It should not be too hard to support cuda on VMWARE (at least for VMWARE developers) and i doubt it would have a serious performance impact, since 1) the communication with the device is in the order of magnitude of 5 mbyte/sec with < 100 kerneles launched per second 2) Once the kernel is started there is no difference on the GPU side 3) the CPU has very little to do
But this is really academical, since a full featured windows client is available today/tomorrow. I use VMware the other way around, and are not sure about this. But > this sounds hard, and something if doable might give a significant > performance drop. > > At least, on my windows guests I don't see the Nvidia card and use > VMware video drivers and not Nvidia drivers. So at least on my setup, > I don't think running CUDA applications is possible. There might me > some tweeks you can do to get more native access to the video > hardware, I haven't looked much on that. > > On Mon, Sep 14, 2009 at 4:46 AM, Jonathan <[email protected]> wrote: > Hey all, > I was wondering what would the performance drop be if I ran the linux > bin > on a Linux Virtual Machine sitting in op of a Windows Host. > > Would virtualizing therefore have an impact? > > Jon > > /****************** > Some Interesting websites: > http://ac3bf1.org - My Personal Blog > http://ninux.org - Wireless Community Rome > ******************/ > _______________________________________________ > A51 mailing list > [email protected] > http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51 > > _______________________________________________ A51 mailing list A51@ > lists.reflextor.com http://lists.lists.reflextor.com/cgi-bin/mailman/ > > listinfo/a51 ______________________________________________________ GRATIS für alle WEB.DE-Nutzer: Die maxdome Movie-FLAT! Jetzt freischalten unter http://movieflat.web.de _______________________________________________ A51 mailing list [email protected] http://lists.lists.reflextor.com/cgi-bin/mailman/listinfo/a51
