest version of DQDRIVER available on the Freeware disk and L can be downloaded from the OpenVMS website.  Feel free to hack away.  If youL can come up with a reasonable set of improvements, send me mail with the newI source code and I'll propose that the changes be added to the master packeJ for a future release.  I cannot make any guarentee that your changes would be accepted.     Paul A. Jacobi Compaq Computer Corporationd! OpenVMS Systems Group, ZKO3-4/U14i 110 Spitbrook Road Nashua, NH 03062-2698n Email: Paul.Jacobi@compaq.coml   ------------------------------    Date: 11 Sep 2000 21:13:44 -05009 From: Kilgallen@eisner.decus.org.nospam (Larry Kilgallen)p. Subject: Re: Off-Topic: DS10 Hardware question+ Message-ID: <IajwrElceBIj@eisner.decus.org>r  Y In article <8pjia7$2gqg$1@lead.zk3.dec.com>, "Paul A. Jacobi" <nospan@nospam.com> writes:y  M > It also unclear when the transistion to USB-only systems will happen in the M > general PC industry.  The "ipaq" designs certainly signals the start of therH > transistion, but it remains a special purpose configuration.  Window98K > cannot support USB-only systems.  Only Windows 2000 and Windows ME (to belN > released this month) can support USB-only systems.  Intel and Microsoft haveN > consistantly mis-judged that time required to transistion the PC industry to > USB.  D This sounds like a Microsoft problem -- all new Macintoshes use USB.   ------------------------------  % Date: Mon, 11 Sep 2000 14:01:12 -0400e3 From: "Gaitan D'Antoni" <gaitan.dantoni@compaq.com>u) Subject: Re: Safearray in COM for openvmsp6 Message-ID: <8pj6mj$c8a$1@mailint03.im.hou.compaq.com>   hresult CA:HI(VARIANT *pval) {A   long a[2];   SAFEARRAY *psa;     SAFEARRAYBOUND bounds[]={1,0};*   psa=SafeArrayCreate(VT_BSTR, 1, bounds);	   a[0]=0;e%   BSTR bstr=SysAllocString(L"abcdd");I"   SafeArrayPutElement(psa,a,bstr);    V_VT(pval) = VT_ARRAY|VT_BSTR;   V_ARRAY(pval)=psa; }h  E bellfra@my-deja.com wrote in message <8pasnn$ntg$1@nnrp1.deja.com>...oD >Is anyone know how to pass a variant type variable in vms such that >safearray can be created? >here is how I implemented!s >hresult CA:HI(VARTYPE *pval)a >{
 >  long a[2];u >  SAFEARRAY *psa;! >  SAFEARRAYBOUND bounds[]={1,0};m+ >  psa=SAFEARRAYCREATE(VT_BSTR, 1, bounds);a
 >  a[0]=0;! >  bstr=SysAllocstring(L"abcdd");e# >  safearrayputelement(psa,a,bstr);a > ; >pval->vt=VT_ARRAY|VT_BSTR; <------- got compilation errorsm: >pval->parray=psa;  <---------------got compilation errors >} >MH >ERROR MESSAGE is :  left operand of the "->" operator must be a pointer >to a classe >r >i >n >U >s' >Sent via Deja.com http://www.deja.com/e >Before you buy.   ------------------------------  % Date: Mon, 11 Sep 2000 19:16:05 +0100nB From: Andrew Harrison SUNUK Consultancy <andrew.nospam@uk.sun.com>* Subject: Re: Sun Hardware problems persist* Message-ID: <39BD2165.25785ECF@uk.sun.com>   jlsue wrote:  G > On Tue, 05 Sep 2000 17:06:48 +0100, Andrew Harrison SUNUK Consultancye# > <andrew.nospam@uk.sun.com> wrote:sC > >Basically yes and the actual latencies are also almost identicalhB > >to the WildFire, the Origin 2000's local latency was 310 ns andH > >the remote if I remember correctly 950 ns. Both better than WildFire. > >mD > >But the Origin 2000 is 4 year old machine and has been superceded > >by the 3000 which is faster.i > D > OTOH, there are many other things about the GS160/320 architecture@ > that are way above what the Origin had 4 years ago.  I've beenH > surveying some techinical information, and it directly addresses Sun's# > (and Andrew's) FUD on this issue.   : Of course it does, it has for example got higher bandwidth6 than the Origin 2000, though not higher than the 3000.  7 Incedentally someone pointed out that the 32 CPU origin 2 which is the equivalent machine to the GS320 had a4 local/remote latency ratio of 1:2 and 1:3 was for 64 CPU systems and above.  3 Does the GS latency get better or worse when the 64  CPU systems appear.'  5 However the bandwidth of the cross-bar is the same ast= that of the E10K cross-bar which is also a 4 year old system.i   >o> > Most notably are these facts:  1) Cross-bar switch, 2) cache" > coherency, and 3) OOO execution. > H > #1 allows a much greater bandwidth, despite the slower latency.  A busB > architecture, as used by Sun and others, will not scale as well.  F Really, you have figures to show this. The Sun uses a cross-bar switchD same as the GS160/320 and it has a bisectional bandwidth of ~11 GB/sE using STREAMS. This is for all local traffic because there is no suchh" thing as remote memory in an E10K.  C The GS320 cross-bar has an unspecified bandwidth, I say unspecified ; because the STREAMS benchmark results Compaq posted for theh9 GS320 were all local on node access no cross-bar traffic.g  F More over the GS320 cross-bar appears to be used for both instructionsA and data, the E10K has separate address and data buses this has a ? big impact on the actual data transfer rate over the cross-bar.e  E > #2 For cache coherency you'll need lots of bandwidth to achieve thetF > speeds needed to keep the CPUs fed.  Our coherency scheme is truly a > distributed shared memory. >vC > #3 means that that 1:3 delay factor won't be nearly as bad as SuniH > would like you to believe.  Now, ask Sun what they know about OOO, and, > how they've implemented it in the UE10000. >   D Why ? and if this is the case why have Compaq just published a TPC-CG TPM result where the system used Oracle Parallel Server in the box, thelH same scheme that Sequent used to try to mask the issues they had gettingB good performance out of the Sequent NUMA-Q box. I know the SequentG and the GS320 are different systems, but its very revealing that Compaqr@ have had to use exactly the same tuning scheme to get reasonable@ performance out of the system. Having to have the same number of> Oracle instances as quads wasn't a big winner for Sequent withE their customers. After all if you do that then you may as well go thei> whole hog and buy 8 x 4 CPU workgroup servers and cluster them: and run OPS on the cluster that should save a bit of cash.   >nD > The general architecture direction is toward NUMA systems, and forG > very good reason:  CPU speeds are increasing very significantly everymG > year, yet memory speeds are not.  Sun FUD would have you believe that F > it is based on "old" technology, however their current bus isn't all > that revolutionary either. >   C Of course it is but the general move is towards NUMA with a small n-> not NUMA with a large N like the GS320. Before you ask how Sun< knows so much about NUMA and why we may just be qualified toD pass comment on the GS320 design you should remember that Sun boughtM the IP/Patents of Thinking Machines and hired nearly all the designers of the3 CM5.   >o= > I personally would love to see a head-to-head comparison on,F > performance of some real applications.  Maybe that'd shut Sun/Andrew > up, but I doubt it.  >c  E You have them already or had you forgot the Oracle Applications 11.03eE benchmark results that Sun, IBM and Compaq have all published numbers  for.     regardst Andrew Harrison3 Enterprise IT Architectn   ------------------------------  % Date: Mon, 11 Sep 2000 19:33:07 +0100sB From: Andrew Harrison SUNUK Consultancy <andrew.nospam@uk.sun.com>* Subject: Re: Sun Hardware problems persist* Message-ID: <39BD2563.3310B349@uk.sun.com>   jlsue wrote:  A > I try not to disparage bad "chip" batches.  These can happen topA > everyone.  The thing about this that really gets me, though, isoH > Andrew's attitude.  He can't ever state "Oops, our mistake...", he hasH > to constantly say that everyone else is out to get them, causing their > problems.s  M No, the origional ecache part was too sensitive to static that was ultimatelyg5 Sun's responsibility regardless of whose part it was.e  L Sun also should have realised that a significant proportion of our customersJ were having problems running their datacenters within sensible parameters.  F Sun should also have realised that customers are not necessarely awareF of the issues such as static, their potential outcomes such as systemsJ failures and the appropriate remedies for this such as furniture, flooring etc.  ? Sun has however addressed these problems by providing customers C with a datacenter assesment service which is free and the customersnA who have taken the advice outlined after the assesment have found G that their failures have been dramatically reduced. This is illustratedn2 by the report at the end of the origional article.  E But please don't get all virtuous with me, when has Digital or CompaqiE ever appologised for their failures. The Compaq Armada notebooks that-E fell over in droves, the non standard Compaq BIOS, the VAX9000, NT onnI Alpha,  Phase V DECNET, there is an endless list of major things that theeH organisation you advocate has done to customers and not appologised for.     Regardst Andrew Harrisonm Enterprise IT Architectu   ------------------------------  % Date: Mon, 11 Sep 2000 15:03:41 -0400i- From: JF Mezei <jfmezei.spamnot@videotron.ca>h* Subject: Re: Sun Hardware problems persist, Message-ID: <39BD2C7B.AAD86571@videotron.ca>  L Just tried to picture the combattants in this holy war in the same room. NotH sure if it would turn into a pillow fight, a sumo wrestling match, or an; all-out pub brawl, breaking windows, tables, chairs etc....n  E Or perhaps they'd just throw site-preparation manuals at each other ?a   ------------------------------  % Date: Mon, 11 Sep 2000 13:25:26 -0600-% From: Dan O'Reilly <dano@process.com>5* Subject: Re: Sun Hardware problems persistC Message-ID: <4.3.2.7.2.20000911132422.01ea1c00@pop.clsp.uswest.net>   & At 01:03 PM 9/11/2000, JF Mezei wrote:M >Just tried to picture the combattants in this holy war in the same room. Not I >sure if it would turn into a pillow fight, a sumo wrestling match, or ano< >all-out pub brawl, breaking windows, tables, chairs etc.... >nF >Or perhaps they'd just throw site-preparation manuals at each other ?  H I vote for Hubble connectors on 10 meter whips at 50 paces...sort of the "electronic mace"...   ------I +-------------------------------+---------------------------------------+cI | Dan O'Reilly                  |                                       |tI | Principal Engineer            |  "Those are my principles. If you     |sI | Process Software              |   don't like them I have others."     | I | http://www.process.com        |                    -- Groucho Marx    |iI +-------------------------------+---------------------------------------+    ------------------------------  % Date: Mon, 11 Sep 2000 20:48:23 +0100?  From: steven.reece@quintiles.com* Subject: Re: Sun Hardware problems persistH Message-ID: <OF71DEF67F.7F04EF7A-ON80256957.006BF76A@qedi.quintiles.com>  4 Dan O'Reilly <dano at process dot com> wrote/quoted:  ) >>>At 01:03 PM 9/11/2000, JF Mezei wrote: I       >Just tried to picture the combattants in this holy war in the same        room. Not I       >sure if it would turn into a pillow fight, a sumo wrestling match,t       or anaB       >all-out pub brawl, breaking windows, tables, chairs etc....       > J       >Or perhaps they'd just throw site-preparation manuals at each other       ?e  H I vote for Hubble connectors on 10 meter whips at 50 paces...sort of the "electronic mace"...<<<   J Nah.  Just copies of the "Site Environmental Preparation Guide" wrapped up in house bricks.    Just for interest, this quotes :K "The recommended operating ambient temperature of 21 deg. C (70 deg. F) andjJ relative humidity of 50% RH has the following tolerance for maximum system
 reliability :x  : "Temperature : 18 deg. to 24 deg. C (65 deg. to 75 deg. F)? "Maximum rate of change 3 deg. C per hour (5.4 deg. F per hour)i  ! "Relative Humidity : 40 to 60% RH-: "Maximum rate of change : 6% RH per hour (nocondensation)"  ; (Digital Site Environmental Preparation Guide, Order NumberD- EK-CSEPG-MA-001, First Edition, October 1989)X   ------------------------------  # Date: Mon, 11 Sep 2000 20:09:56 GMTv4 From: "Michael D. Ober" <mdo.@.wakeassoc.com.nospam>* Subject: Re: Sun Hardware problems persistE Message-ID: <o_av5.3560$%p2.169673@newsread03.prod.itd.earthlink.net>w  " Don't forget the technical errata. --	 Mike Ober-  : "JF Mezei" <jfmezei.spamnot@videotron.ca> wrote in message& news:39BD2C7B.AAD86571@videotron.ca...J > Just tried to picture the combattants in this holy war in the same room. NotmJ > sure if it would turn into a pillow fight, a sumo wrestling match, or an= > all-out pub brawl, breaking windows, tables, chairs etc....I >RG > Or perhaps they'd just throw site-preparation manuals at each other ?|   ------------------------------  % Date: Mon, 11 Sep 2000 20:12:45 -0400i+ From: "Main, Kerry" <Kerry.Main@compaq.com> * Subject: RE: Sun Hardware problems persistJ Message-ID: <910612C07BCAD1119AF40000F86AF0D8052847A2@kaoexc4.kao.dec.com>   Andrew,   F I am assuming you plan to respond to my earlier reply, but I could not resist this comment you made:H  F >>> Sun also should have realised that a significant proportion of ourH customers were having problems running their datacenters within sensible parameters.   I Sun should also have realised that customers are not necessarely aware ofeL the issues such as static, their potential outcomes such as systems failuresI and the appropriate remedies for this such as furniture, flooring etc.>>>-  K ROTFL... Its only Sun systems that are having the "environmental problems",IK so how can you convince any datacenter manager worth his/her salt that they , don't know how to run a datacenter properly?  F As I said in my earlier reply, lowering the temperature is a temporary workaround - not a solution.  L Hey, Digital (RA81), Intel (FP bugs), and likely many other vendors have allI had to face some painful product issues at some point in their past. They-L key to a successful solution is to admit the problem, fix it and get on with	 business.-  L Don't try and tell experienced datacenter managers that they don't know what6 they are doing and put the blame on the environment.     Regards,  
 Kerry Main Senior Consultant,
 Compaq Canada- Professional Services. Voice : 613-592-4660 FAX   : 819-772-7036 Email : kerry.main@compaq.comm       -----Original Message-----' From: Andrew Harrison SUNUK Consultancyu! [mailto:andrew.nospam@uk.sun.com]d( Sent: Monday, September 11, 2000 2:33 PM To: Info-VAX@Mvb.Saic.Comd* Subject: Re: Sun Hardware problems persist     jlsue wrote:  A > I try not to disparage bad "chip" batches.  These can happen totA > everyone.  The thing about this that really gets me, though, ishH > Andrew's attitude.  He can't ever state "Oops, our mistake...", he hasH > to constantly say that everyone else is out to get them, causing their > problems.i  B No, the origional ecache part was too sensitive to static that was
 ultimately5 Sun's responsibility regardless of whose part it was.   L Sun also should have realised that a significant proportion of our customersJ were having problems running their datacenters within sensible parameters.  F Sun should also have realised that customers are not necessarely awareF of the issues such as static, their potential outcomes such as systemsJ failures and the appropriate remedies for this such as furniture, flooring etc.  ? Sun has however addressed these problems by providing customerstC with a datacenter assesment service which is free and the customersuA who have taken the advice outlined after the assesment have foundiG that their failures have been dramatically reduced. This is illustratedr2 by the report at the end of the origional article.  E But please don't get all virtuous with me, when has Digital or CompaqlE ever appologised for their failures. The Compaq Armada notebooks that0E fell over in droves, the non standard Compaq BIOS, the VAX9000, NT on-I Alpha,  Phase V DECNET, there is an endless list of major things that thenH organisation you advocate has done to customers and not appologised for.     Regards1 Andrew Harrison> Enterprise IT Architectd   ------------------------------    Date: 11 Sep 2000 15:45:46 -07003 From: Eric Smith <eric-no-spam-for-me@brouhaha.com>.Y Subject: Re: visual interpretation of algorithms. -> Sound interpretation of    algorythmw0 Message-ID: <qhaedek0o5.fsf@ruckus.brouhaha.com>  * spam+@orion-com.com (Joe Thompson) writes:M > The Chimes of Death on some Macs indicated the problem by variation in the tK > tones (but if you were tone-deaf or just impatient you still had the hex   > codes to guide you). -- Joea  H The point of the sound was for diagnosing Macs with broken video or RAM.3 This is mostly done at the factory or repair depot.d  F On the latest Macs, it is quite a challenge for the software to play a; sound if the RAM is not working correctly, for two reasons:i  H 1)  the sounds are stored in a compressed format in the ROM, and must be     decompressed to be played-  F 2)  the sound may need to be played to USB speakers, and USB typically#     won't work if RAM doesn't work.o  H Most Macs since the Macintosh II have been able to play the error soundsK even with no working RAM.  The original Macintosh through the Plus couldn'tyF do this since the sound buffer was in the RAM.  Later Macintoshes usedI the ASC (Apple Sound Chip) or other dedicated sound systems which did noto depend on main RAM.E   ------------------------------   End of INFO-VAX 2000.510 ************************> consistantly mis-judged that time required to transistion the PC industry to > USB.  D This sounds like a Microsoft problem -- all new Macintoshes use USB.   ------------------------------  % Date: Mon, 11 Sep 2000 14:01:12 -0400e3 From: "Gaitan D'Antoni" <gaitan.dantoni@compaq.com>u) Subject: Re: Safearray in COM for openvmsp6 Message-ID: <8pj6mj$c8a$1@mailint03.im.hou.compaq.com>   hresult CA:HI(VARIANT *pval) {A   long a[2];   SAFE:T-V>R6Xʮ'
x]fڼ#((=F6EKJOy(^Tr(x/s?:2=k=G,aՀyr<Ҭk1R-h~-pNuc%;Oߕ{m. oKzц[B}̪T߻I{V!!x`ם3mM5ڋ)P?;}oa˨.x5Էv1~6-k7}%y0>ov[z.kTyJ!V}qDlnrs
m26 <)r.'1}M=lvh4Q'^{E g߂v㥲TP#$+1nK[jah+EgCB{:[]=_1
:beГtv?]ΤFZ;|]T!䂱xƙ[6/~ߍOcN8:5\S9r-eTJfr-~4VtALfgY>!3uζR{]v-Mv6J輂+':^	:_܅ʭm̉iNwS,Fyn]jG?lK~-nu"k)^Y$G,%?3~B};v=-8ʿ݊ql9t[rp:jz-j׋?Gjex~\L=#m->=+,Rεzc6,Egaj+JOwZ
XN	?D߾ߞ>
k6n|Gwsq,Z=84cnUb
@?dڟ6^o[o*s6ex7(rXqj $y
W~^": O,*4ofK/S(1L	,r^Or)%8YzKfJ^3V'|:!|A\:Π=')3(-+X(9O3zΰf؋X#^6l쀶]	84?{1vVNq,dD:HǙ^oɫV?fD9?
?{3ק2pN󩢺zu<=_rni_FVr'2ޘӸ> #CRN|$(8><vyҏ0g/~*X?o:~ބlXn d0?{$}~Ky.d	'c9F17){DxZm}jsަ~ٹCFa.w!pT8`\%W)UF>wy!_ n'aQBO
g#;C7{}c6RiH6Myb6χ7m^*d){O6ΥNv*;#dAE"_}^ɦʱFfwr;yks]fcOu9q|\͊;qK$;"%h'Wce:ϭt'š=duɇ#Rm/i'˲^>}"WI읺Eb=q	oW{^oڌʼnu_S]1C{rbNod53,]Ɋ}=,mc>/8_6͕/IqxxA9tv>RW&uMm!y{xM7SR8H}xM5cZc	 ~0\ˡfs!}6|L)muhH|Blz֬b}[uF2YYPc~߶P;WS@
AR25hubeCy1CCUd/ z