Linked by Thom Holwerda on Wed 28th May 2008 19:09 UTC
Graphics, User Interfaces Yesterday, during the opening hours of the D6 conference, Walt Mossberg and Kara Swisher jointly interviewed Steve Ballmer and Bill Gates. While the interview dealt mostly with the past, Yahoo, and a bit of Vista, by far the most interesting part was the first ever public appearance of Vista's successor: Windows 7. Earlier today, the team behind D6 posted a video of the demonstration, which was conducted by Microsoft's Julie Larson-Green. From a graphical user interface point of view, there were some interesting things in there.
Thread beginning with comment 316017
To view parent comment, click here.
To read all comments associated with this story, please click here.
orestes
Member since:
2005-07-06

Current systems, as I understand them capture what's being displayed and apply varying degrees of compression to it before sending the output across the network. Theoretically, with a 3D accelerated desktop you could just send the instructions for drawing the primitives the desktop is composed of across the network and have the GPU at the other side render them freeing up processing power on the server side and allowing for higher quality and a richer experience on the client side.

Reply Parent Score: 3

WereCatf Member since:
2006-02-15

Current systems, as I understand them capture what's being displayed and apply varying degrees of compression to it before sending the output across the network. Theoretically, with a 3D accelerated desktop you could just send the instructions for drawing the primitives the desktop is composed of across the network and have the GPU at the other side render them freeing up processing power on the server side and allowing for higher quality and a richer experience on the client side.

There's quite a few catches here actually: the whole windowing subsystem would have to support network transparency all the way from the lowest graphics functions. X does support network transparency so that would be possible with X, but it just hasn't been done. With Windows it is not possible without rewriting the whole thing. There's also a few other caveats: both ends would have to have exactly same fonts or the end result wouldn't look the same, also you'd still have to send all pictures like icons and web page elements over the network.

Reply Parent Score: 0

Jack Burton Member since:
2005-07-06

Current systems, as I understand them capture what's being displayed and apply varying degrees of compression to it before sending the output across the network. Theoretically, with a 3D accelerated desktop you could just send the instructions for drawing the primitives the desktop is composed of across the network and have the GPU at the other side render them freeing up processing power on the server side and allowing for higher quality and a richer experience on the client side.


Not every remote desktop implementation works by capturing a bitmap and sending it across the network. VNC works like this, but RDP or ICA or even X work differently.
They do exactly what you just described: they send graphics command across the network and the client interprets them, by rendering what the server sent.

Reply Parent Score: 1

-oblio- Member since:
2008-05-27

As far as I know, X already does this (sending drawing primitives across the network). On the other hand, its network performance isn't stellar, so there may be problems with either the protocol or the implementation (hence the need for NX or similar).

RDP also has an extension for this: http://msdn.microsoft.com/en-us/library/cc239611.aspx
"For example, instead of sending the bitmap image of a filled rectangle from server to client, an order to render a rectangle at coordinate (X, Y) with a given width, height, and fill color is sent to the client. The client then executes the drawing order to produce the intended graphics result."

The point of sending bitmaps is that you waste bandwidth, but you save CPU. Which can be good for thin clients. I don't think it's a good trade off, but hey, I didn't create these protocols (VNC, RDP without this extension) ;)

Reply Parent Score: 2

sbergman27 Member since:
2005-07-24

As far as I know, X already does this (sending drawing primitives across the network). On the other hand, its network performance isn't stellar, so there may be problems with either the protocol or the implementation (hence the need for NX or similar).

It all depends upon the network's characteristics. X performance on a LAN is quite good. A bit of light compression might help there, but it's debatable whether it would help or hurt. X is, however, a very "chatty" protocol. The serialized round trips are excessive. (Run X over a ppp connection and watch the modem lights.) Latency kills. Even just 10 or 20 ms. The neat thing about NX is that it *is* X, and not one of those framebuffer transmission kludges like VNC. Fast as lightning... and still X. One day, it will be an official extension of X and I can't wait.

Edited 2008-05-29 18:38 UTC

Reply Parent Score: 2