nick@uunet.uu.net (Nicholas Jacobs) (12/15/89)
We are currently evaluating workstations, one of which is a Sparcstation 1. Eventually we may want to run X windows on the machine and we are concerned about a suitable configuration of hardware. One of the possible configurations is a machine with 8 Mbytes and a single disk (~100 Mbytes) for local swap and personal files for the user of the workstation. Is this sufficient? Obviously, the more memory the better but we would like to keep initial costs down. Please email me any responses. Thanks in advance, Nicholas Jacobs
jim@kanga.lcs.mit.edu (Jim Fulton) (01/03/90)
|We are currently evaluating workstations, one of which is a Sparcstation |1. Eventually we may want to run X windows on the machine and we are |concerned about a suitable configuration of hardware. MIT's R4 server (watch comp.windows.x for announcements) runs quite nicely on a 16meg SPARCstation 1 with a vanilla cg3 8-bit framebuffer. We haven't tried it on an 8meg system, but since the server's working set is much smaller than R3, it should pretty usable.
nick@uunet.uu.net (Nicholas Jacobs) (01/09/90)
In article <3982@brazos.Rice.edu> toro!nick@uunet.uu.net (Nicholas Jacobs) writes: >X-Sun-Spots-Digest: Volume 8, Issue 227, message 12 of 18 > >We are currently evaluating workstations, one of which is a Sparcstation >1. Eventually we may want to run X windows on the machine and we are >concerned about a suitable configuration of hardware. I got several replies (about 7 or 8). The general (but not universal) feeling seems to be that 8 Mbytes of memory is the bare minimum. Most people seemed to think that 12 or 16 was better. A Sun rep. here in NYC said that the unofficial policy is that if you want run X, you need to have 16 Mbytes. If people want copies of the summary, please feel free to email me. Thanks again for your help, Nicholas Jacobs +-----------------------+----------------------------+----------------------+ | UUCP: uunet!toro!nick | Internet: nick@toro.uu.net | AT&T: (212) 236-3230 | +-----------------------+----------------------------+----------------------+
mlandau@bbn.com (Matt Landau) (01/29/90)
jim@kanga.lcs.mit.edu (Jim Fulton) writes: >MIT's R4 server (watch comp.windows.x for announcements) runs quite nicely >on a 16meg SPARCstation 1 with a vanilla cg3 8-bit framebuffer. We >haven't tried it on an 8meg system... I'm running it on an 8-meg color machine, and it's blindingly fast at most operations. (Interestingly enough, though, we've found in analyzing some of our X11 programs that doing any graphics using a fill_style of FillOpaqueStippled is *really* slow in the X11R4 compared to the Sun OpenWindows server. The MIT server doesn't seem to optimize the case where the stipple is a "good" size, like 32 x 32 -- it just uses the fully generic "fill anything with anything" code, which is slow.)
pop@sun.com (Steve Tufty) (01/30/90)
:>We are currently evaluating workstations, one of which is a Sparcstation :>1. Eventually we may want to run X windows on the machine and we are :>concerned about a suitable configuration of hardware. : :I got several replies (about 7 or 8). The general (but not universal) :feeling seems to be that 8 Mbytes of memory is the bare minimum. Most :people seemed to think that 12 or 16 was better. A Sun rep. here in NYC :said that the unofficial policy is that if you want run X, you need to :have 16 Mbytes. I was wondering what the recommended memory minimums would be for the client and the server, providing that these processes are on different machines. My guess is that the client is the cpu hog. Comments anyone? Steve Tufty (..!sun!cohesive!mercury!pop)