info-vax@ucbvax.ARPA (03/14/85)
From: zar%xhmeia%hamlet.arpa@lbl.arpa Can anyone give me a reason why I shouldn't set my WSMAX sysgen parameter and all user's WSEXTENT quotas to the size of physical memory (excluding system reserved memory). It seems to me that this should give a user all physical memory if he has a program that needs it. The only drawback is the increase of reserved system memory to account for raising the WSMAX parameter (does anyone know how much this would be for a 4Meg mem system?). Thanks in advance... From The Great Zar Daniel Zirin Caltech Chem Dept (818) 356-6011
walton%Deimos@CIT-HAMLET.ARPA (09/07/85)
Here at Caltech astronomy we have two 780's, each equipped with 4Mb of memory. One of them is sufficiently old that it has 16K chips on its memory board, while the other has 64K chips. My impression is that prices for VAX memory are quite ridiculous. The least expensive upgrade to 8 Mb for both machines which anyone here has come up with is $24,000 or so, which includes a new controller to allow 64K chips to be used in the older machine and 12 Mb of memory boards with 64 Kb chips. This is $2000 per megabyte, which is something like 10 times what computer memory costs in the IBM PC world. Why?? Is it simply because DEC's prices are so high that their competitors can charge high prices while still underselling DEC? I realize that VAX memory has to be error correcting and not just error checking, and that it has to be faster than a PC's memory, but still... (PDP-11 memory seems to be somewhat higher priced, as well. I have seen ads for 4 Mb on a single Q-bus board priced at $4000.) A releted question is whether or not we have missed a trick--is there a cheaper way to do this? In particular, are there VAX memories made up of 256K bit chips? What's the cheapest memory anyone out there has purchased for a 780? Send to me directly, and I'll summarize to the net. Steve Walton Caltech Solar Astronomy BITNET: walton@citdeimo ARPAnet: swalton%deimos@cit-hamlet
W._Michael_Terenyi%QZCOM.MAILNET@MIT-MULTICS.ARPA (09/12/85)
VAX memories manufactured by National Semiconductor are much chaeper than equivalent DEC memories. Regards, Michael
OC.GARLAND%CU20B@COLUMBIA.EDU (Richard Garland) (09/19/85)
We have used National Semiconductor memory for years and had good success. They have socketed chips with extras on each board. In about 10 Megabyte-years I've had to swap maybe 3 chips (there is a whole book on how to do it). The memory is not on service since the cost of service for the warantee period (2 years) would buy a new board. Furthermore changing chips is about the only service I imagine will ever be needed and I do that fine. So consider the service cost "free". (they promise 24 turn-around on boards returned under warantee.) Other managers whom I asked before buying this stuff all had found a different supplier (about 5 vendors were mentioned) and all were happy. Some have warantee periods of 5 years. It looks like memory is one area where youcan't go far wrong. By the way, another site at Columbia just did a 4 Meg/old controller for 8 Meg /new controller deal with EMC and they are happy with cost and performance. Local DEC will service them also (DEC Field service largesse extends at least from NYU to Columbia). (If I were they, I would put the controller on contract and the memory off contract as per above price vs. warantee comparison). Rg PS -- How about a survey on the wonderful world of microVAX-II memory prices. Now that's an exciting market (for users, pity the vendors). Rg -------
PMANERA%NYBVX1.BITNET@WISCVM.ARPA (09/19/85)
We upgraded our 780 from 4 megs of the old C-controller memory to 8 megs of E-controller memory using all EMC hardware, i.e. memory, controller, backplane, and power supply (they swap parts in the DEC memory power-supply box). DEC field service was on-site for the swap, but not "officially", if you will. DEC does have an on-going relationship with EMC - both in and out of court. Here in New York, field service WILL maintain EMC boards. But given EMC's responsivness, we cut a special labor-only contract against the controller/backplane and "first board" of EMC memory. This was just to avoid any possible finger-pointing should there be problems. I don't know that this is any kind of policy, but DEC has made it one here in New York, at our Center, if no where else. As to cost, EMC's was very competitive. We had purchased other memory from Trendata (Standard Memories) with some success also, but EMC's prices were better. I've had one problem with one board since installation. I started logging soft errors, called EMC, they sent a new board out over night, I swapped it in the next day, and that was that. I recommend it. Peter Manera NYU - (212) 285-6184
walton%Deimos@CIT-HAMLET.ARPA (09/26/85)
Well, I have input on the subject of VAX memory. Basically, it's expensive, but I didn't realize how much we needed to buy. To review and correct: we would like to upgrade two 780's from 4 to 8 Mb by buying a new backplane, controller, power supply, and an 8 Mb board for the one and move the old-style 4 Mb board to the expansion chassis which is connected to the other machine. Most of the $24,000 cost for this is for the new backplane, controller, and power supply required. So, the cost of VMS memory is probably "only" 5 times as many $/bit as in the micro world. I received no messages on the whys of this disparity, except for speculation that DEC's prices for memory are so high that other vendors can undercut DEC and still charge alot and make a tidy profit. Caltech Solar Astronomy BITNET: swalton@citdeimo ARPAnet: walton%deimos@cit-hamlet