The only trouble with the ram limit is it doesn't work on the 3d system. So if you have a heavy 3d section of the script it can still go over.
-deke On Wed, Apr 20, 2011 at 11:43, Nathan Rusch <[email protected]> wrote: > Same question as before: What percentage of the system RAM is Nuke being > allowed to use, and how much RAM does that actually equate to? ('cache > memory usage' a.k.a. 'CacheLimit' knob in the Preferences). > > If you're hitting the ceiling of Nuke's allotted system RAM, your nodes will > start panic-swapping, and when Nuke shuts down, Windows will need to free > that swapped virtual memory before the program can exit (which can take a > loooong time on a single spinning-metal drive). > > -Nathan > > -----Original Message----- From: Francois Lord > Sent: Wednesday, April 20, 2011 11:36 AM > To: Nuke user discussion > Subject: Re: [Nuke-users] Nuke slow to close on Windows 7 > > We set the cache explicitely on the D: drive of each machine. > We launch only one render per machine with multithread. > The swap is set to the same min/max on every machine. > Anything else? This is the kind of bug that can take me weeks to find. > [sigh] > > On 20/04/2011 04:26, Deke Kincaid wrote: >> >> Set the swap to 25 gig and have the min/max the same so windows >> doesn't keep on dynamically resizing it. Also make sure your cache is >> explicitly set. Especially if you have floating profiles under >> windows then you run into the problem where lame windows puts tmp in a >> directory which lives on the server and or it gets copied back to the >> server every time the system logs out. >> >> 3rd, on your farm are you launching a render per core or are you >> multithreading? I would multithread if you don't have the ram since a >> frame per core uses far more ram. >> >> -deke >> >> On Tue, Apr 19, 2011 at 14:02, Francois Lord<[email protected]> >> wrote: >> > Are we the only ones to have problems closing Nuke after working on big >> > frames (4K) on Windows 7? It can take up to 20 minutes to close and >> during >> > that time, we can see the memory slowly being emptied in the task >> manager. >> > We didn't have that problem in Windows XP. We can kill the process >> and it >> > frees the memory instantly, but it's not a viable solution on the render >> > farm and currently we have frames that take 10 minutes to render >> when they >> > take only 1 minute in the UI. >> > Anything we can do about it, other then switching the farm to linux? >> > Thanks. >> > _______________________________________________ >> > Nuke-users mailing list >> > [email protected] >> > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users >> > >> _______________________________________________ >> Nuke-users mailing list >> [email protected] >> http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > > _______________________________________________ > Nuke-users mailing list > [email protected] > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > _______________________________________________ > Nuke-users mailing list > [email protected] > http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users > _______________________________________________ Nuke-users mailing list [email protected] http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-users
