I love USB displays!
|
Prev: pipes/streams/channels for parallelism (was: Is C close ...)
Next: Intel Investing Millions for Supercomputer Research in Europe
From: Terje Mathisen on 25 Nov 2009 01:37 Andy "Krazy" Glew wrote: > I love displays! I love looking at large numbers of pixels, relatively > large pixels for my aging eyes. I've been using a 1920x1200 + 1600x1200 combination both at home and at work for a few years now, these are both connected to a little docking slice. For our trading department I tested out a PC-Card graphics adapter some years ago, using this allowed another pair of relatively fast 2D screens. Since PC-Card slots have gone away, it makes sense to move the power-hungry parts to the other end of a sufficiently fast USB2 connection. I suppose you need external power for these, or do they run on 500 mA? > comp.arch relevance: what sort of computers are good for processing such > large displays/ Not necessarily GPUs, since not necessarily real time > graphics. External screens needs external power anyway, so why not embed the needed USB/display hardware in the screen itself? I'm pretty sure I've seen such beasts, at least in the form of projectors supporting both VGA and USB connections. Having the driver available is the crucial problem for a meeting room projector however. Terje -- - <Terje.Mathisen at tmsw.no> "almost all programming can be viewed as an exercise in caching"
From: "Andy "Krazy" Glew" on 25 Nov 2009 16:53 Terje Mathisen wrote: > Andy "Krazy" Glew wrote: >> I love displays! I love looking at large numbers of pixels, relatively >> large pixels for my aging eyes. > Since PC-Card slots have gone away, it makes sense to move the > power-hungry parts to the other end of a sufficiently fast USB2 connection. > > I suppose you need external power for these, or do they run on 500 mA? The USB display adapters have no external power. They must be USB powered, unless they can snarf power from DVI or VGA. 2 of my displays have built in USB hubs. I.e. the display is powering it's own USB display adapter.
From: ChrisQ on 27 Nov 2009 09:05 Andy "Krazy" Glew wrote: >> >> I suppose you need external power for these, or do they run on 500 mA? > > The USB display adapters have no external power. > > They must be USB powered, unless they can snarf power from DVI or VGA. > > 2 of my displays have built in USB hubs. I.e. the display is powering > it's own USB display adapter. > I replaced my old tube monitor for a large tft earlier this year, 1600 x 1200 vs 1280x1024 on the tube and a fraction of the power. Order of magnitude clearer as well and less eye strain when using it all day. I wouldn't have thought that usb would have enough bandwidth for video rates, though perhaps if you're not doing anything else with the usb controller. NIf no psu, so how bright is it ?. Most of the power for tft's is in the backlight, often 10-20 watts on a larger display... Regards, Chris
From: Peter Grandi on 29 Nov 2009 08:52 > I love displays! I love looking at large numbers of pixels, > relatively large pixels for my aging eyes. Actually it should be relatively small pixels with scalable fonts and icons -- they give much better contrast and are easier for the eyes to focus on than coarse pixels masked by antialiasing. [ ... ] > Best of all, I can almost, *almost*, act as if my love of large > display surfaces is work related. It sure does help to be able > to look at really, really, wide spreadsheets (although really, > really, wide spreadsheets are a bit of an abomination). For programming it is much worse than abomination. I have noticed that many people inexplicably (to me) like to have a single maximized window on the screen, and that many programmers tend to write lines as long as their window (and I guess many know that kind of programmer who also likes really tiny fonts with dark colors on black backgrounds). The result I see is that lines in programs become longer and longer. The extreme case I have seen so far is that coworker who used a tiny font on a large monitor, with a maximized editor window 400 columns wide (and 200 high). I asked him what was the point, and he said that with very long lines he would write many C functions entirely on one line, and he could often write a whole module that would fit in one screenful too. Had to spend several days reformatting and indenting his code. [ ... ] > (After asking IT, who said that I could only have two monitors > if they were smaller, 1400x1050. Which rather misses the point.) I see that Mordac style characters are persecuting you. Where I work currently some people have 3x 1920x1200 monitors on their desks. That's pyshing it a bit too. [ ... ] > I decided to drive my two 1900x1200 monitors from Hillsboro to > Bellevue, carefully wrapped in sleeping bags and clothes. So > now, on my big Biomorph desk at work (another piece of personal > equipment) I have 5 monitors: Another case of edging closer to being a "free agent" supplying your own tools/machinery for work. > two 1900x1200 in landscape mode, and the two 1050x1680, in > portrait mode. Plus the laptop LCD display. The 1920x1200 would not be really necessary if native-portrait monitors were available to. Ideally they would be grayscale too. Unfortunately most monitors are targeted to consumers who can only think of playing movies on them, and want them as wide and colorful as possible; for office/programming work high DPI portrait greyscale monitors would be far better (speaking from experience). [ ... ] > Let's see, that's 8.88 megapixels, if I have done my math > correctly. Most of it driven by USB. Probably no good for > video or games, but good enough to throw a lot of data up where > I can look at it. More! I want more! More slow pixels! If I > could plug in e-paper displays all about my office, I would. My observation is that the most precious computer entities are displays and memory (that is, visual and program memory) as proven by the degree of multiplexing/caching they are subjected to. Right now I have a 24" LCD, with the following levels of multiplexing: * KVM to switch between 2 computers. * Window manager with multiple virtual desktops. * Multiple overlapping windows within a desktop. * Multiple tabs within a window. * Multiple buffers (Emacs) within a tab. It can get pretty confusing and distracting. Never mind the crazy performance implications of many layer of memory caching each with their own inappropriate replacement policy. > We're on the verge of LCDs and e-paper being cheap enough to > replace the whiteboards that are ubiquitous in offices. I am very very much against replacing any tech that works well, requires no power or batteries, no cabling, little maintenance, and has excellent viewing properties, with something else, just for the sake of bringing a "Blade Runner" style world forward. [ ... ] > Eventually, we must get rid of refresh. LCDs don't refresh... > comp.arch relevance: what sort of computers are good for > processing such large displays/ Not necessarily GPUs, since not > necessarily real time graphics. Well, 'comp.arch' is about computer system architecture in general, not just processors, and the technology of the surface between the human an computer perceptual worlds is part of that. There is an angle on processors: current monitors are in effect display computers, as the signals received from the main unit are processes (e.g. zooming, sharpening, ...) before being rendered to the LCD. In effect what one sees on the monitor is a processed movie of the contents of the frame buffer (that's why some LCD monitors offer a sharpening setting even for DVI input), and that is also what those USB adapters you use do too, as do the various KVM-over-Ethernet-or-IP products. Not too different from current disk drives, which are in effect complicated block device servers. The overall architectural relevance here is that modern systems are ever more asymmetrical distributed system with specialized embedded processors (and not just printers and disks, monitors too, never mind disk host adapters, network cards, ...). The result is flexibility, but also increased confusion as the definition of "working" becomes rather fuzzy for even simple systems. I wonder mhow many of the readers of this newsgroup realize that sometimes current TVs crash and have to be rebooted, and so do current LCD displays (and trains -- once I was stuck on a train for a couple of hours as the locomotive engineers had difficulty rebooting to a stable state the engine's controllers).
From: Bernd Paysan on 29 Nov 2009 12:57
Peter Grandi wrote: > The 1920x1200 would not be really necessary if native-portrait > monitors were available to. Ideally they would be grayscale too. > > Unfortunately most monitors are targeted to consumers who can > only think of playing movies on them, and want them as wide and > colorful as possible; for office/programming work high DPI > portrait greyscale monitors would be far better (speaking from > experience). High resolution: yes, please. Grayscale: no. I use syntax highlighting, and even when composing a Usenet posting like this, my editor colors quoted text differently - that's very useful, I would miss it on a grayscale monitor. IMHO, a number of things go wrong which is why advance of resolution is slow. My take at this is here: http://www.jwdt.com/~paysan/hires.html Executive summary: Use bayer pattern for the screen, double effective resolution without decreasing feature size (i.e. with the same technology and yield). > I wonder mhow many of the readers of this newsgroup realize that > sometimes current TVs crash and have to be rebooted, and so do > current LCD displays Fortunately, I'd neither to take my TV nor my LCD to an unexpected forced reboot yet. The complexity still remains low enough to keep the software robust. Suggestions like mine above would require some slightly more complex software in the screen as long as the interface stays RGB. -- Bernd Paysan "If you want it done right, you have to do it yourself" http://www.jwdt.com/~paysan/ |