madd@bu-cs.BU.EDU (Jim Frost) (09/18/88)
(This discussion doesn't really belong here so I've redirected followups to comp.unix.wizards) In article <657@mace.cc.purdue.edu> mtr@mace.cc.purdue.edu (Miek Rowan) writes: |I have a color Sun 3 in my office and it is just unacceptably slow. |Especially if you want to use X11R2, but it carries on to Sunview |also. It can get so that scrolling is no better than 1200 baud dial |up. HP's color workstation is about 100 times faster, although they |have thier own problems with networking. Hmm. I use a Sun 2/120 with x11r2 and it's not "unacceptably slow" for most things. If you put xterm into "jump" mode you get good speed on the scrolling. Of course you can't read it then but most people can't follow stuff running out at more than 4800 baud anyway; that's why "more" is there. On a Sun 3/50 x11r2 performance is much better, although still slow for really intensive graphics applications. This is mostly due to the lack of a good x11 server for the sun. A quote from the documentation of the Sun server: "The drivers are completely untuned and have inferior performance." If someone got on the ball and tuned the server you'd see much better performance, especially on the color systems (like yours) where the server is *really* untuned. I maintain that it's good enough to get work done on, though. If you want real performance I suggest using a Sun 3/50 (or better a Sun 4/110) as a smart graphics terminal to something like an Encore. Yanking all the non-graphics stuff off the Sun and putting it on a scream machine gives very nice performance and is fairly transparent under x11; even my Sun 2 runs nicely when I offload the application to a Silicon Graphics 4D. It's true that Sun's are sometimes plagued with hardware problems, but my experiences with their support were all favorable. In addition they understand that not all users are stupid and allow them to do board swapping and the like, something that I appreciate. I haven't any experience with their newer machines (we have 2 Roadrunners and a few Sun 4's coming, but nothing yet) so I can't comment on increasing or decreasing reliability. So far I've had one mono card blow up on me, a disk problem, and a couple of mono screens have flipped out, but Sun dealt with them quickly so I've no complaints. I have no idea just how long my Sun 2 has been around, but considering that it's a Sun 2 I'd say that it's been awhile. As for their software, it's a good and bad thing. I *like* SunOS, at least 3.5. Some of the networking stuff -- like the yp server -- is pretty hairy and not so reliable, but if you don't have a big network you don't need it and it runs very cleanly. NFS setup is simple and very easily maintained. Security is a problem but it ALWAYS is when the user has direct access to the hardware, not one manufacturer can say otherwise. It still takes a little ingenuity to really screw things up, to Sun's credit, and putting a password on single-user boots really blocks up some holes that exist in 3.5. In summary I don't think you've voiced a valid complaint. There is virtually nothing else that works as well as Sun workstations in their price range. 386 PC's don't have anywhere near the networking support that Sun's do, almost never have good support, and cost nearly as much. Higher-end workstations (eg Silicon Graphics) often address these problems but they're for a more specific audience and cost a lot more. jim frost madd@bu-it.bu.edu
mtr@mace.cc.purdue.edu (Miek Rowan) (09/18/88)
In article <24947@bu-cs.BU.EDU>, madd@bu-cs.BU.EDU (Jim Frost) writes: > > Hmm. I use a Sun 2/120 with x11r2 and it's not "unacceptably slow" > for most things. If you put xterm into "jump" mode you get good speed > on the scrolling. Of course you can't read it then but most people > can't follow stuff running out at more than 4800 baud anyway; that's > why "more" is there. I wasn't just speaking of X11R2, but obviously "acceptable" means two different things between us. Even with more (or less), it can take more that 3-4 seconds to paint a window of text, and not just in X11R2. Come on, you can't tell me this is ok to you? > work done on, though. If you want real performance I suggest using a > Sun 3/50 (or better a Sun 4/110) as a smart graphics terminal to > something like an Encore. Yanking all the non-graphics stuff off the > Sun and putting it on a scream machine gives very nice performance and > is fairly transparent under x11; even my Sun 2 runs nicely when I > offload the application to a Silicon Graphics 4D. Why not use an X Terminal then? Why bother with the headache of Sun's OS and all the related problems? My biggest point was the state of thier source distribution, and it carried over to other things. The only security problem is that a user could have direct access to a ethernet line which is a problem with the Sun workstations too. > > It's true that Sun's are sometimes plagued with hardware problems, but > my experiences with their support were all favorable. In addition > they understand that not all users are stupid and allow them to do > board swapping and the like, something that I appreciate. I haven't > any experience with their newer machines (we have 2 Roadrunners and a > few Sun 4's coming, but nothing yet) so I can't comment on increasing > or decreasing reliability. So far I've had one mono card blow up on > me, a disk problem, and a couple of mono screens have flipped out, but > Sun dealt with them quickly so I've no complaints. I have no > idea just how long my Sun 2 has been around, but considering that it's > a Sun 2 I'd say that it's been awhile. From talking to some engineers, the quality of the internal hardware is pretty low (I wouldn't know though), and the external hardware (mouse, keyboard..) is definatly low grade. > > As for their software, it's a good and bad thing. I *like* SunOS, at > least 3.5. Some of the networking stuff -- like the yp server -- is > pretty hairy and not so reliable, but if you don't have a big network > you don't need it and it runs very cleanly. NFS setup is simple and Considering Sun's claims about networking and their products, that is a pretty big thing to just blow off. > In summary I don't think you've voiced a valid complaint. There is I think I voiced quite a few valid complaints and you mostly avoided them with work-arounds. Gee, if you don't use this and if you fixed this the system is really nice. And, oh, don't do anything locally, let another machine do all your work. > virtually nothing else that works as well as Sun workstations in their > price range. 386 PC's don't have anywhere near the networking support > that Sun's do, almost never have good support, and cost nearly as > much. Higher-end workstations (eg Silicon Graphics) often address > these problems but they're for a more specific audience and cost a lot > more. Consider programmer-time when talking about price ranges. I think that you will find a Sun network *much* more expensive to purchase, setup and maintain than a lot of other systems availble. mtr
phil@titan.rice.edu (William LeFebvre) (09/20/88)
In article <24947@bu-cs.BU.EDU> madd@bu-it.bu.edu (Jim Frost) writes: >(This discussion doesn't really belong here so I've redirected >followups to comp.unix.wizards) > >In article <657@mace.cc.purdue.edu> mtr@mace.cc.purdue.edu (Miek Rowan) writes: >|I have a color Sun 3 in my office and it is just unacceptably slow. >|Especially if you want to use X11R2, but it carries on to Sunview >|also. It can get so that scrolling is no better than 1200 baud dial >|up. HP's color workstation is about 100 times faster, although they >|have thier own problems with networking. > >Hmm. I use a Sun 2/120 with x11r2 and it's not "unacceptably slow" >for most things. If you put xterm into "jump" mode you get good speed >on the scrolling. [ I won't do it I won't do it I won't do it I won't do it I won't do it I won't post a followup to this message I won't I won't I won't YAAAHHHHHHH! ] Oh well. This should really be discussed in Sun-Spots (was that a shameless plug or what?). If Miek Rowan has problems with Sun and Sun hardware/software, he would do well to discuss them *in detail* in sun-spots (or in comp.sys.sun). His machine is slow, by the way, because it is a color machine. The color frame buffers are really slow. He might try using just the B&W plane, but I don't know if you can coerce X into doing that. I understand that there are also a few problems with the color implementation of X11R2, so the fault might not lie strictly with the hardware. There are other things that could also be slowing down the machine: insufficient real memory for the demand (you should have at least 4 meg to run SunView acceptably), loaded network, loaded disk server, slow disk server. By the way, you can get sun-spots by either reading the moderated Usenet list comp.sys.sun or by sending in a request to subscribe (please include your Internte address) to "sun-spots-request@rice.edu". To submot an article, either post it to comp.sys.sun (it will get mailed to the right place if you are running the right version of the news software) or mail it to "sun-spots@rice.edu". Turnaround time last week was about 1 to 2 days. But the hurricane has disrupted my life and put me a few days behind. William LeFebvre Sun-Spots moderator Department of Computer Science Rice University <phil@Rice.edu>
mtr@mace.cc.purdue.edu (Miek Rowan) (09/20/88)
I know, I know....This *really* didn't start to be a big free-for-all-Sun-Bash....Geez, one inadvertant comment and whamo, you are defending yourself to the grave :-( The beauty of the net I guess.... mtr
peter@ficc.uu.net (Peter da Silva) (09/20/88)
Is this a bash where they're giving away Suns, or is Sun putting on a free bash, or is this a benefit for incarcerated workstations, or what? -- Peter da Silva `-_-' Ferranti International Controls Corporation. "Have you hugged U your wolf today?" peter@ficc.uu.net
pavlov@hscfvax.harvard.edu (G.Pavlov) (09/21/88)
In article <675@mace.cc.purdue.edu>, mtr@mace.cc.purdue.edu (Miek Rowan) writes: > > I know, I know....This *really* didn't start to be a big > free-for-all-Sun-Bash....Geez, one inadvertant comment and ^^^^^^^^^^^ > whamo, you are defending yourself to the grave :-( > The beauty of the net I guess.... > > mtr Before, the problem was all the code you "cleaned up", now its the fault of the net. A simple apology or withdrawal of statement a week ago would have quieted things down much sooner....
mtr@mace.cc.purdue.edu (Miek Rowan) (09/21/88)
In article <624@hscfvax.harvard.edu>, pavlov@hscfvax.harvard.edu (G.Pavlov) writes: > > Before, the problem was all the code you "cleaned up", now its the fault > of the net. A simple apology or withdrawal of statement a week ago would > have quieted things down much sooner.... The problem still exists, and I never blamed anything on the net. The **only** thing that I would appologize for is not following up to sun-spots when the discussion took off. Why would I want to withdraw my statements? They are all accurate ;-) mtr