home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.sysv386
- Path: sparky!uunet!stanford.edu!leland.Stanford.EDU!fangchin
- From: fangchin@leland.Stanford.EDU (Chin Fang)
- Subject: Re: What is the fastest 80x86 SVR4 X window hardware?
- Message-ID: <1992Sep9.193419.26104@leland.Stanford.EDU>
- Sender: news@leland.Stanford.EDU (Mr News)
- Organization: DSG, Stanford University, CA 94305, USA
- References: <1992Sep8.162409.19441@tropix.uucp> <BuB644.GCw@gator.rn.com> <1992Sep9.153201.18582@cbnewsj.cb.att.com>
- Date: Wed, 9 Sep 92 19:34:19 GMT
- Lines: 77
-
- In article <1992Sep9.153201.18582@cbnewsj.cb.att.com> dwex@cbnewsj.cb.att.com (david.e.wexelblat) writes:
- >In article <BuB644.GCw@gator.rn.com> larry@gator.rn.com (Larry Snyder) writes:
- >> jdc@tropix.uucp (Jim Cronin) writes:
- >>
- >> >As the subject line says, what is the "fastest" X-windows display
- >> >for a 80386 or 80486 running SVR4 Unix? Are there particular
- >> >chipsets (ET4000 or 8414 for example) that are really great or
- >> >really bad? Are there any boards out there that are much better
- >> >or much worse? Are there any that stand out at low resolution
- >> >but slow down at higher resolutions? And how about dual-ported
- >> >video ram?
-
- Depending whether you want free/(with src) or want commerical implementation.
- As David pointed out below. Go with VRAM helps quite much in many instances.
- ET4000 is a dumb frame buffer, but if you can find local bus card using
- this chipset, it should be sufficient for many daily work.
-
- >> How about the ATI ultra? With Dell SVR4 something like 86,000
- >> xstones is delivered using their X11R5 making it faster than
- >> many sparcs..
-
- I am running Dell's SVR4 issue 2.2 on a 20Mhz 386 with 8 megs right
- now. at work. A few days ago, I put in an early model ATI 8514
- Graphics Ultra with MS compatible 3button bus mouse into this box just
- for curiosity. Well, the results were pleasing, but due to the fact
- that this box has only 8 MEGs, and the default Dell kernel is fat
- (~1.8 Meg, I haven't got around to trim it yet, my work load these
- days is too heavy :( so disk swapping caused a lot slow down. But the
- potential of the ATI board is really obvious.
-
- BTW, the early model of ATI Graphics Ultra (with 1MEG VRAM) only offers
- 75Mhz as the highest dot clock. This is really confining. As dictated
- by the basics of video timing, if one wants 70Hz or greater refresh rate
- then only 0.7 Meg pixel is possible (eg. 1024x768). More on this below.
-
- >A big piece of the puzzle is "do you want free X, with source?" If this
- >is a concern for you, then ET4000 is the best you are going to do. An
- >Orchid ProDesigner IIs in my 486/33 box is currently doing 11.6k xStones
- >with XFree86 1.1 (beta - expected release at the end of the month). We
- >are hoping to eventually squeeze a bit more out, but that's the about the
- >best you're going to do.
-
- David, have you got a local bus ET4000 card for testing yet? I imagine
- the higher bandwidth and higher clock should help. I don't have one thou.
-
- [the rest deleted]
-
- In general, when ask X graphics performance, the desired resolution should
- be made clear. Various graphics hardware combo can support different
- resolution at different performace levels. Here are some simple rule
- of thumb myself use to evaluate hardware quality for high-res work (i.e.
- at least one Meg pixels)
-
- Assuming you want X% of Meg pixels as your resolution (e.g.
- 1024x768 ~ 70% meg pixels, 1152x900 ~ 90% meg pixels, 1120x830 ~ 80% meg
- pixels etc.) and assuming that you always want 70 Hz vertical refresh
- rate or higher for non-flickering display, then the following rules
- should hold for most instances (only variable frequency monitors are
- considered)
-
- (1) the highest dot clock available from your video adapter should
- be X Mhz or higher, (e.g. if X = 70, then 70 Mhz should be available),
- (2) the upper limit of your monitor's horizontal sync frequency should
- be at least the above dot clock /(0.8 x your desired horizontal resolution)
-
- I think if you go by these two rules in selecting your video subsystem
- components, with a sufficiently generous budget for a coprocessor
- assisted board (for commerical implementations, of course), then you
- should be fine in most instances.
-
- Regards,
-
- Chin Fang
- fangchin@leland.stanford.edu
- ps. Please don't reply to my Stanford account, as I have graduated in
- June and am now working in Los Angeles as a Member Technical Staff again
-
-