1 INFO-VAX	Sun, 11 Jun 2006	Volume 2006 : Issue 322       Contents:P Re: Brokeback Mountain II - The IT Guys  (Was Re: So how representative is this P Brokeback Mountain II - The IT Guys  (Was Re: So how representative is this expe' Re: Cost of used alphas vs 8086+charron ' Re: Cost of used alphas vs 8086+charron ' RE: Cost of used alphas vs 8086+charron ' Re: Cost of used alphas vs 8086+charron ' Re: Cost of used alphas vs 8086+charron ' Re: Cost of used alphas vs 8086+charron , FA: ULTRIX-11 V2 Manual Set and Install Tape Re: My Boot Camp trip report# Re: New HP web based email package!  Re: Results of my straw poll.   F ----------------------------------------------------------------------  % Date: Sat, 10 Jun 2006 19:04:13 -0700 # From: "Tom Linden" <tom@kednos.com> Y Subject: Re: Brokeback Mountain II - The IT Guys  (Was Re: So how representative is this  ) Message-ID: <op.tayldbpwzgicya@hyrrokkin>   4 On Sat, 10 Jun 2006 18:50:20 -0700, Richard Maher  =  $ <maher_rj@hotspamnotmail.com> wrote:   > Hi Tom, John,  >  >> Tom LINDEN wrote:- >> > What do you do for Cobol, wrt alignment?  > I > Most COBOL sites I know of compile for /NOALIGNMENT and let the compil=  erI > generate the bazillion extra instructions needed to prevent an alignme=  ntI > fault. During many VAX to Alpha COBOL migrations, a Phase II was plann=  edI > where performance critical sections of the code were to be re-visited =  and I > the compiler-directives (Yes we have those :-) *DC SET ALIGNMENT would=   be C > used so as to achieve the performance boost. As Alpha ended up so I > blisteringly fast (certainly for batch jobs) no one really bothered go=  ing  > back to tune the code. >  > "John Reagan"  wroteE >> The lack of pointers make it hard to lie and get alignment faults.  > I > We've got all the Pointers you want (Even a Pointer-64 apparently) It'=  s  =   > justI > that we don't have a lot of verbs that can do anything with them :-)  =    > Apart F > from POINTER VALUE REFERENCE ws_var. and SET ws_pointer TO REFERENCEI > ws/ls_var there's not much that can be done (Besides pointer-arithmati=  c  =   > and I > USING BY VALUE ws_pointer). I believe IBM COBOL has a SET ADDRESS OF  =    > ws_varI > TO ws_pointer (But I don't think it does what I hoped and I couldn't s=  ee  =    > a 9 > corresponding MAP statement or  *DC SET NOALLOCATE etc)  >  >> I'll be fixing those 8 >> over the next few months as I jump into the compiler. > F > Curious. Did Don Braffit die at his keyboard due to underfunding andI > atrophy? Any point in asking the new-broom for a list of new features =   =   > thatI > are coming up for COBOL (Local Storage? Recursion? Conforming to lates=  t 
 > standards?)  > I > Don't get me wrong, I'm over the moon if you're in charge of COBOL! (A=  s  =   > longI > as you haven't had to give up MACRO) And it's only 'cos I've been happ=  ily I > married for 11 years and am a man at peace with his sexuality, that I =   =   > don't I > mind saying that I'm quite fond of you John. (And I'm not just saying =   =   > thatI > 'cos I'm running out of anyone else to talk to :-) You don't say too m=  uch F > but I've never seen you turn down a Pascal or MACRO question here  =   > (certainlyI > not mine) and some of them obviously took a lot longer than 5mins to  =   	 > answer. ! > Why can't they all be like you?  > I > It's just that it looks like HP/VMS has 50 developers for each variant=   of I > it's many (this time for sure :-) C compilers, and one John Reagan for=   A > everything else. Did you pick up Don's SORT comitments as well?  >  > Regards Richard Maher  > I > PS. Still, it's not *all* doom and gloom for COBOL. At least we're in =   =   > better. > shape than PL/I 8-O OK, I'll get my coat :-)  G Two points worthy of note,  We continue getting new busines for PL/I on I Alphas and unlike Cobol programmers, PL/I programmers don't need viagra.=   F seriously Richard, I can't understand a man of your stature, depth andE complexity trifling with Cobol, when you could have the world in your 5 palm with PL/I.  Now, the genie is out of the bottle.    > I > PPS. I have no idea of Don Braffit's state of health or employment sta=  tus I > and zero knowledge (or interest) in anyone's personal foibles, so if I=  've I > managed to find another foot to stick in my mouth, it was unintentiona=  l. > 5 > "John Reagan" <john.reagan@hp.com> wrote in message - > news:R5hfg.1284$dG4.137@news.cpqcorp.net...  >> Tom LINDEN wrote: >> >> >- >> > What do you do for Cobol, wrt alignment?  >> > >> > >>C >> COBOL is different than the other compilers by defaulting to VAX I >> alignment rules.  You can asking for padding, but you don't get it by=   D >> default.  Other than that, the COBOL should behave like the other > compilers. >>I >> Binary fields that are known to be aligned can be accessed with a sin=  gle  >> instruction.  >>F >> Binary fields that are known not to be aligned can be accessed with@ >> multiple instructions that don't generate an alignment fault. >>I >> Character data is manipulated with RTL routines that are byte-oriente=  d . >> so there shouldn't be aligned faults there. >>E >> The lack of pointers make it hard to lie and get alignment faults.  >>F >> Now, with all that said, there are a couple of bugs in the compilerI >> where the front-end knows a field is unaligned, but it accidently tol=  d B >> GEM that the field was aligned.  GEM believes the front-end andI >> generates single instruction accesses that fault.  I'll be fixing tho=  seI >> over the next few months as I jump into the compiler.  Right now, I h=  ave I >> several COBOL examples that appear to be much slower on I64 than Alph=  a  >> all due to alignment faults.  >> >> --  >> John Reagan2 >> HP Pascal/{A|I}MACRO for OpenVMS Project Leader >> Hewlett-Packard Company >  >  >    ------------------------------  % Date: Sun, 11 Jun 2006 09:50:20 +0800 3 From: "Richard Maher" <maher_rj@hotspamnotmail.com> Y Subject: Brokeback Mountain II - The IT Guys  (Was Re: So how representative is this expe 1 Message-ID: <e6fspp$o73$1@news-02.connect.com.au>   
 Hi Tom, John,    > Tom LINDEN wrote: , > > What do you do for Cobol, wrt alignment?  H Most COBOL sites I know of compile for /NOALIGNMENT and let the compilerH generate the bazillion extra instructions needed to prevent an alignmentH fault. During many VAX to Alpha COBOL migrations, a Phase II was plannedI where performance critical sections of the code were to be re-visited and I the compiler-directives (Yes we have those :-) *DC SET ALIGNMENT would be A used so as to achieve the performance boost. As Alpha ended up so I blisteringly fast (certainly for batch jobs) no one really bothered going  back to tune the code.   "John Reagan"  wroteD > The lack of pointers make it hard to lie and get alignment faults.  L We've got all the Pointers you want (Even a Pointer-64 apparently) It's justJ that we don't have a lot of verbs that can do anything with them :-) ApartD from POINTER VALUE REFERENCE ws_var. and SET ws_pointer TO REFERENCEK ws/ls_var there's not much that can be done (Besides pointer-arithmatic and K USING BY VALUE ws_pointer). I believe IBM COBOL has a SET ADDRESS OF ws_var J TO ws_pointer (But I don't think it does what I hoped and I couldn't see a7 corresponding MAP statement or  *DC SET NOALLOCATE etc)    > I'll be fixing those7 > over the next few months as I jump into the compiler.   D Curious. Did Don Braffit die at his keyboard due to underfunding andJ atrophy? Any point in asking the new-broom for a list of new features thatG are coming up for COBOL (Local Storage? Recursion? Conforming to latest  standards?)   L Don't get me wrong, I'm over the moon if you're in charge of COBOL! (As longI as you haven't had to give up MACRO) And it's only 'cos I've been happily K married for 11 years and am a man at peace with his sexuality, that I don't J mind saying that I'm quite fond of you John. (And I'm not just saying thatI 'cos I'm running out of anyone else to talk to :-) You don't say too much L but I've never seen you turn down a Pascal or MACRO question here (certainlyL not mine) and some of them obviously took a lot longer than 5mins to answer. Why can't they all be like you?   I It's just that it looks like HP/VMS has 50 developers for each variant of F it's many (this time for sure :-) C compilers, and one John Reagan for? everything else. Did you pick up Don's SORT comitments as well?    Regards Richard Maher   L PS. Still, it's not *all* doom and gloom for COBOL. At least we're in better, shape than PL/I 8-O OK, I'll get my coat :-)  I PPS. I have no idea of Don Braffit's state of health or employment status I and zero knowledge (or interest) in anyone's personal foibles, so if I've H managed to find another foot to stick in my mouth, it was unintentional.  3 "John Reagan" <john.reagan@hp.com> wrote in message + news:R5hfg.1284$dG4.137@news.cpqcorp.net...  > Tom LINDEN wrote:  >  > > , > > What do you do for Cobol, wrt alignment? > >  > >  > B > COBOL is different than the other compilers by defaulting to VAXG > alignment rules.  You can asking for padding, but you don't get it by C > default.  Other than that, the COBOL should behave like the other 
 compilers. > J > Binary fields that are known to be aligned can be accessed with a single > instruction. > E > Binary fields that are known not to be aligned can be accessed with ? > multiple instructions that don't generate an alignment fault.  > H > Character data is manipulated with RTL routines that are byte-oriented- > so there shouldn't be aligned faults there.  > D > The lack of pointers make it hard to lie and get alignment faults. > E > Now, with all that said, there are a couple of bugs in the compiler H > where the front-end knows a field is unaligned, but it accidently toldA > GEM that the field was aligned.  GEM believes the front-end and I > generates single instruction accesses that fault.  I'll be fixing those J > over the next few months as I jump into the compiler.  Right now, I haveH > several COBOL examples that appear to be much slower on I64 than Alpha > all due to alignment faults. >  > --  
 > John Reagan 1 > HP Pascal/{A|I}MACRO for OpenVMS Project Leader  > Hewlett-Packard Company    ------------------------------  % Date: Sat, 10 Jun 2006 20:08:37 +0200 3 From: Wilm Boerhout <w5OLD.boerhout@PAINTplanet.nl> 0 Subject: Re: Cost of used alphas vs 8086+charron6 Message-ID: <448b0aa5$0$29769$ba620dc5@nova.planet.nl>  ' Dave Froble wrote on 10-6-2006 19:31...   0 > The slowest Alpha is much faster than an 8086! > $ > You're both being subverted by JF. > 0 > Perhaps you wish to discuss x86-64, or AMD-64?   ???   G We're discussing CHARON-VAX on Intel vs. CHARON-VAX on Alpha, starting  C from JF's question about CHARON-AXP on 8086. Are you still with us?    /Wilm    ------------------------------  % Date: Sat, 10 Jun 2006 15:17:55 -0400 2 From: "Stanley F. Quayle" <squayle@insight.rr.com>0 Subject: Re: Cost of used alphas vs 8086+charron; Message-ID: <448AE2A3.4787.2ED2FAB0@squayle.insight.rr.com>   ' On 9 Jun 2006 at 21:42, JF Mezei wrote: I > If one were to pitch VMS to a new customer, would a used Alpha come out 1 > cheaper than a 64 bit 8086 with charron-alpha ?   F Speaking as a CHARON-AXP reseller [Shameless Plug Alert (tm)], let me B help.  First, I'd need to clarify the performance you need.  If a A $1,000 DS-10L from Island Computer is all you need, and if David  A still has some, that would be your direction.  Especially if you  ? don't have some Alpha licenses to transfer to the "new" system.   ? > What sort of MHZ/ghz would a modern 8086 server require, with ; > charron-alpha, to match  a ds15/25 machine's perfomance ?   D The currently-available CHARON-Alpha models are available on my web  site:   )    http://www.stanq.com/charon-alpha.html   J > i.e. will industry standard servers with the charron alpha emulator soonF > start to surpass existing alphas in terms of performance and cost of > ownership ?   E I'm counting on Moore's Law for the foreseeable future.  By starting  C NOW with Alpha emulation, we're in position for 5-10-15 years down  F the road.  Like a surfer, if we start swimming now, the big wave will 	 catch us.   H > or must we wait for a native VMS port to the 64 bit 8086 before we see< > the 8086 beat alpha in performance and cost of ownership ?  E A version of VMS that's native on platform "X" is always going to be  @ faster that emulating it on "X".  But don't hold your breath -- F everyone at VMS Engineering swears that they're NOT working an 8086.  = If you want VMS on Intel today, get an Itanium.  Or CHARON-*.   
 --Stan Quayle  Quayle Consulting Inc.  
 ----------8 Stanley F. Quayle, P.E. N8SQ  Toll free: 1-888-I-LUV-VAX3 8572 North Spring Ct., Pickerington, OH  43147  USA < stan-at-stanq-dot-com   http://www.stanq.com/charon-vax.html) "OpenVMS, when downtime is not an option"    ------------------------------  % Date: Sat, 10 Jun 2006 15:51:11 -0400 ' From: "Main, Kerry" <Kerry.Main@hp.com> 0 Subject: RE: Cost of used alphas vs 8086+charronT Message-ID: <FA60F2C4B72A584DBFC6091F6A2B86840155E3EF@tayexc19.americas.cpqcorp.net>   > -----Original Message-----9 > From: JF Mezei [mailto:jfmezei.spamnot@teksavvy.com]=20  > Sent: June 9, 2006 9:42 PM > To: Info-VAX@Mvb.Saic.Com . > Subject: Cost of used alphas vs 8086+charron >=20 > Just curious:  >=20= > If one were to pitch VMS to a new customer, would a used=20  > Alpha come out1 > cheaper than a 64 bit 8086 with charron-alpha ?  >=20 >=20? > What sort of MHZ/ghz would a modern 8086 server require, with ; > charron-alpha, to match  a ds15/25 machine's perfomance ?  >=20? > i.e. will industry standard servers with the charron alpha=20  > emulator soon F > start to surpass existing alphas in terms of performance and cost of > ownership ?=20 >=20H > or must we wait for a native VMS port to the 64 bit 8086 before we see< > the 8086 beat alpha in performance and cost of ownership ? >=20    D Mmmm... Don't forget that on top of normal VMS licensing, the CharonG license is definitely not free and is a big expense to be considered .. G Not sure what the latest list prices are, but they are not small (or at 1 least for the VAX 66xx versions I saw quoted).=20   B Emulation has its place, but native ports and upgrading to currentD versions is imho, the preferred approach if one does not have source/ code or ISV vendor challenges with upgrades.=20    Regards   
 Kerry Main Senior Consultant  HP Services Canada Voice: 613-592-4660  Fax: 613-591-4477  kerryDOTmainAThpDOTcom (remove the DOT's and AT)=20  4 OpenVMS - the secure, multi-site OS that just works.   ------------------------------  % Date: Sat, 10 Jun 2006 15:43:17 -0400 6 From: =?ISO-8859-1?Q?Arne_Vajh=F8j?= <arne@vajhoej.dk>0 Subject: Re: Cost of used alphas vs 8086+charron. Message-ID: <jhFig.40898$fG3.20562@dukeread09>   Larry Kilgallen wrote:F > That surprises me, seeming to indicate that an "8086" is faster than( > all but the highest performance Alpha.  = Pentium D 950 (3.4 GHz) which is what you 14 year old son may : be using in his home PC have a SPECint2000 rate around 38.  2 An EV68 1000 MHz have a SPECint2000 rate around 7.  , You will need a GS80 to get more CPU integer power.   Absurd. But that is how it is.   Arne  0 PS: Whether SPECint2000 rate is a good metric or      not depends, but ...    ------------------------------  % Date: Sat, 10 Jun 2006 17:42:32 -0400 . From: JF Mezei <jfmezei.spamnot@vaxination.ca>0 Subject: Re: Cost of used alphas vs 8086+charron, Message-ID: <448B3CC8.F3D1F978@teksavvy.com>   "Main, Kerry" wrote:F > Mmmm... Don't forget that on top of normal VMS licensing, the CharonI > license is definitely not free and is a big expense to be considered ..   @ Correct. I think that in the case of VAX, those who pay hardwareH maintenance for their VAX at exhorbitant price for museum hardware mightD find that running VMS in a industry standard server with charron VAXR emulator would end up costing roughly the same or less (with greater performance.)  D Obviously, alpha maintenance costs will rise so perhaps at one pointG soon, there may come a time when the charron solution may come in handy ' to continue to run VMS on new hardware.   G It is a shame though because VAX is the one with the software (and Qbus B hardware) that ties customers to VMS, but Alpha is the one that isC developped and with access to new hardware (PCI etc). So you cannot  integrate both.   E In hindsight, Digital should have provided more Q-BUS support for its ! alphas to get people to migrate.    @ In terms of porting VMS to the 8086, of course VMS engineers andE management will deny it. Until HP corporate says so, nobody within HP  can discuss such a topic.   G But once Intel and HP annouce the end of line for that IA64 thing, lets G just hope that first boot of VMS on the 8086 will come within weeks and  not months ;-)  G The danger here is that if VMS management do NOT tell HP Corporate that F customers are interested in a native port of VMS to the 8086 platform,H then HP corporate may let VMS go down the drain when the end of line for IA64 is announced.  A So while I expect VMS engineers to publically deny the concept of G porting VMS to the 8086, I expect these very same people to be fighting I for it internally. If there is no push internally, then we have to worry.     E The fact that SRI is so succesful is an indication that HP has a huge  untapped market.   ------------------------------    Date: 10 Jun 2006 17:09:50 -0500- From: Kilgallen@SpamCop.net (Larry Kilgallen) 0 Subject: Re: Cost of used alphas vs 8086+charron3 Message-ID: <B60A$AFPdoqE@eisner.encompasserve.org>   ] In article <448B3CC8.F3D1F978@teksavvy.com>, JF Mezei <jfmezei.spamnot@vaxination.ca> writes:   G > In hindsight, Digital should have provided more Q-BUS support for its # > alphas to get people to migrate.     Why ?   ? The Logical Company provided that, and I used it several times. A Is the goal to drive away all companies providing add-on products 	 for VMS ?    ------------------------------  # Date: Sun, 11 Jun 2006 05:23:06 GMT , From: Jeff Shirley <spam-fan@mindspring.com>5 Subject: FA: ULTRIX-11 V2 Manual Set and Install Tape B Message-ID: <_MNig.4346$lf4.1128@newsread1.news.pas.earthlink.net>  
 Greetings.  J I just listed an ULTRIX-11 V2 manual set and install tape on Ebay, in case9 anyone is interested.  See the URL and description below.    Thanks,    Jeff. 
 ----------= http://cgi.ebay.com/ws/eBayISAPI.dll?ViewItem&item=8826822043   M This auction is for the manuals and 9-track install tape for the Ultrix-11 V2 J operating system for Digital Equipment Corporation (DEC) PDP-11 computers. The tape label reads:    BB-X302B-BC  ULTRIX-11 V2 16USR BIN 16MT9 1984  N Three of the binders comprise the Programmer's Manual set, labeled as follows:  	 ULTRIX-11  Programmer's Manual  Volume 1, 2A, and 2B  N The fourth binder contains the System Management and Operation Manuals, and is labeled as follows: 	 ULTRIX-11 ' System Management and Operation Manuals 
 Release Notes  Installation Guide Software Product Description Software Technical Description System Management Guide   L One binder has a split that has been sewn back together, and the manuals areM in otherwise excellent condition.  I cannot guarantee that the tape can still L be read due to its age, so please bid accordingly.  No software licenses are	 included.   J This item can be shipped at the buyer's expense within the continental US.H Weight would be around 25 pounds, and it would ship from ZIP code 91773. Thanks for looking.  --   Jeff Shirley spam-ahoy@mindspring.comO "Bill Gates is filthy rich, but that doesn't mean I want to be married to him."    ------------------------------    Date: 10 Jun 2006 15:26:07 -0700) From: "Sue" <susan_skonetski@hotmail.com> % Subject: Re: My Boot Camp trip report C Message-ID: <1149978367.176109.275630@c74g2000cwc.googlegroups.com>   C I have no intention of changing the price but to tell the truth the ; price changes depends on how much things go up and how many E registrations actually get paid and we are looking good for this year  since we changed our policy.   Sue    Ian Miller wrote: 5 > This year the cost was $1495. Next year, who knows.    ------------------------------    Date: 10 Jun 2006 15:32:46 -0700) From: "Sue" <susan_skonetski@hotmail.com> , Subject: Re: New HP web based email package!C Message-ID: <1149978766.242201.273120@y43g2000cwc.googlegroups.com>   G Bob I would be interested in your feedback on trysecuresverer I have an F account so send me mail. %%%mamavms@trysecureserver.com take out the %F signs in case you are wondering I did not pick the username but I like it.    Anyone else as well.   sue      bob@instantwhip.com wrote: > https://trysecureserver.com/   ------------------------------  % Date: Sat, 10 Jun 2006 18:59:48 -0400 3 From: "Richard B. Gilbert" <rgilbert88@comcast.net> & Subject: Re: Results of my straw poll.: Message-ID: <QvqdndFbIYp40xbZnZ2dnUVZ_q2dnZ2d@comcast.com>   Michael Kraemer wrote:   > Tom Linden schrieb:  >  >> Michael,  >>! >> So why did GSI give up on VMS?  >> >> Tom >> > % > well, it isn't completely dead yet, 4 > particle accelerator controls still run under VMS.< > However, this isn't due to "quality" but rather due to theH > complexity of that particular application grown over 20 or more years,= > but the guys in charge are bailing out as fast as they can.  <snip> > and:< > "Traditionaly, VMS was famous for compatibility. It is our< > experience that this is no longer true. We find that every; > new release of anything (e.g. Fortran compiler, C, socket ; > libraries, VMS itself) breaks something. We spend alot of B > time coping with this and it is getting increasingly difficult." > (as of Oct-1994).   I I find the above a little puzzling.  It has been my experience over many  C years that applications that stick to the documented and supported  D interfaces work without problems on newer versions of VMS.  Can you F provide examples of things that broke while using only the documented  and supported interfaces?    ------------------------------   End of INFO-VAX 2006.322 ************************