On Thu, 28 May 2009 14:04:34 +0100 "Daniel P. Berrange" <d...@berrange.com> wrote:
> On Thu, May 28, 2009 at 01:54:20PM +0200, Pierre Ossman wrote: > > > > I belive the problem is that WMVi is fundamentally broken. RFB is all > > about putting complexity at the server and letting the client use > > whatever formats it wants. As such, WMVi will always be a pain to > > handle for clients, > > > > What should be done is to have an extension where the server informs > > the client of the server pixfmt, not shoves it down the client's > > throat. A client can then decide on its own if it wants/can follow this > > change. > > IIUC, forcing it on the client is an explicit goal of this extension. > Historically the VNC server associated with a virtual machine has had > to maintain a duplicate copy of the video RAM, track changes between > the guest and what the VNC client has, and do pixel format conversions. > This imposes both a memory and CPU overhead on the host, which becomes > non-negligble with lots of guest machines running on a single host. This overhead exists if you have a classical terminal server based on VNC as well, so I think this is a problem that can be managed. Also, virtualisation already has a lot of overhead in a lot of areas. > > With the WMVi extension enabled, the VNC client is set to always have > its framebuffer in exactly the same format as the guest OS has. In > combination with elimination of the video RAM duplicate, the VNC server > now avoids the memory & CPU overheads, pushing these conversion out to > each individual client. > How do these server deal with clients that do not support the WMVi extension? Do they disconnect them upon seeing SetPixelFormat? > NB, this is assuming you have a low latency, high bandwidth link between > client & server, but that's not totally unreasonable given the goal of > this extension is to push resource utilization off the server to the > client - which is sort of an opposite goal to traditional VNC desktop > scenarios. As such it wouldn't make sense to recommend use of WMVi > extension for most VNC servers, where the more common goal is to keep > the overhead/complexity in the server. Probably not. But EDS is desirable on those servers if they want to export multi-head capabilities, so the combination of EDS and WMVi is not unreasonable. Rgds -- Pierre Ossman OpenSource-based Thin Client Technology System Developer Telephone: +46-13-21 46 00 Cendio AB Web: http://www.cendio.com
signature.asc
Description: PGP signature
------------------------------------------------------------------------------ Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT is a gathering of tech-side developers & brand creativity professionals. Meet the minds behind Google Creative Lab, Visual Complexity, Processing, & iPhoneDevCamp as they present alongside digital heavyweights like Barbarian Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com
_______________________________________________ tigervnc-rfbproto mailing list tigervnc-rfbproto@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/tigervnc-rfbproto
