klamer@mi.eltn.utwente.nl (Klamer Schutte) (05/27/91)
In alt.sys.sun and comp.sys.hp i wrote the following query: > We are considering buying a new workstation. The two candidates are: > 1) Sun sparcstation 2 > 2) HP 730 > > In price they are about considerable (at least for us). In speed the HP > is the clear winner. On compatiblity the sun is the winner (we have 5 > sun sparcs here, and no HP (yet)). > > The question is: > Is the HP really the promised factor 2+ faster compared to the sun? Not only > on benchmarks, but also on "real" programs? (We do both integer and float > programs. I/O or display are less important). > and > Does fitting 1 HP (with 840 Mb disk) in a sun network give big problems? > Can we easily compile && run our programs on the HP? > Is the basic MOTIF HP interface and the OPENLOOK interface of sun no > problem? Can we use the display of the one to let the program run on the > other? > > Replies by email will be summarized. I got in total 20 queries, in which 4 asked for a forwarding of the info i got. Here is my summary of the replies. I will split the answers in three parts: speed, compatibility and general. The speed comparision. Here the answer is simple: the HP machine is faster. The only point is how much faster. Answers indicate speedup is between the promised 2+ to 10. The compatibility problem. Here are very different answers. It ranges from: } I think there will not be problems if you put HP 730 into network } composed by Suns. by Andrea Borgato <borgato@ghost.unimi.it> to this answer: } I tend to be a Sun booster so beware. However, that said, I think you } are better off with another Sun. } } I haven't run benchmarks, but I do believe HP's performance claims. } However, I also believe that you vastly underestimate the support costs of } each new vendor OS you bring in. How much time and effort have you and } your students put into understanding SunOS? I would guess between 1/2 and } two person-years. You can expect to invest a similar amount of time on } the HP (less because some of the experience is transferrable, more because } now you have to deal with compatibility problems rather than just 2 } separate systems). } } When I managed a computer science department computing facility, I } estimated an ongoing cost of between .25 and .75 FTE for each new platform } for my department, plus startup costs. I had lots of unpaid grad } students, 4 systems programmers, and a technical manager (me) supporting 6 } Unix and 4 non-Unix architectures at the time. } } If we figure a 4 year useful life on the hardware you buy today, .25 FTE } translates into 1 person-year. Can you buy an additional SPARCstation II } for the cost of 1 person-year? } } In your calculations, you should not forget the ongoing cost of hardware } and software maintenance from the vendor. I don't know what HP's rates } are. My guess is that 2 SS II machines would be slightly more expensive } than 2 HP 730s. by JQ Johnson <jqj@oregon.uoregon.edu>. The problem is nicely summarized by Michael Ames <mike@jupiter.nmt.edu>: } I would say that if your code is strictly number } crunching and fairly standard C or Fortran it should work fine. The } programs we have had problems with need to talk to various parts of } the OS or to the tty drivers. Also, as Kelvin Don Nilsen <kelvin@kickapoo.cs.iastate.edu> and Nadia Pitacco (morgaine) <modus!simm!polluce!root@rutgers.edu> indicate, the system administration procedures are different. The general section: Michaela Harlander <harmic@marvin.e17.physik.tu-muenchen.de> indicates that you might have problems with the delivery: } You should perhaps check how long it will take to get the HP } standing on your desk. During the presentation of the 700 HP series } in Munic I heard that they may be able to do it within 14 months! } I don't know if you have time to wait so long.... this was for our site } a severe argument against the HP. Kelvin Don Nilsen <kelvin@kickapoo.cs.iastate.edu> indicated a difference in memory prices: } Something to consider is that the HP requires much more expensive memory than } is available for the Sparc 2 (by a factor of 4 or so!). That's it. I think that which choice you take is dependent of the kind of software you want to run on it. When it is your own code, and you know that it is not system dependent, you could go for the HP. On the other hand, when you have included a lot of SunOs isms, or it is not your own code and not (yet) ported to the HP's, a Sun can be a better choice. I wish to thank all who responded. They are, in order of arrival of the reply, fmonaldo@aplcomm.jhuapl.edu (Monaldo Francis M. S1R x8648 ) jahonen@ujocs.joensuu.fi (Jarmo Ahonen) Michaela Harlander <harmic@marvin.e17.physik.tu-muenchen.de> mike@jupiter.nmt.edu (Michael Ames) "raymond thomas pierrehumbert" <rtp1@midway.uchicago.edu> Fred Appelman <fred@cv.ruu.nl> kelvin@kickapoo.cs.iastate.edu (Kelvin Don Nilsen) John DiMarco <jdd@db.toronto.edu> jqj@duff.uoregon.edu JQ Johnson tdycapd!oliveau@uunet.uu.net (Greg Oliveau) "(Alain Brossard EPFL-SIC/SII)" <brossard@sasun1.epfl.ch> Andrea Borgato <borgato@ghost.unimi.it> schultz@halley.serc.3m.com (John C. Schultz) Nadia Pitacco (morgaine) <modus!simm!polluce!root@rutgers.edu> sutton@lamar.colostate.edu (Richard Sutton) jimb@silvlis.com (Jim Budler) yih%atom@cs.utah.edu (Benny Yih) hunter@phoenix.pub.uu.oz.au (James Gardiner [hunter]) Jim Wright <jwright@cfht.hawaii.edu> Klamer -- Klamer Schutte Tel: +31-53-892786 Fax: +31-53-340045 Faculty of electrical engineering -- University of Twente, The Netherlands preferred: klamer@mi.eltn.utwente.nl SMTP: klamer@utelmi01.el.utwente.nl
harmic@hotblack.e17.physik.tu-muenchen.de (Michaela Harlander) (05/28/91)
In the summary on "Sun sparc 2 or HP 730" I have been cited with a statement that you may have to wait as long as 14 months for your HP. This turned out to be a bad rumor: we have now been told by our local HP salesperson that at the moment we would have to wait about 3 months. He said, however, that this may get longer within the next months. \Michaela. Michaela K. Harlander | internet: harmic@physik.tu-muenchen.de Institut fuer Theor. Physik | Techn. Univ. Muenchen | D-8046 Garching, Germany |
bill@hpuorfa.HP.COM (Bill Chidester) (05/29/91)
} You should perhaps check how long it will take to get the HP } standing on your desk. During the presentation of the 700 HP series } in Munic I heard that they may be able to do it within 14 months! Ooops! That should be 14 weeks, not months -- orders we have entered as of last Friday are showing 14 week availability. Response to the product has been extremely strong, but customer orders will always come first. It's us poor folks with internal orders who need to worry.... The product has been in production quantities and in the third party developer's hands since January and is currently shipping. } I don't know if you have time to wait so long.... this was for our site } a severe argument against the HP. If that was indeed important, please contact your sales representative and relay that message. It may well have been a slip of the tongue at the presentation. ------------------------ hpuorfa!bill Bill Chidester HP Orlando Sales Office (407) 826-9228 This does not represent an official statement of the Hewlett-Packard Company and is provided as informal information. My company never listens to me, and has been quite successful in spite of it.
jason@hpcndjdz.CND.HP.COM (Jason Zions) (05/30/91)
> we have now been told by our local HP salesperson >that at the moment we would have to wait about 3 months. >He said, however, that this may get longer within the next months. > >Michaela K. Harlander | internet: harmic@physik.tu-muenchen.de >Institut fuer Theor. Physik | >Techn. Univ. Muenchen | >D-8046 Garching, Germany | I wonder if this may have more to do with regulatory approval in Germany or export paperwork with the U.S. Government. US-based deliveries should occur sooner (I hope). Inside HP, it takes an Act of Ghod (or a personally-written note from John Young) to get a 730. -- This is not an official statement of The Hewlett-Packard Company. No warranty is expressed or implied. The information included herein is not to be construed as a committment on HP's part. The devil made me do it. This won't save me from the lawyers' wrath, but it can't hurt. Jason Zions The Hewlett-Packard Company Colorado Networks Division 3404 E. Harmony Road Mail Stop 102 Ft. Collins, CO 80525 USA jason@cnd.hp.com (303) 229-3800