home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.os.vms
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!cs.utexas.edu!sun-barr!decwrl!deccrl!news.crl.dec.com!news!nntpd.lkg.dec.com!tallis.enet.dec.com!kirk
- From: kirk@tallis.enet.dec.com (Matthew B. Kirk)
- Subject: Re: Plans for mixed Alpha/VAX cluster with common system disk?
- Message-ID: <1992Nov20.154517.627@nntpd.lkg.dec.com>
- Sender: usenet@nntpd.lkg.dec.com (USENET News System)
- Organization: Digital Equipment Corporation
- Date: Fri, 20 Nov 1992 15:29:41 GMT
- Lines: 71
-
-
- In article <1992Nov19.125053.5849@ais.com>, bruce@ais.com (Bruce C. Wright) writes...
- >In article <1992Nov19.125756.7901@slcs.slb.com>, brydon@asl.slb.com (Harvey Brydon (918)250-4312) writes:
- >> In article <1992Nov13.100832.3777@dxcern.cern.ch>, julian@vscn08.cern.ch
- >> (Julian James Bunn) writes:
- >>>What worries me more about a mixed architecture VAX/Alpha VMS Cluster is
- >>>how the poor innocent users are going to sort out the confusion between
- >>>VAX and Alpha executables. [...]
-
- We've been running a test system as a mixed cluster in my group (my
- primary workstation, a 3000 model 500, is a member of it). We haven't
- had problems with 'gee, is that a OpenVMS VAX executable or an OpenVMS AXP
- executable'. We do have separate system disks though, which is
- necessitated by the fact that we frequently have each OpenVMS AXP node in the
- cluster running a different test version of OpenVMS.
-
- >
- >This is one of those situations where you can't win. If the default
- >extension for executable images were changed, then many of the programs
- >that spawn subprocesses or do dynamic mapping to sharable images are
- >likely to break when they're recompiled for Alpha. (For the most part
- >you don't need to go through your programs to look for things like
- >that when you move them to Alpha -- though you may want to do so for
- >other reasons, such as improving efficiency on the new architecture).
- >
-
- Exactly correct. We run into a similar problem with translated images
- that do something with another image (e.g. vfork) and specify the image
- name as 'FOO.EXE'. If FOO.EXE is translated (and hence renamed FOO_TV.EXE),
- then the user must do one of two things:
-
- 1. rename FOO_TV.EXE to be FOO.EXE. This runs the risk of confusing
- translated and native images (which are different in some cases,
- even if they would otherwise share the same name).
-
- 2. modify the source to simply specify 'FOO', then $DEFINE FOO FOO_TV
- on the OpenVMS AXP system.
-
- You would run into the same problem if .EXEs on OpenVMS AXP were renamed
- .AXEs, or .AEXEs (both were suggested at one point).
-
- >The idea one person had, about embedding both types of executable
- >code in one image, also has similar problems but probably a lot less
- >code depends on the image format than on the file names. If any code
- >does depend on the image format it will break for Alpha anyway.
-
- The problem here is that the images get much larger, so more disk space
- is required. Also, image formats are different.
-
- >What I don't understand is why the image activator in Alpha doesn't
- >crank up VEST in interpretive mode in such cases (or possibly only if
- >a flag is set in the image header, like the debug flag). It seems to
- >me that this would get you very much of the effect you'd want for
- >very little penalty (other than the interpretive overhead, of course,
- >which is substantial but still faster than most VAXen).
-
- VEST is a static translator. The Translated Image Environment (TIE)
- contains a fall-back interpreter but it's not particularly fast.
- One of the reasons we went to great effort to develop a translator
- instead of an interpreter (which would have been far simpler) is that
- interpretation would never have given acceptable performance. There
- are too many things that the VAX does (like maintain condition codes
- all the time) that couldn't be optimized out in interpretation. Translation
- allows us to do all sorts of optimization.
-
- =============================================================================
- Matthew Kirk kirk@tallis.enet.dec.com
-
- All opinions expressed here are mine and do not reflect those of Digital
- Equipment Corporation.
- =============================================================================
-