1 INFO-VAX	Thu, 09 Sep 2004	Volume 2004 : Issue 501       Contents: Backup and ACLs  Re: Backup and ACLs  Re: Definition for SOR$GK_* ?  Re: Definition for SOR$GK_* ? G freeware's (5.0) mailcount from a system account? (orphaned .mai files) P Re: freeware's (5.0) mailcount from a system account? (orphaned .mai files) file< Re: Itanic Adoption Survey from 23-August-2004 Computerworld< Re: Itanic Adoption Survey from 23-August-2004 Computerworld% Re: Itanium not meeting Intel's goals 5 Re: JAVA deleted without warning by VMS 7.3-2 upgrade 4 Problem using SCP/SFTP between HP-TCPIP and Multinet Re: Sending mail via dbl-cobol Re: Sending mail via dbl-cobolP RE: Status of PL/I on Itanium (Was: Re: Canada's Equitable Life begins         mP Status of PL/I on Itanium (Was: Re: Canada's Equitable Life begins moving off VM Re: VMS marketing opportunity  RE: VMS marketing opportunity  Re: VMS marketing opportunity  Re: VMS marketing opportunity " Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" RE: why mainframes are still used?" RE: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?" Re: why mainframes are still used?  F ----------------------------------------------------------------------   Date: 9 Sep 2004 00:25:28 -0700 # From: sinobato@yahoo.com (Sinobato)  Subject: Backup and ACLs= Message-ID: <ec657b64.0409082325.38f735be@posting.google.com>    Hello!  D I need to backup directory and all files/sub-directories beneath it.A Also, I need to preserve the current ACL that the directories and ( files has. The backup command I've used:  = $ backup/log disk1:[test...] disk2:[backup]test_back.bck/save   E However, when I try to restore it on another node but almost the same C setup (ie. same user, same identifiers), I am not able to apply the F correct original ACLs. Here is the command I'm using to restore, after! copying the saveset to this node:   1 $ backup/log disk$backup:[test]test_back.bck/save  disk$prod:[*...]/by_own=orig  F Can somebody help me on this? I'm just trying to avoid re-applying the; ACLs again on the restored files, which might take a while.    Thanks,  Sinobato   ------------------------------  % Date: Thu, 09 Sep 2004 12:53:41 +0200 2 From: Karl Rohwedder <emil.mustermann@t-online.de> Subject: Re: Backup and ACLs* Message-ID: <chpcn9$20f13@doiweb4.b2x.vwg>   Sinobato wrote:  > Hello! > F > I need to backup directory and all files/sub-directories beneath it.C > Also, I need to preserve the current ACL that the directories and * > files has. The backup command I've used: > ? > $ backup/log disk1:[test...] disk2:[backup]test_back.bck/save  > G > However, when I try to restore it on another node but almost the same E > setup (ie. same user, same identifiers), I am not able to apply the H > correct original ACLs. Here is the command I'm using to restore, after# > copying the saveset to this node:  > 3 > $ backup/log disk$backup:[test]test_back.bck/save  > disk$prod:[*...]/by_own=orig > H > Can somebody help me on this? I'm just trying to avoid re-applying the= > ACLs again on the restored files, which might take a while.  > 	 > Thanks, 
 > Sinobato  5 Make sure, that the identifiers have the same values. S Make sure, that you have the required privs (I think SYSPRV) to set to another UIC.    ------------------------------   Date: 9 Sep 2004 02:55:03 -0700 0 From: chris_doran@postmaster.co.uk (Chris Doran)& Subject: Re: Definition for SOR$GK_* ?= Message-ID: <948f0720.0409090155.35e47822@posting.google.com>   u chris_doran@postmaster.co.uk (Chris Doran) wrote in message news:<948f0720.0409031706.1498f3ee@posting.google.com>... d > Dale Dellutri <ddelQQQlutr@panQQQix.com> wrote in message news:<chabd1$lc8$1@reader1.panix.com>...E > > I'm trying to write a FORTRAN program which uses the SOR$ROUTINES D > > (Sort/Merge Utility routines).  According to the OpenVMS Utility> > > Routines Manual, Sort/Merge Routines section, when callingF > > SOR$BEGIN_SORT, one needs to specify the sort process by passing a > > byte containing one of:  > > ! > >   SOR$GK_RECORD (record sort)  > >   SOR$GK_TAG (tag sort) # > >   SOR$GK_ADDRESS (address sort)  > >   SOR$GK_INDEX (index sort)  > > C > > I specifically need SOR$GK_RECORD.  However, I can't find this  C > > definition anywhere.  It is not in $SORDEF or SOR$ROUTINES (in  J > > SYS$LIBRARY:FORSYSDEF.TLB), nor any of the other TLB's in SYS$LIBRARY. > > < > > Where are these values defined, or what is their values? > >  > > (OpenVMS 7.3-1 on Alpha.)  > > I > > A web search seems to indicate that SOR$GK_RECORD = 1, but I'd reall  8 > > like to get the standard definition from the system. > F > If you're programming in C, just declare them as globalvalues (DEC CC > extension) and the linker will find them "from the system", e.g.:  >  > #include <stdio.h> > F > globalvalue SOR$GK_RECORD, SOR$GK_TAG, SOR$GK_ADDRESS, SOR$GK_INDEX; >  > main() > { 0 > printf("SOR$GK_RECORD = %d\n", SOR$GK_RECORD);* > printf("SOR$GK_TAG = %d\n", SOR$GK_TAG);2 > printf("SOR$GK_ADDRESS = %d\n", SOR$GK_ADDRESS);. > printf("SOR$GK_INDEX = %d\n", SOR$GK_INDEX); > }  >  > which gives (VAX/VMS 6.2): >  > SOR$GK_RECORD = 1  > SOR$GK_TAG = 2 > SOR$GK_ADDRESS = 4 > SOR$GK_INDEX = 3 > G > When/if I have time, I'll dig out the (ancient) microfiche to confirm  > this.   ? Having dug out the (5.4!) 'fiche, I verify that these and other D symbols are declared as global symbols and available from SORTSHR to5 the linker as I and others have previously indicated.   F There is therefore no need for them to be in include files, and indeedB it would be unwise to duplicate their definitions that way. Bitter= experience shows that no matter how careful you are, defining E something "identically" in two or more places sooner or later results - in someone forgetting to update them in sync.   D Why some things are in include files and some in shareable images isA no doubt something on which we could have a long debate :) Please  create a new topic if you must.    Chris    ------------------------------  % Date: Thu, 09 Sep 2004 13:18:52 +0100 - From: John Laird <nospam@laird-towers.org.uk> & Subject: Re: Definition for SOR$GK_* ?8 Message-ID: <edi0k011fiseo7bv3h9got9fv4fek611jl@4ax.com>  H On 9 Sep 2004 02:55:03 -0700, chris_doran@postmaster.co.uk (Chris Doran) wrote:  @ >Having dug out the (5.4!) 'fiche, I verify that these and otherE >symbols are declared as global symbols and available from SORTSHR to 6 >the linker as I and others have previously indicated. > G >There is therefore no need for them to be in include files, and indeed C >it would be unwise to duplicate their definitions that way. Bitter > >experience shows that no matter how careful you are, definingF >something "identically" in two or more places sooner or later results. >in someone forgetting to update them in sync.  H Not if the build process creates the include files and the global symbolE definitions from the same sources(*).  This is VMS, not some twilight  hacker's paradise, remember !    (*)  SDL, most likely.   --  4 Press any key to continue or any other key to quit.    Mail john rather than nospam...    ------------------------------   Date: 9 Sep 2004 07:49:04 -0700 & From: pjohnson@wcu.edu (Patti Johnson)P Subject: freeware's (5.0) mailcount from a system account? (orphaned .mai files)= Message-ID: <ca28f268.0409090649.61aaae53@posting.google.com>   
 Hello all,D just a quick question:  I've downloaded & used mailcount to identifyA orphaned mail$...mai files.  It's great.  But how can I use it to F identify these files for others' accounts without logging in to them? B As the system administrator, I'd like to clean up others' accounts7 w/out having to involve them to login to their account.   0 Am I missing something that's painfully obvious?   tia  pj   ------------------------------  % Date: Thu, 09 Sep 2004 16:19:31 +0100 - From: Roy Omond <Roy.Omond@BlueBubble.UK.Com> Y Subject: Re: freeware's (5.0) mailcount from a system account? (orphaned .mai files) file * Message-ID: <2qbaouFtepoeU1@uni-berlin.de>   Patti Johnson wrote:   > Hello all,F > just a quick question:  I've downloaded & used mailcount to identifyC > orphaned mail$...mai files.  It's great.  But how can I use it to H > identify these files for others' accounts without logging in to them? D > As the system administrator, I'd like to clean up others' accounts9 > w/out having to involve them to login to their account.    1)	$ submit/user=xxxx        or  4 2)	Download and install Hunter Goatley's HGLOGIN at:  = 	http://vms.process.com/scripts/fileserv/fileserv.com?HGLOGIN     	 Roy Omond  Blue Bubble Ltd.   ------------------------------   Date: 9 Sep 2004 07:10:00 -0700 % From: tadamsmar@yahoo.com (Tom Adams) E Subject: Re: Itanic Adoption Survey from 23-August-2004 Computerworld < Message-ID: <ea44f5a1.0409090610.e7918e5@posting.google.com>  v koehler@eisner.nospam.encompasserve.org (Bob Koehler) wrote in message news:<RGnxz2uw4ye1@eisner.encompasserve.org>...g > In article <ea44f5a1.0409080525.61494fbf@posting.google.com>, tadamsmar@yahoo.com (Tom Adams) writes:  > > I > > I did see somewhere that HP is expecting most users to migrate before F > > 2007.  But the survey seems to indicate that most have not started0 > > planning, so the migrations could run later. > > >    I know of shops that plan to keep right on running VAXen.  E I guess we might end up running Alphas for a long.  I am not sure its / a good option for us.  I think we will migrate.   E I was pricing Alphas (400s, DS10S) in the aftermarket from early 2003 C till recently.  These low end Alphas seemed to be more available in ; early 2003.  The market seemed a lot tighter by early 2004.    ------------------------------  $ Date: Thu, 9 Sep 2004 11:19:42 -0400# From: "John Smith" <a@nonymous.com> E Subject: Re: Itanic Adoption Survey from 23-August-2004 Computerworld , Message-ID: <xfGdnWQzzKcR6d3cRVn-tA@igs.net>   Tom Adams wrote:@ > koehler@eisner.nospam.encompasserve.org (Bob Koehler) wrote in9 > message news:<RGnxz2uw4ye1@eisner.encompasserve.org>... @ >> In article <ea44f5a1.0409080525.61494fbf@posting.google.com>,* >> tadamsmar@yahoo.com (Tom Adams) writes: >>> B >>> I did see somewhere that HP is expecting most users to migrateE >>> before 2007.  But the survey seems to indicate that most have not 8 >>> started planning, so the migrations could run later. >>? >>    I know of shops that plan to keep right on running VAXen.  > G > I guess we might end up running Alphas for a long.  I am not sure its 1 > a good option for us.  I think we will migrate.  > G > I was pricing Alphas (400s, DS10S) in the aftermarket from early 2003 E > till recently.  These low end Alphas seemed to be more available in = > early 2003.  The market seemed a lot tighter by early 2004.     K Don't kid yourself....HP will be ratcheting the Alphaserver production line F to a close, most likely creating shortages in certain product lines asG similarly equivalent (in HP's eyes) Itanic-based systems are brought to  market.   K It won't occur to HP that the customer base will continue to need those new L Alpha's for at least 2 years after their apps may become available on ItanicK in order to satify themselves that the 'ported' Itanic apps behave the same E as the Alpha apps and that the IT guys have sufficient time to become J comfortable with the new stuff before any migration off Alpha begins. ThatF and the need to accomodate user/processing growth during the migration period.     J Just because HP makes VMS-qualified Itanics available for GA on day X doesF not mean that a users apps will also be available on day X or even dayI X+365. ISV's are going to attempt to gauge the uptake of Itanic among VMS E shops, and I'd wager that many ISV Alpha/VMS apps are never ported to * Itanic.  HP will, of course, blame others.   ------------------------------  $ Date: Thu, 9 Sep 2004 09:22:37 -0400# From: "John Smith" <a@nonymous.com> . Subject: Re: Itanium not meeting Intel's goals, Message-ID: <1p-dnYk-tuKGxN3cRVn-jw@igs.net>   David J Dachtera wrote:  > JF Mezei wrote:  >> >>> L http://news.com.com/Itanium+not+meeting+Intel%27s+goals/2100-1010_3-5349588. html?tag=nefd.top  >>D >> Will not quote the whole article here. It is a comment made by an? >> Intel exec during the dveelopper forum, but refers mostly to D >> short/medium term goals, but IA64 is still on track for long termE >> goals even though those goals were downscaled from total dominance  >> of the market.  >>E >> The 64 bit 8086 has also thrown the IA64 folks off track with XEON ? >> now offering serious competition to IA64, relegating IA64 to 7 >> combatting only Sparc and Power in high end systems.  > G > Y'know, I was pondering this today and had a chill run down my spine.  > Consider:  > G > There's another a thread that mentions an article indicating that the / > last Alpha chips are coming off the FABs now.  > C > Compaq was pushing NT/Alpha like crazy just before the Alphacide.  > H > hp is "pushing" VMS (in so far as it ever has) right now. With all dueG > respect to Sue, perhaps even she is being led down the primrose path.  > G > Intel appears to be suggesting that Itanic may be sinking faster than * > we first thought, or ever dared suggest. > F > OVMS-IA32 and OVMS-x86/64 are both viewed as non-doable (for reasonsG > known only to VMS, the rest of the market seems to disagree - tacitly @ > - by fully embracing those technologies). That leaves VMS with9 > potentially no path forward and certainly no path back.        see L http://story.news.yahoo.com/news?tmpl=story&cid=1738&ncid=1208&e=8&u=/zd/200 40908/tc_zd/134847    " Chip Race Is On Between Intel, IBM   Wed Sep 8, 8:11 AM ET  Mark Hachman - eWEEK  F SAN FRANCISCO-The enterprise processor contest is now a two-horse raceD between IBM and Intel Corp., Intel's enterprise platforms chief said Tuesday.  D In a briefing with reporters at the Intel Developer Forum here, AbhiI Talwalkar, vice president of Intel's Enterprise Platforms Group, in Santa L Clara, Calif., said in the future only IBM's Power5 architecture and Intel'sE family of processors will be battling for OEM sockets. Talwalkar also I admitted in a question-and-answer session earlier in the day that Intel's A Itanium sales have not met "aggressive" internal sales forecasts.   K Sun Microsystems Inc. isn't even a contender in the race, he said. "Most of D you watching Sun will appreciate where Sun is going from a long-termH standpoint," Talwalkar said, apparently referring to Sun's collaborationC with Fujitsu Ltd. on the future development of the SPARC processor.   L While Intel has fared well in the low-end server segment, RISC architecturesK continue to dominate servers selling for more than $25,000, said Talwalkar, C who added that legacy applications have tied IT administrators into * supporting the so-called "legacy systems."  H "Power5 is a very competitive platform to Itanium 2," he said, primarilyF because of its multicore architecture. Intel executives demonstrated aJ dual-core "Montecito" Itanium processor here on Tuesday, a processor Intel will launch next year.  L Talwalkar also said Tuesday that the company has not been happy with ItaniumE sales, although he quickly qualified the statement. In a Q&A session, K Talwalkar was asked if the Itanium should be considered a failure. Although I he said it is not a failure, he added: "Are we meeting system-level goals = for this year? No, not at the aggressive levels that we set."   J The company, however, is satisfied with the "types" of customers that have4 chosen to buy the Itanium processor, Talwalkar said.      H ....kinda reminds me of the denials by the Iraqi Minister of Information3 about the presence of US troops outside Baghdad....    ------------------------------   Date: 9 Sep 2004 09:38:08 -0700 % From: whohe@whoever.com (DL Phillips) > Subject: Re: JAVA deleted without warning by VMS 7.3-2 upgrade< Message-ID: <af0dc2ea.0409090838.173b0cf@posting.google.com>  b Verne Britton <verne-REMOVE-THIS@wvnet.edu> wrote in message news:<413F1AAD.C4A6AFCC@wvnet.edu>...5 > (problem and solution... in case this bites others)  > C > We run Oracle on VMS, and over the weekend upgraded from 7.3-1 to  > 7.3-2. > C > Later my Oracle people said they were getting Java errors, namely  > - >     No version of JDK installation detected  > C > Years ago I was surprised to find Java 1.1.8 on my system as part C > of some VMS upgrade;  now it has disappeared just as quietly  :-)  > 0 > ITRC says (gosh I do miss DSNlink), in article > 2 >    Oracle error after upgrading to OpenVMS 7.3-2 >    Date: 2/10/04 >    Document id: VO040206_CW01  > 
 >    [...] > F >    This is because Java was not included as part of the base OpenVMS > 7.3-2 I >    installation. As a result, customers who relied on Java being on the  > systemI >    (typically due to an OpenVMS Alpha installation or upgrade) may have  > to  H >    manually re-install Java. The non-inclusion of JAVA was intended to > be  I >    stated in the 7.3-2 Release Notes, but was inadvertantly overlooked.  >   D *Little* oversights like this seem to be happening too often lately.. Have you checked the V7.3-2 patch site lately?   >  > C > HEY ENGINEERING... next time can you have a message on the screen G > during the upgrade itself to mention these kinds of things to us  :-)     D I'll go you one better. A modern upgrade procedure should detect theC component and layered software that is installed and provide notice E about possible conflicts and allow you to upgrade any or all those as 	 you wish.      > D > I would have thought if a new version was not being installed that7 > at least the old version WOULD HAVE BEEN LEFT BEHIND.   ' That would be a reasonable expectation.      Doug   *** 8 ** Never anthropomorphize computers. They don't like it. ***    ------------------------------   Date: 9 Sep 2004 07:48:05 -0700 4 From: francesco.gennai@iat.cnr.it (Francesco Gennai)= Subject: Problem using SCP/SFTP between HP-TCPIP and Multinet < Message-ID: <72f5654.0409090648.7a2865b8@posting.google.com>  < I'm trying to use SCP (or SFTP) with a client (HP-TCPIP) on:  :  HP TCP/IP Services for OpenVMS Alpha Version V5.4 - ECO 2=  on a COMPAQ AlphaServer DS20E 833 MHz running OpenVMS V7.3-2     and a server (Process Multinet):  J  Process Software MultiNet V4.4 Rev A-X, COMPAQ AlphaServer DS20E 833 MHz,  OpenVMS AXP V7.3-2    but I get the following error:  E  Warning: packet length mismatch: expected 42, got 8;  connection to    non-standard server?   N  FATAL: BUILD4$:[TCPIP_V54_BL15.SRC_PAT.SSH]SSHFC_RECURSE.C;2:605 SshFCRecurseK  (function name unavailable) Unreachable code failed: Invalid code reached.     7  %TCPIP-F-SSH_FATAL, non-specific fatal error condition     
 Any help ?  	 Francesco    ------------------------------  $ Date: Thu, 9 Sep 2004 09:17:20 -0400* From: "Syltrem" <syltremzulu@videotron.ca>' Subject: Re: Sending mail via dbl-cobol 3 Message-ID: <yFY%c.2192$KF.17466@tor-nn1.netcom.ca>   L Look up the OpenVMS utility routines manual, specifially the MAIL$ routines. MAIL$SEND_BEGIN, MAIL$SEND_ADD_ATTRIBUTE, MAIL$SEND_ADD_BODYPART,  MAIL$SEND_ADD_ADDRESS, MAIL$SEND_MESSAGE, MAIL$SEND_END,  L Everything you need is there. If you send through SMTP, just enable the SMTP& service in UCX/TCPIP/MULTINET/WHATEVER  = You can call those from any language you have a compiler for.    HTH    --   Syltrem    OpenVMS 7.3-1 + Oracle 8.1.7.4H http://pages.infinit.net/syltrem (OpenVMS related web site, en franais)% ---zulu is not in my email address--- H "Ste /aka=Stardust" <stefores@toglimi.tin.it> a crit dans le message de2 news:t2a0k010s5hj2553k4bfv1ovohtrmfmhs1@4ax.com...G > I want to send mails automatically from my synergex's dbl (cobol-like  > language) apps. B > I think the best way is to open a tcp/ip communication to a mailF > server witch is not the same openvms system but it's inside the sameG > network, send smtp commands and data, then close communication (doing + > this using telnet from dcl prompt works). ! > Is there a best way to do this? C > I have searched a lot and found some examples but all of them are E > written in C and use a C tcp/ip socket library. It's possible to do  > something in dbl-cobol? A > Thanks in advance for the answers, and sorry for my english :-)  >  >  > Ste /aka=Stardust  > --@ > Non c' cattivo pi cattivo di un buono quando diventa CATTIVO   ------------------------------  % Date: Thu, 09 Sep 2004 15:46:17 +0200 1 From: Ste /aka=Stardust <stefores@toglimi.tin.it> ' Subject: Re: Sending mail via dbl-cobol 8 Message-ID: <u1n0k055u5m4r0svsg362l6h0f7q9v7ktq@4ax.com>  5 Thanks for the answer, but I have other questions :-) . The system is OpenVMS 7.2-1 running on a ES40.D When you say "just enable the SMTP service", do you mean that i needB the TCPIP$SMTP_*.EXE to be loaded? (I saw there is a batch file to make those run).E I also need a mailbox to be configured locally on the openvms system? B In Compaq doc's I read "The callable interface of the Mail utilityE (MAIL) lets you send messages to users on your system or on any other E computer connected to your system with DECnet". This mean I must send E message to a local mail server on the same network (configuring node) 6 because I can't send message directly to the internet?
 TIA again,        , Il Thu, 9 Sep 2004 09:17:20 -0400, "Syltrem"& <syltremzulu@videotron.ca> ha scritto:  M >Look up the OpenVMS utility routines manual, specifially the MAIL$ routines.  >MAIL$SEND_BEGIN,  >MAIL$SEND_ADD_ATTRIBUTE,  >MAIL$SEND_ADD_BODYPART, >MAIL$SEND_ADD_ADDRESS,  >MAIL$SEND_MESSAGE,  >MAIL$SEND_END,  > M >Everything you need is there. If you send through SMTP, just enable the SMTP ' >service in UCX/TCPIP/MULTINET/WHATEVER  > > >You can call those from any language you have a compiler for. >  >HTH   Ste /aka=Stardust  --> Non c' cattivo pi cattivo di un buono quando diventa CATTIVO   ------------------------------  $ Date: Thu, 9 Sep 2004 04:54:15 -0700# From: "Tom Linden" <tom@kednos.com> Y Subject: RE: Status of PL/I on Itanium (Was: Re: Canada's Equitable Life begins         m 9 Message-ID: <NDEMLKKEBOIFBMJLCECICEEKDMAA.tom@kednos.com>   E We are planning to start a port in Q1 2005 and expect to take 6 to 9  C months to get to a field test version.  These plans could change if 7 HP drops Itanium.  Bart, what is your interest in PL/I?    < -----Original Message-----. < From: Bart Zorn [mailto:Bart.Zorn@xs4all.nl], < Sent: Thursday, September 09, 2004 4:27 AM < To: Info-VAX@Mvb.Saic.Com F < Subject: Status of PL/I on Itanium (Was: Re: Canada's Equitable Life < begins moving off VMS) <  < 1 < "Tom Linden" <tom@kednos.com> wrote in message  7 < news:<NDEMLKKEBOIFBMJLCECICEEHDMAA.tom@kednos.com>...   < > < -----Original Message----- <  <    [ S n i p ... ] < 1 < > < So what's the status of PL/1 on Itanic/VMS?  < >  < > Nothing new report < 4 < I would not mind if you repeat the current status! <  < Bart Zorn  <  < --- ( < Incoming mail is certified Virus Free.< < Checked by AVG anti-virus system (http://www.grisoft.com).B < Version: 6.0.742 / Virus Database: 495 - Release Date: 8/19/2004 <  --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.742 / Virus Database: 495 - Release Date: 8/19/2004   ------------------------------   Date: 9 Sep 2004 04:26:55 -0700 % From: Bart.Zorn@xs4all.nl (Bart Zorn) Y Subject: Status of PL/I on Itanium (Was: Re: Canada's Equitable Life begins moving off VM = Message-ID: <a98cd882.0409090326.676f8266@posting.google.com>   d "Tom Linden" <tom@kednos.com> wrote in message news:<NDEMLKKEBOIFBMJLCECICEEHDMAA.tom@kednos.com>... > < -----Original Message-----      [ S n i p ... ]  / > < So what's the status of PL/1 on Itanic/VMS?  >  > Nothing new report  2 I would not mind if you repeat the current status!  	 Bart Zorn    ------------------------------   Date: 9 Sep 2004 12:02:46 GMT ( From: bill@cs.uofs.edu (Bill Gunshannon)& Subject: Re: VMS marketing opportunity* Message-ID: <2qav36Ftt2i0U1@uni-berlin.de>  , In article <SeydnQO_JMySKKLcRVn-jw@igs.net>,& 	"John Smith" <a@nonymous.com> writes: > L > Not every company operating in Florida is headquartered in Florida....someM > companies might get the message that VMS for disaster tolerance is a pretty ( > good idea if it were conveyed to them.   D Considering that the third major hurricane in about as many weeks isF headed in their direction one would think the smart ones would get theB idea that being located in Florida at all is a really bad business	 decision.    bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  $ Date: Thu, 9 Sep 2004 04:59:30 -0700# From: "Tom Linden" <tom@kednos.com> & Subject: RE: VMS marketing opportunity9 Message-ID: <NDEMLKKEBOIFBMJLCECIIEEKDMAA.tom@kednos.com>    < -----Original Message-----F < From: bill@gw5.cs.uofs.edu [mailto:bill@gw5.cs.uofs.edu]On Behalf Of < Bill Gunshannon , < Sent: Thursday, September 09, 2004 5:03 AM < To: Info-VAX@Mvb.Saic.Com ( < Subject: Re: VMS marketing opportunity <  < . < In article <SeydnQO_JMySKKLcRVn-jw@igs.net>,( < 	"John Smith" <a@nonymous.com> writes: < > > < > Not every company operating in Florida is headquartered in < Florida....some C < > companies might get the message that VMS for disaster tolerance 
 < is a pretty * < > good idea if it were conveyed to them. < F < Considering that the third major hurricane in about as many weeks isH < headed in their direction one would think the smart ones would get theD < idea that being located in Florida at all is a really bad business < decision.   . Unless your business is selling trailer homes. <  < bill <  < --L < Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesF < bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. < University of Scranton   |@ < Scranton, Pennsylvania   |         #include <std.disclaimer.h> <  < --- ( < Incoming mail is certified Virus Free.< < Checked by AVG anti-virus system (http://www.grisoft.com).B < Version: 6.0.742 / Virus Database: 495 - Release Date: 8/19/2004 <  --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.742 / Virus Database: 495 - Release Date: 8/19/2004   ------------------------------   Date: 9 Sep 2004 12:16:23 GMT ( From: bill@cs.uofs.edu (Bill Gunshannon)& Subject: Re: VMS marketing opportunity* Message-ID: <2qavsnFtt2i0U2@uni-berlin.de>  9 In article <NDEMLKKEBOIFBMJLCECIIEEKDMAA.tom@kednos.com>, & 	"Tom Linden" <tom@kednos.com> writes: >  >  > < -----Original Message-----H > < From: bill@gw5.cs.uofs.edu [mailto:bill@gw5.cs.uofs.edu]On Behalf Of > < Bill Gunshannon . > < Sent: Thursday, September 09, 2004 5:03 AM > < To: Info-VAX@Mvb.Saic.Com * > < Subject: Re: VMS marketing opportunity > <  > < 0 > < In article <SeydnQO_JMySKKLcRVn-jw@igs.net>,* > < 	"John Smith" <a@nonymous.com> writes: > < > @ > < > Not every company operating in Florida is headquartered in > < Florida....some E > < > companies might get the message that VMS for disaster tolerance  > < is a pretty , > < > good idea if it were conveyed to them. > < H > < Considering that the third major hurricane in about as many weeks isJ > < headed in their direction one would think the smart ones would get theF > < idea that being located in Florida at all is a really bad business
 > < decision.  > 0 > Unless your business is selling trailer homes. >   B You still don't want your business located there.  Nothing to sellB when the 160mph winds blow all your stock into the next county. Of? course, I guess that would give a whole new meaning to the term  "rolling stock". :-)   bill   --  J Bill Gunshannon          |  de-moc-ra-cy (di mok' ra see) n.  Three wolvesD bill@cs.scranton.edu     |  and a sheep voting on what's for dinner. University of Scranton   |A Scranton, Pennsylvania   |         #include <std.disclaimer.h>       ------------------------------  $ Date: Thu, 9 Sep 2004 09:30:13 -0400# From: "John Smith" <a@nonymous.com> & Subject: Re: VMS marketing opportunity, Message-ID: <wZidndjp-r15x93cRVn-oA@igs.net>   Bill Gunshannon wrote:; > In article <NDEMLKKEBOIFBMJLCECIIEEKDMAA.tom@kednos.com>, ' > "Tom Linden" <tom@kednos.com> writes:  >> >> >> < -----Original Message----- F >> < From: bill@gw5.cs.uofs.edu [mailto:bill@gw5.cs.uofs.edu]On Behalf >> Of < Bill Gunshannon / >> < Sent: Thursday, September 09, 2004 5:03 AM  >> < To: Info-VAX@Mvb.Saic.Com+ >> < Subject: Re: VMS marketing opportunity  >> < >> <1 >> < In article <SeydnQO_JMySKKLcRVn-jw@igs.net>, * >> < "John Smith" <a@nonymous.com> writes: >> < >A >> < > Not every company operating in Florida is headquartered in  >> < Florida....someF >> < > companies might get the message that VMS for disaster tolerance >> < is a pretty- >> < > good idea if it were conveyed to them.  >> <F >> < Considering that the third major hurricane in about as many weeksF >> is < headed in their direction one would think the smart ones wouldF >> get the < idea that being located in Florida at all is a really bad >> business < decision.  >>1 >> Unless your business is selling trailer homes.  >> > D > You still don't want your business located there.  Nothing to sellD > when the 160mph winds blow all your stock into the next county. OfA > course, I guess that would give a whole new meaning to the term  > "rolling stock". :-)    I I want to have about 40 semi's loaded with 3/4" plywood and bottled water G drive into FLA each week during hurricane season and park in the target J zones for the week leading up to landfall. Cash sales only. Load your own.   ------------------------------  % Date: Thu, 09 Sep 2004 02:53:41 -0400 ( From: David Froble <davef@tsoft-inc.com>+ Subject: Re: why mainframes are still used? , Message-ID: <413FFDF5.6060100@tsoft-inc.com>   Main, Kerry wrote:   >>-----Original Message-----7 >>From: JF Mezei [mailto:jfmezei.spamnot@teksavvy.com]  ! >>Sent: September 8, 2004 8:41 PM  >>To: Info-VAX@Mvb.Saic.Com - >>Subject: Re: why mainframes are still used?  >>  > >>Once you've built an ecosystem around your mainframe, it is  >>quite difficult to, >>remove/change the heart of that ecosystem. >> >> >  > JF -   > $ > You have hit the nail on the head. > I > Most Customers, vendors and propeller head analysts that think they can I > easily move a heavily customized application that has been finely tuned > > for 10-15-20+ years have no idea of what they are taking on.  : The same was true for many VMS shops.  A constant diet of H convert-convert-convert caused many of them to take that difficult step.   --  4 David Froble                       Tel: 724-529-04504 Dave Froble Enterprises, Inc.      Fax: 724-529-0596> DFE Ultralights, Inc.              E-Mail: davef@tsoft-inc.com 170 Grimplin Road  Vanderbilt, PA  15486    ------------------------------  % Date: Thu, 09 Sep 2004 09:08:21 +0100 < From: Andrew Harrison <andrew_remove__s._harrison@s_u_n.com>+ Subject: Re: why mainframes are still used? 0 Message-ID: <chp31m$f1u$1@new-usenet.uk.sun.com>   JF Mezei wrote: # > Mainframes have 2 selling points:  > L > On a technical basis, I think that where IBM mainframes historically had aS > large edge was in IO throughput. (I think this could be debated nowadays though).  >   @ The Sun E10K took the I/O crown away from IBM and it hasn't been	 returned.   D SGI fans might claim that SGI snatched it first but SGI Origins haveG never really ventured into the large commercial batch processing space.    Regards  Andrew Harrison N > From a management point of view, it isn't so much the mainframe per say, butO > the software that runs on it. If you are a bank and your banking software and P > databases are already on the mainframe since they have always been there, thenK > you are more likely to continue to run that solid software and either add O > software to your mainframe, or connect some lesser machines to your mainframe_) > to make use of the mainframe databases.a > O > Consider banking web site or telephone applications. The web or telephone app.N > may run on UNXI, but they connect to the IBM mainframe to perform the actualK > transactions. Same for ATM and interbank networks which may run on tandem O > machines for total uptime, but the tandem ends up sending the transactions tol > the IBM mainframe. > P > Once you've built an ecosystem around your mainframe, it is quite difficult to, > remove/change the heart of that ecosystem.   ------------------------------  * Date: Thu, 9 Sep 2004 12:01:08 +0000 (UTC)( From: m.kraemer@gsi.de (Michael Kraemer)+ Subject: Re: why mainframes are still used? 5 Message-ID: <chpgm4$olj$1@lnx107.hrz.tu-darmstadt.de>p  i In article <e3muj0h1bntovb5cm2ge9pi3dv1df28prg@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes:? > H > I don't know what they had in a 3090.  The X/MP 48 had 4 CPUs, and theC > 890 MFLOPS number was using all of them (it was done by one of myEA > co-workers for a pre-sales demo using customer-supplied Fortran?G > code...with some mods by us).  The "48" in an X/MP's model name tells E > you the CPU count and memory size: 48 = 4 CPUs and 8 million 64-bit 8 > words of memory.  The one I worked on at NRL was a 12.  ' Now this sounds more reasonable too me.u% A single CPU doing around 250 MFLOPS.?9 I further assume that this was on highly vectorized code.e6 Given the vector speedup of 10 you quoted below leaves/ a single CPU with 25 MFLOPS scalar performance.43 According to you a 3090 does also 25 MFLOPs scalar. $ So where exactly is the difference ?D (to be honest: I doubt the IBM number, it was more 15 than 25, IIRC)   > >aJ > >Well, to be fair you should add that those Crays used to be specialized@ > >vector computers whereas the 3090 was a general purpose box - > F > Cray's were general purpose computers as well.  They had both scalarD > and vector capabilities, with the vector operations being about 10H > times faster than scalar (by nearly eliminating fetch/store delays andF > reducing instruction decoding times for repeated operations mostly). > S > >and probably quite a bit cheaper than a Cray (as far as multimillion $ computers_ > >could be called "cheap"). > G > Not sure what a 3090 cost, but they weren't cheap either.  Cost isn'tpG > really the point though.  The point was that IBM claimed to be in the F > same class, and they weren't even close.  A kid's go cart is cheaperG > than an Indy car, but that isn't really releveant when they start thed@ > race, and doesn't make the go kart a contender in the class in > question.   @ Now again this is a biased comparison. IBM cycle times were also, in the nanosecond (two-digit, though) range.@ No doubt Cray's were "real" supercomputers and faster than IBM'sB but not by your 890/25 ratio. IIRC you could buy a 6-CPU 3090 withG vector features, capable of doing several 100 MFLOPs on good vectorizedo8 code. *This* is the number to be compared with your 890.G And I highly doubt that IBM praised their barebone scalar 3090 as beingn5 supercomputer, it was certainly meant including a VF.r   / > >For a "little" extra money you could hook uppJ > >a vector "coprocessor" (per CPU) which could deliver around 100 MFLOPS,Q > >IIRC. This makes it less laughable than you suggest, since e.g. any universitysG > >running a 3090 anyway could add vector facilities without having to   > >buy a full Cray.r > E > What would it cost to get a 3090 that could do 890 MFLOPS?  If suche > was possible at all?  K Cost ? Dunno exactly, but I seem to recall annual leasing rates of DEM 5Mioe, ($ 2.5Mio), over a period of around 4 years.  tF > I know who the competition was for Cray at the time.  IBM wasn't it.F > Fujitsu and Hitachi, with a few early worries from CDC's spinoff ETAC > (which ended up buying Crays to fulfill contracts when the ETA-100C > never got working (melted itself I heard ;-), but never IBM.  IBMQG > wasn't in the supercomputer arena...though some of their ads tried to' > claim otherwise. > E > >The same logic, however, holds also for vectorized vs scalar code,nD > >so your Cray was only worth its money for highly ( maybe >> 80% ) > >vectorizable code.n > @ > Perhaps, but vector processing is really common in science andG > engineering (lots of array operations), which is where supercomputersrG > are needed.  With business processing it's usually more a case of I/OtF > speed than raw CPU power that matters.  You are usually doing simpleB > things to lots of data, not complex things to a limited dataset.  " I think you can't generalize this.G Certainly there are such codes working on large arrays with high degreeI? of vectorization, but there are as well such with lower degree,rO for which an IBM solution was more cost effective (and sure cost is a point !).nH And there are scientific problems, e.g. in particle physics, which can't@ be vectorized at all, but run happily on parallel architectures.    D > Cray also had multiprocessing down pat...spreading a given programE > across several CPUs with little or no need for programmers to worrylG > about it...the compilers took care of it automatically.  If there was C > somthing in the code that would prevent this from being done, then+ > compilers would note it in the listing.  f > G > Cray also had a really efficient OS (COS), with some hardware supportOA > for doing things like figuring out which software interrupts toR; > service next (for system service calls...dispatch was oneID > instruction...it was based on a bit vector).  If VMS is a CadillacD > with ride comfort and lots of gadgets, and MS-DOS is a go kart forA > playing around with in your spare time, then COS was a top fueldG > dragster...anything that didn't help it go fast was stripped off, andd4 > comfort weasn't a factor considered in the design. > ; > Even when running scalar only a Cray was faster than mostiE > computers...that 8.5 ns clock rate, separate ALUs that could run ateH > the same time in many cases (allowing instruction execution overlap ifE > coded correctly, and the compilers were pretty good at doing that), H > and a bunch of other techniques since copied by the better single chipH > CPUs, along with use of ECL, short path lengths, no busses to conflictF > (all point-to-point wiring to connect modules), etc., and the resultB > was the fastest computers on the planet for many years.  NeatestC > looking ones too...both the "bench seat" Cray 1 and X/MPs and the  > "aquarium" Cray 2 types.  & Sure, but what a waste of CPU cycles !G I remember that on some supercomputer sites they required codes to have B a high degree of vectorization in order to be allowed to use their expensive Crays.C And again, a 8.5 ns clock rate is not orders of magnitude differenta4 from a 3090 or any other supercomputer at that time.    o3 > >The fraction of such codes is probably even less D > >than that of parallel codes. Which might be a reason why we don'tX > >see that many vector machines anymore these days (apart from Earth Simulator maybe).  > F > If you consider all software, maybe.  In the science and engineeringH > area I doubt that that's true.  Arrays and operations on them are veryH > common.  There's a reason the Livermore Loops and Whetstone benchmarks > were heavy on them.s  P Again, don't generalize. Today's more realistic performance measures (eg SpecFP)% aren't vector centric anymore, AFAIK.t   ------------------------------  # Date: Thu, 09 Sep 2004 12:43:23 GMT9# From: nospam@nouce.bellatlantic.neta+ Subject: Re: why mainframes are still used?o8 Message-ID: <lii0k0lhs11nc20thdqnecr1v3thlac0ea@4ax.com>  0 On Thu, 09 Sep 2004 00:51:13 -0400, Mike Bartman" <omni@foolie.omniphile.com> wrote:  8 >Just about all computers at the time were larger than aD >refrigerator...that was about 1978 when the 11/780 was first built.@ >At that time the only small computers were basically toys.  TheF >Commodore PET, the "Trash-80" (Radio Shack's TRS-80), the Imsai 8080,D >the Apple I, and maybe some S-100 bus machines that were verging onD >actually useful for more than playing with for fun and education.    A In 1978 the LSI-11 (PDP-11) was desktop sized as were other smally machines like the DG Nova.  C Serious error there.  There were such machines as the PDP-8, PDP-11dB and  I can't list the IBM small machines that were in the class of4 larger than a hatbox and smaller than refigerator.    C For the "hobby" machines like TRS80 (always a toy) and Apple][ 1978.D was the landfall year and many of the S100 crates were just startingC to get serious enough disk systems to make them functional as smalleE business systems.  The real differentiator for that class of machinestE was storage systems and software.  It would take killer apps such as  E Wordstar, Visicalc and Dbase to move them into the serious use arena.:  G >"Super-mini" was an apt term, as the "minis" were things like PDP-11s,l? >which were smaller in terms of memory capacity, CPU speed, and0C >sometimes even physical size (depending on the model and options).R  @ Yes, exactly.  Superminis were those designs that  exceeded that: performance level in that, for the same size they were far? faster/capable.  They had to be smaller than their contemporaryaC mainframes but easily far faster and exapndable than that companies4D top of the line.  Examples were DEC VAX11/780 and DG Eclipse.   CostB was also a factor, Superminis gave users near mainframe power and - far less than mainframe costs (and overhead).t  @ By the mid '80s the idea of mainframe and mini was blured as the> differences were often hard to easily catagorize.  During that@ timeframe the ide that Superminies lacked the IO processors thatD mainframes posessed was one such differentiator but with clustering,F network based LAT servers and devices like the HSC50 the VAX line took on mainframe charateristics.  C There was never a hard wall from mainframe to smaller systems.  The D declining cost of computing hardware was a deciding factor and maybe; the only one.  The enduring difference between contemporary4F distributed systems and the classic mainframe was centralized softwareB and control.  Even this is challenged  when you consider a rack of blade servers in a small room.   Allisono   ------------------------------   Date: 9 Sep 2004 08:00:46 -0500g; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)i+ Subject: Re: why mainframes are still used?g3 Message-ID: <pCGtTdo8l+Js@eisner.encompasserve.org>    In article <gunvj0t3qte5sekht4293fnolpnla2ooq1@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes:> On 8 Sep 2004 14:08:13 -0500, koehler@eisner.nospam.encompasserve.org > 9 > Just about all computers at the time were larger than ak > refrigerator  E    Our LSI-11, PDP-11/34, PDP-11/23, MicroPDP-8, ... were not bigger  E    than a refigerator.  I'm not sure DEC ever built a PDP-8 that was  F    bigger than a refrigerator, although the layout tended to use more ;    floor space.  DECSYSTEM 2020 were also physically small.a  F    Of these I think only the PDP-11/23 did not predate the VAX 11/780.   ------------------------------  $ Date: Thu, 9 Sep 2004 06:06:00 -0700# From: "Tom Linden" <tom@kednos.com>r+ Subject: RE: why mainframes are still used? 9 Message-ID: <NDEMLKKEBOIFBMJLCECIOEELDMAA.tom@kednos.com>e   < -----Original Message-----% < From: nospam@nouce.bellatlantic.neti( < [mailto:nospam@nouce.bellatlantic.net], < Sent: Thursday, September 09, 2004 5:43 AM < To: Info-VAX@Mvb.Saic.Comr- < Subject: Re: why mainframes are still used?a <  < 2 < On Thu, 09 Sep 2004 00:51:13 -0400, Mike Bartman$ < <omni@foolie.omniphile.com> wrote: < : < >Just about all computers at the time were larger than aF < >refrigerator...that was about 1978 when the 11/780 was first built.B < >At that time the only small computers were basically toys.  TheH < >Commodore PET, the "Trash-80" (Radio Shack's TRS-80), the Imsai 8080,F < >the Apple I, and maybe some S-100 bus machines that were verging onF < >actually useful for more than playing with for fun and education.   < C < In 1978 the LSI-11 (PDP-11) was desktop sized as were other small. < machines like the DG Nova. < E < Serious error there.  There were such machines as the PDP-8, PDP-11sD < and  I can't list the IBM small machines that were in the class of6 < larger than a hatbox and smaller than refigerator.   < E < For the "hobby" machines like TRS80 (always a toy) and Apple][ 1978nF < was the landfall year and many of the S100 crates were just startingE < to get serious enough disk systems to make them functional as small G < business systems.  The real differentiator for that class of machinesgG < was storage systems and software.  It would take killer apps such as eG < Wordstar, Visicalc and Dbase to move them into the serious use arena.T < I < >"Super-mini" was an apt term, as the "minis" were things like PDP-11s,rA < >which were smaller in terms of memory capacity, CPU speed, and E < >sometimes even physical size (depending on the model and options).n < B < Yes, exactly.  Superminis were those designs that  exceeded that< < performance level in that, for the same size they were farA < faster/capable.  They had to be smaller than their contemporarypE < mainframes but easily far faster and exapndable than that companiespF < top of the line.  Examples were DEC VAX11/780 and DG Eclipse.   CostD < was also a factor, Superminis gave users near mainframe power and / < far less than mainframe costs (and overhead).c  I I think Supermini was used to designate the first 32 bit minis like Prime 	 Interdata  < B < By the mid '80s the idea of mainframe and mini was blured as the@ < differences were often hard to easily catagorize.  During thatB < timeframe the ide that Superminies lacked the IO processors thatF < mainframes posessed was one such differentiator but with clustering,H < network based LAT servers and devices like the HSC50 the VAX line took < on mainframe charateristics. < E < There was never a hard wall from mainframe to smaller systems.  The F < declining cost of computing hardware was a deciding factor and maybe= < the only one.  The enduring difference between contemporary H < distributed systems and the classic mainframe was centralized softwareD < and control.  Even this is challenged  when you consider a rack of  < blade servers in a small room. < 	 < Allisont <  < ---M( < Incoming mail is certified Virus Free.< < Checked by AVG anti-virus system (http://www.grisoft.com).B < Version: 6.0.742 / Virus Database: 495 - Release Date: 8/19/2004 <  --- & Outgoing mail is certified Virus Free.: Checked by AVG anti-virus system (http://www.grisoft.com).@ Version: 6.0.742 / Virus Database: 495 - Release Date: 8/19/2004   ------------------------------  $ Date: Thu, 9 Sep 2004 09:29:38 -04005 From: "Bochnik, William J" <William_Bochnik@acml.com>i+ Subject: RE: why mainframes are still used?tQ Message-ID: <2D75787AAF09C64481BDFD89113BE6D504ADA81F@ac2kama0102.ac.lp.acml.com>I  G >But the question I have for those Windoze folks: how many Server boxes0E would you have to use to achieve the same results?  >All of the usersiG being able to access the same databases for a particular region, at thes same time? =20  H and how many sysadmins would have been needed to mainain all the windoze machines=20t   -----Original Message-----5 From: Michael Jenkins [mailto:mwjenkins@excel.com]=20,, Sent: Wednesday, September 08, 2004 10:18 PM To: Info-VAX@Mvb.Saic.Com + Subject: Re: why mainframes are still used?i    E At my last assignment, the client company had six alpha machines, the D largest being a GS1280.  At one particular installation, I monitored over 5000 (yes,tG thousand) users logged in and 5000 concurrent processes, all running atsE the same time.  One operating system, OVMS 7.3, several applications,hE several databases, per single alpha machine.  The O/S & the apps were D the same on each machine. The basic database structure was the same,! only the contents were different.e  H Personal Computers and/or "dumb" terminals were used by the users.  I do= not know how many of either were being used at each site. =20   F But the question I have for those Windoze folks: how many Server boxesD would you have to use to achieve the same results?  All of the usersG being able to access the same databases for a particular region, at thet same time? =20   MikeWJ  ) arjdombivli@indiatimes.com (ashok) wrote:t  
 >dear all, >i >i > G >In the today's age of supercomputer ,why mainframes are still used?=207J >which are the operating systems for mainframes? How they are different=20> >from the PC? which are the mainframe manufacturing companies? > # >sorry to trouble you like this....f >  >a >regds,  >ypj  & http://www.geocities.com/mwjenkins001/    ) ----------------------------------------- L The information contained in this transmission may contain privileged and c=L onfidential information and is intended only for the use of the person(s) n=L amed above. If you are not the intended recipient, or an employee or agent =L responsible for delivering this message to the intended recipient, any revi=L ew, dissemination, distribution or duplication of this communication is str=L ictly prohibited. If you are not the intended recipient, please contact the=L  sender immediately by reply e-mail and destroy all copies of the original =L message. Please note that we do not accept account orders and/or instructio=L ns by e-mail, and therefore will not be responsible for carrying out such o= rders and/or instructions.L If you, as the intended recipient of this message, the purpose of which is =L to inform and update our clients, prospects and consultants of developments=L  relating to our services and products, would not like to receive further e=L -mail correspondence from the sender, please "reply" to the sender indicati=L ng your wishes.  In the U.S.: 1345 Avenue of the Americas, New York, NY 101= 05.i   ------------------------------  $ Date: Thu, 9 Sep 2004 11:11:10 -0400# From: "John Smith" <a@nonymous.com>w+ Subject: Re: why mainframes are still used? , Message-ID: <mLednffmIKYR793cRVn-vw@igs.net>   Bochnik, William J wrote:-C >> But the question I have for those Windoze folks: how many Serverm >> boxesG > would you have to use to achieve the same results?  >All of the userseE > being able to access the same databases for a particular region, atp > theg > same time? > B > and how many sysadmins would have been needed to mainain all the	 > windoze 
 > machines    > Dontcha know? One instance of Windows Server running on one HPL industry-standard server using one Itanic processor (any variety, since they9 are all extremely powerful) ought to handle the job.  ;-)    ------------------------------   Date: 9 Sep 2004 10:15:31 -0500a; From: koehler@eisner.nospam.encompasserve.org (Bob Koehler)n+ Subject: Re: why mainframes are still used?P3 Message-ID: <Jjw+Ctos9cLy@eisner.encompasserve.org>r  i In article <gunvj0t3qte5sekht4293fnolpnla2ooq1@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes:n > 9 > Just about all computers at the time were larger than a  > refrigerator  @    Olivetti/Underwood produced a desktop computer in the 1960's.E    IIRC it had 2 registers, 16 words of data storage, and 32 words ofoE    instruction storage.  Which made it less powerfull than my 10 year &    old programmable pocket calculator.  B    To program it there were buttons representing each register and>    operation, and a numeric keypad for the data and addresses.   ------------------------------  % Date: Thu, 09 Sep 2004 11:40:46 -0400M. From: Mike Bartman <omni@foolie.omniphile.com>+ Subject: Re: why mainframes are still used?n8 Message-ID: <s1s0k09h0f52hvunc1drompb3ko2hcj1hb@4ax.com>  B On Thu, 9 Sep 2004 12:01:08 +0000 (UTC), m.kraemer@gsi.de (Michael Kraemer) wrote:e  j >In article <e3muj0h1bntovb5cm2ge9pi3dv1df28prg@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes: >> cI >> I don't know what they had in a 3090.  The X/MP 48 had 4 CPUs, and the D >> 890 MFLOPS number was using all of them (it was done by one of myB >> co-workers for a pre-sales demo using customer-supplied FortranH >> code...with some mods by us).  The "48" in an X/MP's model name tellsF >> you the CPU count and memory size: 48 = 4 CPUs and 8 million 64-bit9 >> words of memory.  The one I worked on at NRL was a 12.  >k( >Now this sounds more reasonable too me.& >A single CPU doing around 250 MFLOPS.: >I further assume that this was on highly vectorized code.7 >Given the vector speedup of 10 you quoted below leavesd0 >a single CPU with 25 MFLOPS scalar performance.4 >According to you a 3090 does also 25 MFLOPs scalar.% >So where exactly is the difference ?v  4 1) We don't know how many CPUs the 3090 was using.    A Since I'm not an IBM person, I have no idea if it was a multi-CPUa system or not.  E 2) You are comparing one system's slowest capabilities to the other's-
 fastest.    C Making the Indy car run in first gear to make the go kart look lessv= pitiful isn't really much of a demonstration for the go kart.a   See the difference?u  A >No doubt Cray's were "real" supercomputers and faster than IBM's*C >but not by your 890/25 ratio. IIRC you could buy a 6-CPU 3090 with H >vector features, capable of doing several 100 MFLOPs on good vectorized9 >code. *This* is the number to be compared with your 890.i  A Then IBM probably should have advertised that setup instead.  I'm F basing the 25 MFLOPS figure on an IBM ad I read at the time (late 80s)F which claimed this 25 MFLOPS machine was a supercomputer.  Since I wasF working for Cray and using real supercomputers on a daily basis at the time, I found it laughable.4  D I wasn't in sales, but I heard about our competition all the time inC meetings.  Fujitsu, Hitachi and ETA were mentioned frequently.  IBMtD never came up.  Those who were in the market for supercomputers cameC to the above named companies, and generally had each one prove thatDF they were fastest for the work to be done, then bought that one.  ThisF was usually Cray.  If IBM had a real supercomputer they'd have been inC the running, and I'd have heard about them in those meetings.  That  didn't happen.  H >And I highly doubt that IBM praised their barebone scalar 3090 as being6 >supercomputer, it was certainly meant including a VF.  D The IBM ad I read claimed that the 3090 was a supercomputer and thatA it could do 25 MFLOPS.  Whether that is "praising" it or not I'll  leave up to you.  F >> >The same logic, however, holds also for vectorized vs scalar code,E >> >so your Cray was only worth its money for highly ( maybe >> 80% )t >> >vectorizable code. >> lA >> Perhaps, but vector processing is really common in science andIH >> engineering (lots of array operations), which is where supercomputersH >> are needed.  With business processing it's usually more a case of I/OG >> speed than raw CPU power that matters.  You are usually doing simpleaC >> things to lots of data, not complex things to a limited dataset.s >m# >I think you can't generalize this.EH >Certainly there are such codes working on large arrays with high degree@ >of vectorization, but there are as well such with lower degree,P >for which an IBM solution was more cost effective (and sure cost is a point !).I >And there are scientific problems, e.g. in particle physics, which can'trA >be vectorized at all, but run happily on parallel architectures.g  F Of course.  It's not whether codes exist, but what ratio of processing? a given area requires.  Business needs less vector, science and D engineering need more.  Business users might occasionally be able toA make use of vector capabilities, but probably not often enough to-D warrant the cost.  Science and engineering folks will sometimes needD scalar only, but their need for vector is frequent enough to make it worth the cost.  b  F For those problems that are nearly 100% parallel, more CPUs are alwaysA better.  I said up front that there were such problems I believe.NF Cray Research was really aware of what their customers needed in termsD of processing...it took an average of 2 years to sell a machine, andE the salesperson was in constant contact with the customers concerningt> their actual needs the whole time.  This information went intoB planning future designs.  Seymore stated on more than one occasionF that Cray's customers could make use of a maximum of 15 CPUs on almostB all codes run.  The economic balance point, where cost and utility? met, was at 4 CPUs...hence the X/MP and Cray 2 limit of 4 CPUs.nC Apparently this point changed later, after I left the company, as Ie= believe the Y/MP could take up to 15 CPUs.  The point is thateD apparently most people who needed supercomputers ran a lot of vectorF code that wasn't very parallel.  For those who ran almost all parallelB code, there were massively parallel processors available, and theyC were usually cheaper than a Cray (half to a third the cost, but notlC capable of running very fast unless the problem was almost entirely.F parallel).  Some places, like NRL (Naval Research Lab), got both sorts- of machines, and used each where it did best.t  E >> Cray also had multiprocessing down pat...spreading a given program-F >> across several CPUs with little or no need for programmers to worryH >> about it...the compilers took care of it automatically.  If there wasD >> somthing in the code that would prevent this from being done, the, >> compilers would note it in the listing.   >> oH >> Cray also had a really efficient OS (COS), with some hardware supportB >> for doing things like figuring out which software interrupts to< >> service next (for system service calls...dispatch was oneE >> instruction...it was based on a bit vector).  If VMS is a Cadillac E >> with ride comfort and lots of gadgets, and MS-DOS is a go kart forrB >> playing around with in your spare time, then COS was a top fuelH >> dragster...anything that didn't help it go fast was stripped off, and5 >> comfort weasn't a factor considered in the design.b >> s< >> Even when running scalar only a Cray was faster than mostF >> computers...that 8.5 ns clock rate, separate ALUs that could run atI >> the same time in many cases (allowing instruction execution overlap ifrF >> coded correctly, and the compilers were pretty good at doing that),I >> and a bunch of other techniques since copied by the better single chip I >> CPUs, along with use of ECL, short path lengths, no busses to conflicttG >> (all point-to-point wiring to connect modules), etc., and the resultnC >> was the fastest computers on the planet for many years.  NeatesttD >> looking ones too...both the "bench seat" Cray 1 and X/MPs and the >> "aquarium" Cray 2 types.  >-' >Sure, but what a waste of CPU cycles !.   Excuse me?  How so?i  H >I remember that on some supercomputer sites they required codes to haveC >a high degree of vectorization in order to be allowed to use their  >expensive Crays.   D That makes sense.  If your system is overloaded you get rid of thoseF that can't really make best use of it first, and leave it to those who can.  D >And again, a 8.5 ns clock rate is not orders of magnitude different5 >from a 3090 or any other supercomputer at that time.   E True, but the other speed features were pretty rare...that's why Cray C sold machines, even at the cost they had.  If you needed speed, the D Cray was it, and speed costs money...though in some environments NOTA having speed costs more.  That's why people paid what Cray asked.   C For example, oil companies.  They were a big part of Cray's market. > They did things like taking echo sounding data and calculatingC geologic strata from it.  They'd then simulate various drilling and ? extraction techniques to see which would work best on the given C geologic structure.  One potential customer of Cray at one time was D using IBM 370 machines to do this stuff.  It took them over 24 hoursC to do a run on a single dataset.  As part of their decision making,aB they set up their codes on a test machine at Cray.  When their guyE called the Cray salesman to find out the results so they could see ifeF they matched what they'd gotten on their existing machine, the resultsA files had been accidentally deleted by an operator (it was a testn> machine... ;-).  "Drat, I'll call you again this time tomorrowA then..." said the unhappy oil guy.  "No problem, hang on and I'lli= re-run them for you...just take a few minutes." said the CrayeE salesman.  15 minutes later they had results to compare...and the oillD guy was seriously amazed at the speed difference of the machine they were contemplating buying.  F At another company they'd run extraction technique simulations on partD of the north coast of Alaska field (20 billion barrels estimate) andE reached a conclusion.  Then they bought a Cray, and during the 30 day D acceptance testing they decided to re-run their simulation, but at a> higher resolution than they'd been able to run it on their oldD machine.  They found that a different extraction technique would getE them 2% more oil out of the ground than the one they'd have gone witht7 based on their older runs.  2% of 20 billion barrels atiD $20-$30/barrel?  That's 8 to 12 billion dollars.  That Cray paid forC itself many times over before it was even accepted.  The Cray salessE guy did try to renegotiate the deal so we got 2% of their 2%, but wase turned down.  ;-)o  4 >> >The fraction of such codes is probably even lessE >> >than that of parallel codes. Which might be a reason why we don'taY >> >see that many vector machines anymore these days (apart from Earth Simulator maybe).   >>  G >> If you consider all software, maybe.  In the science and engineeringeI >> area I doubt that that's true.  Arrays and operations on them are verytI >> common.  There's a reason the Livermore Loops and Whetstone benchmarks  >> were heavy on them. >eQ >Again, don't generalize. Today's more realistic performance measures (eg SpecFP)w& >aren't vector centric anymore, AFAIK.  B Who's generalizing here?  Seems you are.  The Livermore Loops were> entirely realistic.  They were taken from code actually run atD Livermore Labs, as a representation of what was actually in use on aA daily basis.  The Whetstone benchmark was from a similar exercise-E using engineering codes from many places.  That's not general, that'siC specific, and based on real world use.  Where did SpecFP come from?   
 -- Mike B.@ ----------------------------------------------------------------=   To reply via e-mail, remove the 'foolie.' from the address.a%   I'm getting sick of all the SPAM... @ ----------------------------------------------------------------   ------------------------------  % Date: Thu, 09 Sep 2004 11:53:53 -0400 . From: Mike Bartman <omni@foolie.omniphile.com>+ Subject: Re: why mainframes are still used?>8 Message-ID: <fdu0k0hafjvs86dq5rqealtsvtpukbt420@4ax.com>  F On Thu, 09 Sep 2004 12:43:23 GMT, nospam@nouce.bellatlantic.net wrote:  1 >On Thu, 09 Sep 2004 00:51:13 -0400, Mike Bartmani# ><omni@foolie.omniphile.com> wrote:  >19 >>Just about all computers at the time were larger than a E >>refrigerator...that was about 1978 when the 11/780 was first built.-A >>At that time the only small computers were basically toys.  TheaG >>Commodore PET, the "Trash-80" (Radio Shack's TRS-80), the Imsai 8080,eE >>the Apple I, and maybe some S-100 bus machines that were verging oniE >>actually useful for more than playing with for fun and education.  - >-B >In 1978 the LSI-11 (PDP-11) was desktop sized as were other small >machines like the DG Nova._  A LSI-11 was a very specific version of the PDP-11...and yes it was < small.  There was one in the main cabinet of the 11/780 thatF functioned as a console controller and boot controller (it's what readC the 8" floppy with the WCS info and boot progs).  Most PDP-11s weregF refrigerator sized though...like the 11/70.  They were in standard 19"C racks (the ones I saw anyway).  Those were not only standalone, butsD were also embedded in other systems...such as the PDP 11/40 (I think? that's the model) that was used as a terminal controller in thee= DECsystem-10 (PDP-10) and referred to as a "DN-87" (size of a  refrigerator).  ? >There was never a hard wall from mainframe to smaller systems.   C True.  It was mostly a marketting distinction, though most folks innB the industry could look at one and tell you what it was.  Giving aF hard and fast definition that always worked was pretty much impossibleD though.  Any such distinction was usually based on a comparison withD other systems, and which other systems were around to be compared toC had as much to do with the definition as anything else.  VAXen wereTD "super-minis" because they were bigger and more powerful than PDP-11F "minicomputers".  "Microcomputers" were that because they were smallerD and less powerful than the minis.  Mainframes were those things thatB were bigger and more powerful and with greater resource needs thanE super-minis.  Take the VAX-9000.  Was it a mainframe or just the nexti? larger super-mini in the VAX line?  Call it any way you see it..  
 -- Mike B.
 -- Mike B.  
 '04 FLSTCI  @ ----------------------------------------------------------------=   To reply via e-mail, remove the 'foolie.' from the address.e%   I'm getting sick of all the SPAM...a@ ----------------------------------------------------------------   ------------------------------  % Date: Thu, 09 Sep 2004 12:28:50 -0400w. From: Mike Bartman <omni@foolie.omniphile.com>+ Subject: Re: why mainframes are still used? 8 Message-ID: <7u01k0h5bkmojih5t2rrc21q2ndh03jhvo@4ax.com>  E On 9 Sep 2004 10:15:31 -0500, koehler@eisner.nospam.encompasserve.orgu (Bob Koehler) wrote:  j >In article <gunvj0t3qte5sekht4293fnolpnla2ooq1@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes: >>  : >> Just about all computers at the time were larger than a >> refrigeratore > A >   Olivetti/Underwood produced a desktop computer in the 1960's.wF >   IIRC it had 2 registers, 16 words of data storage, and 32 words ofF >   instruction storage.  Which made it less powerfull than my 10 year' >   old programmable pocket calculator.l > C >   To program it there were buttons representing each register andM? >   operation, and a numeric keypad for the data and addresses.i  A I guess it depends how you define "computer" then.  That's a niceoE calculator, but not much of a general purpose computer...not compared 6 to real computers available in the 60s (IBM 360, etc.)  ? It might also depend on how you define "refrigerator" (Walk in? > Upright?  Side-by-side?  Chest?  Dorm?  ;-)  I was picturing a' standard upright or chest model.  YMMV.d  5   -- Mike  "this is starting to get silly" Bartman --o
 -- Mike B.  
 '04 FLSTCI  @ ----------------------------------------------------------------=   To reply via e-mail, remove the 'foolie.' from the address.h%   I'm getting sick of all the SPAM...p@ ----------------------------------------------------------------   ------------------------------  % Date: Thu, 09 Sep 2004 12:42:52 -0400 # From: sol gongola <sol@adldata.com>e+ Subject: Re: why mainframes are still used? + Message-ID: <4140880C.C14CA446@adldata.com>    Mike Bartman wrote:i > G > On 9 Sep 2004 10:15:31 -0500, koehler@eisner.nospam.encompasserve.orgt > (Bob Koehler) wrote: > l > >In article <gunvj0t3qte5sekht4293fnolpnla2ooq1@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes: > >>< > >> Just about all computers at the time were larger than a > >> refrigeratoro > >oC > >   Olivetti/Underwood produced a desktop computer in the 1960's.hH > >   IIRC it had 2 registers, 16 words of data storage, and 32 words ofH > >   instruction storage.  Which made it less powerfull than my 10 year) > >   old programmable pocket calculator.o > >rE > >   To program it there were buttons representing each register and A > >   operation, and a numeric keypad for the data and addresses.r > C > I guess it depends how you define "computer" then.  That's a nicebG > calculator, but not much of a general purpose computer...not compared08 > to real computers available in the 60s (IBM 360, etc.) > A > It might also depend on how you define "refrigerator" (Walk in?5@ > Upright?  Side-by-side?  Chest?  Dorm?  ;-)  I was picturing a) > standard upright or chest model.  YMMV.t > 7 >   -- Mike  "this is starting to get silly" Bartman --C > -- Mike B.  > My college had the same or similar device around that time. It: was a central unit with multiple keyboard/display stations> spread around the room. The computer was a fancy programmable ? calculator for that time period and it was also a time sharing i	 system :)    soll   ------------------------------  # Date: Thu, 09 Sep 2004 16:57:11 GMTn# From: nospam@nouce.bellatlantic.netc+ Subject: Re: why mainframes are still used?r8 Message-ID: <s821k0d9u92u4ssvt1so5bkadjlm2m6tq1@4ax.com>  0 On Thu, 09 Sep 2004 11:53:53 -0400, Mike Bartman" <omni@foolie.omniphile.com> wrote:B >LSI-11 was a very specific version of the PDP-11...and yes it was= >small.  There was one in the main cabinet of the 11/780 thatsG >functioned as a console controller and boot controller (it's what readeD >the 8" floppy with the WCS info and boot progs).  Most PDP-11s wereG >refrigerator sized though...like the 11/70.  They were in standard 19"eD >racks (the ones I saw anyway).  Those were not only standalone, butE >were also embedded in other systems...such as the PDP 11/40 (I thinki@ >that's the model) that was used as a terminal controller in the> >DECsystem-10 (PDP-10) and referred to as a "DN-87" (size of a >refrigerator).g  D Gee that makes the 11/44 I used with  RL02 in a short cab(44") a bar fridge then.    C The PDP10 in my mind was never anything less than a small mainframea= and I do remember the 4 CPU PDP-10 cluster as rather large.  o  @ If all you saw were the larger implmentations then they were allD large.   The bulk of the PDP-11 series had processor boxes under 20"@ high (most were under 11").  So it was often a system in a rack.A Whereas the PDP-11/780 the CPU nearly filled a rack and was not au0 system without a few more cabs never mind disks.  E To make sense of that you have to keep the timeframe and contemporarya systems in mind.       Allisony   ------------------------------  # Date: Thu, 09 Sep 2004 17:06:12 GMT'# From: nospam@nouce.bellatlantic.netn+ Subject: Re: why mainframes are still used?t8 Message-ID: <6u21k0h6lc56frbud7amkirhcefarver1i@4ax.com>  E On 9 Sep 2004 08:00:46 -0500, koehler@eisner.nospam.encompasserve.orgh (Bob Koehler) wrote:   >In article <gunvj0t3qte5sekht4293fnolpnla2ooq1@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes:> On 8 Sep 2004 14:08:13 -0500, koehler@eisner.nospam.encompasserve.org  >> d: >> Just about all computers at the time were larger than a >> refrigerator, > F >   Our LSI-11, PDP-11/34, PDP-11/23, MicroPDP-8, ... were not bigger F >   than a refigerator.  I'm not sure DEC ever built a PDP-8 that was G >   bigger than a refrigerator, although the layout tended to use more P< >   floor space.  DECSYSTEM 2020 were also physically small.  A The PDP12 (pdp-8+link8)  were 1965-67 and they were a bit larger.r; The 11/23 was ony a little (not more than a year) after VAXe
 introduction.    Allisonr   ------------------------------  * Date: Thu, 9 Sep 2004 17:32:32 +0000 (UTC)( From: m.kraemer@gsi.de (Michael Kraemer)+ Subject: Re: why mainframes are still used?f4 Message-ID: <chq43g$ag$1@lnx107.hrz.tu-darmstadt.de>  i In article <s1s0k09h0f52hvunc1drompb3ko2hcj1hb@4ax.com>, Mike Bartman <omni@foolie.omniphile.com> writes:tD > On Thu, 9 Sep 2004 12:01:08 +0000 (UTC), m.kraemer@gsi.de (Michael > Kraemer) wrote:e  6 OK, I'm not very keen on arguing over issues long gone% (3090 and Cray vector supercomputers)m< and certainly not keen on entering a Bob vs Andrew dogfight, so here's my last one ...s     > >,* > >Now this sounds more reasonable too me.( > >A single CPU doing around 250 MFLOPS.< > >I further assume that this was on highly vectorized code.9 > >Given the vector speedup of 10 you quoted below leavese2 > >a single CPU with 25 MFLOPS scalar performance.6 > >According to you a 3090 does also 25 MFLOPs scalar.' > >So where exactly is the difference ?5 > 6 > 1) We don't know how many CPUs the 3090 was using.   >  > Since I'm not an IBM person, t  I Nor am I. Just a person who has worked on IBM boxes, vector and parallel.a  & > I have no idea if it was a multi-CPU > system or not.  > So you can't bring up your 25 vs 890 MFLOPs comparison either.  RG > 2) You are comparing one system's slowest capabilities to the other'sh > fastest.   > E > Making the Indy car run in first gear to make the go kart look less ? > pitiful isn't really much of a demonstration for the go kart.l  G Your example quotes the Indy (I admit I know nothing about racing cars)hD running in its highest gear, and the "go kart" in some unknown gear.) I tried to compare them in the same gear.b  a > See the difference? C > >No doubt Cray's were "real" supercomputers and faster than IBM'seE > >but not by your 890/25 ratio. IIRC you could buy a 6-CPU 3090 withtJ > >vector features, capable of doing several 100 MFLOPs on good vectorized; > >code. *This* is the number to be compared with your 890.f > C > Then IBM probably should have advertised that setup instead.  I'msH > basing the 25 MFLOPS figure on an IBM ad I read at the time (late 80s)H > which claimed this 25 MFLOPS machine was a supercomputer.  Since I wasH > working for Cray and using real supercomputers on a daily basis at the > time, I found it laughable.   = I remotely remember this attitude of Cray people towards IBM.iD Inspecting the current HPC chart I guess IBM has had the last laugh.  tF > I wasn't in sales, but I heard about our competition all the time inE > meetings.  Fujitsu, Hitachi and ETA were mentioned frequently.  IBMhF > never came up.  Those who were in the market for supercomputers cameE > to the above named companies, and generally had each one prove thatmH > they were fastest for the work to be done, then bought that one.  ThisH > was usually Cray.  If IBM had a real supercomputer they'd have been inE > the running, and I'd have heard about them in those meetings.  Thatp > didn't happen.  4 If it's about top speed, maybe, but also in terms ofI units shipped ? Where's the limit to qualify a machine as a supercomputeri anyway ?   > F > >And again, a 8.5 ns clock rate is not orders of magnitude different7 > >from a 3090 or any other supercomputer at that time.o > G > True, but the other speed features were pretty rare...that's why CrayeE > sold machines, even at the cost they had.  If you needed speed, thelF > Cray was it, and speed costs money...though in some environments NOTC > having speed costs more.  That's why people paid what Cray asked.   E Sure, weather services come to mind. You'd better finish the forecast ' before the weather actually arrives ...pA But for most others Crays were not economical, for them a 3090 VFs@ was enough, with still respectable performance ( >> 25 MFLOPs ).D That's what I wanted to say, there's nothing "laughable" about that.    E > For example, oil companies.  They were a big part of Cray's market.e@ > They did things like taking echo sounding data and calculatingE > geologic strata from it.  They'd then simulate various drilling andDA > extraction techniques to see which would work best on the given E > geologic structure.  One potential customer of Cray at one time waseF > using IBM 370 machines to do this stuff.  It took them over 24 hoursE > to do a run on a single dataset.  As part of their decision making,PD > they set up their codes on a test machine at Cray.  When their guyG > called the Cray salesman to find out the results so they could see ifsH > they matched what they'd gotten on their existing machine, the resultsC > files had been accidentally deleted by an operator (it was a test-@ > machine... ;-).  "Drat, I'll call you again this time tomorrowC > then..." said the unhappy oil guy.  "No problem, hang on and I'll ? > re-run them for you...just take a few minutes." said the CraysG > salesman.  15 minutes later they had results to compare...and the oil@F > guy was seriously amazed at the speed difference of the machine they > were contemplating buying.  # Oh come on, this is a story of sort1? "compare the oldest box you want to get rid of with the new onew5  you want to buy and wonder how slow the old one is".8 Doesn't proof anything.,   ------------------------------  # Date: Thu, 09 Sep 2004 17:39:25 GMTw# From: nospam@nouce.bellatlantic.netr+ Subject: Re: why mainframes are still used?.8 Message-ID: <1m31k05qbm74j00q9rka0mth08kcabc7jp@4ax.com>  0 On Thu, 09 Sep 2004 11:53:53 -0400, Mike Bartman" <omni@foolie.omniphile.com> wrote:   > B >LSI-11 was a very specific version of the PDP-11...and yes it was >small.a    I know that, I still have a few.  7 >  There was one in the main cabinet of the 11/780 thathG >functioned as a console controller and boot controller (it's what readd2 >the 8" floppy with the WCS info and boot progs).   D Microcode loader and boot.  the F11 and J11 and T11 series of micros1 were widely used by DEC internal to many systems.f   > Most PDP-11s wereeG >refrigerator sized though...like the 11/70.  They were in standard 19"sD >racks (the ones I saw anyway).  Those were not only standalone, butE >were also embedded in other systems...such as the PDP 11/40 (I think @ >that's the model) that was used as a terminal controller in the> >DECsystem-10 (PDP-10) and referred to as a "DN-87" (size of a >refrigerator).i  D Well my PDP-8f is a deskside system and it plus disk fits in roughlyE 24" of 19" rack (left side of a desk actually). My expanded 11/73 is  F in a 50inch rack but that's more for the RL02, RX02 and two bookshevesE in the rack for docs.   I have a rather large and eclectic collection D of old computers including many Q-bus PDP-11s, VAXen (UVAX-II/gps in< BA123 plus numerous uVAX3100s) and  S100 Z80 and T-11 based 	 machines.e  @ The average PDP-11 I'd been exposed to was two to three racks ofA 44" or 72" in height.  The usual 72" racked PDP-11 was 60% air as C the racks were far from full and the cpu only accounted for a small 9 part of that.  Not large at all compared to the Ki-10. ;)h  D >True.  It was mostly a marketting distinction, though most folks inC >the industry could look at one and tell you what it was.  Giving a,  F At one time yes, later no.  Though at one point a distinction I'd comeD to know was CPUs with word lengths greater than 32-36 bits.  I think+ CDC with their 60bit monster promoted that.l  < Then again back in the days of the 7094 and 9020s small was ; anything smaller than a Buick.  As systems (especially CPU rE and memory) got smaller never mind faster the idea of super anything tE was a distinction that it was top of the line in performance and that & often it was packaged smaller as well.  @ A more useful distingtion was the "3M" (1megaflops, 1meg memory,C 1megapixel display) on a desk dating to the workstation wars of theoC late 80s/early 90s (sun/apollo/dec).  That distinction was a little  lower in vapor content. ;)     Allisonh   ------------------------------   End of INFO-VAX 2004.501 ************************e: 495 - Release Date: 8/19/2004   ------------------------------  $ Date: Thu, 9 Sep 2004 09:29:38 -04005 From: "Bochnik, William J" <William_Bochnik@acml.com>i+ Subject: RE: why mainframes are still used?tQ Message-ID: <2D75787AAF09C64481BDFD89113BE6D504ADA81F@ac2kama0102.ac.lp.acml.com>I  G >But the question I have for those Windoze folks: how many Server boxes0E would you have to use to achieve the same results?  >All of the usersiG being able tyz7\ՃI,?IkoPO39%+t$Ϝf@vٞodiWbfl%.X%j_ab؝jJ! 7cHH=^*koxFSˍ79ψzÜ덠Sxg/Q'&=ᴭ>ᓂ-de#uD=(D{8ݎ>QGicB}Pˑ}-	r7ȍѢ+%^	X,nkzjoE1;23: R)4~Сgv(iUi,?Ve4U%m*5/bhvn܂aM7a\Jy| 1qcs= }
 9X,8yv(O߹jk7bƆ;( |GBb3>=fme`c`f(TX
]o,ijlָ.ڸU7W6ܵym}Ƕ5oa!0Xf	45v\ӐEIɃ,d},Sr\umd1dP蓿j߁R7[M1$gTp
*29*ݏTN"O勫֬ܗ.V#/	,kd]wA9jrB4OPyYdǘ#T:S`{e,5THz7GxuOnLSYT>S1iD5|io$AnB<'8%tv7mcJi(
5WLR$RvGSCêM
$YҦݍ9Xv)IXPӫwA<}WCC 1
J6 W5d =D=Ac1<JB.D{e$Ar=TPV5b`֍ܥv