On Wed, Apr 10, 2002 at 06:23:32AM -0700, Sam King wrote: > I ran some simple benchmarks measuring the performance > of the Plex86 I/O subsystem, and there is room for > improvement. By moving part of the logic for devices > to the vmm, we can minimize the number of world > switches needed for device emulation, while still > using the host os for all I/O. Although this will > mean loosing bochs compatibility, it could be a major > performance improvement.
Well, having compatibility with bochs is also a bit important I think, because they have a lot of developers. Maybe we should try and coorperate to design an interface to allow both project to share code without losing too much performance. > Also, what is the state of the binary re-writing. I > did some measurements and looked through the source, > it appears that ALL instructions are being emulated. > Is this true? I tried running the dt version, and it > was even slower than the 100% emulation vm. Hmm, I don't think so, but if it is we know where the performance problem is. I don't think the rewrite was actually finished. Jeroen Dekkers -- Jabber supporter - http://www.jabber.org Jabber ID: [EMAIL PROTECTED] Debian GNU supporter - http://www.debian.org http://www.gnu.org IRC: jeroen@openprojects
