Jim, I am one of the first to report the CPU increase issue. I am using a NVIDIA GeForce 6200 (latest drivers & DirectX).
Turning off write combing made little difference. Eliminating hardware acceleration made CPU utilization go higher. -Tim --- Tim Ellison ( [EMAIL PROTECTED] ) Integrated Technical Services ( http://www.itsco.com ) -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Jim Lux Sent: Saturday, February 11, 2006 6:42 PM To: richard allen Cc: FlexRadio Systems email reflector Subject: Re: [Flexradio] [Fwd: Re: Preview 14 significant CPU increase heretoo] At 02:55 PM 2/11/2006, richard allen wrote: >You are correct, Jim. The thing we are trying to determine, however, >is why the cpu loading increased from preview 13 to 14 and only on >certain folks systems. The behaviour relative to covering the window >has been present in earlier versions. Any changes in the underlying .net libraries? I would surmise that one cause of the disparity among users might be the type of graphics card and if the user is using acceleration on the card. I used to have a Matrox card that Matlab graphics (i.e. ordinary figures and plots, nothing fancy) would run extremely slow on (sucking up 100% Kernel CPU) if acceleration was turned on. Turn off acceleration, and while everything else repainted very slowly, Matlab at least worked. Turned out it had to do with the OpenGL interface and implemention in Matrox's drivers. A good test is to go to Right click on desktop, select properties to bring up display properties dialog Select Settings tab Click Advanced Select Troubleshoot tab Set hardware acceleration to other values (and or try other options, e.g. Enable write combining) Try the Rev 13, Rev 14 comparison There's probably some new feature in the new rev that triggers "bad" behavior on the repaints (i.e. triggers more painting than it should, or asks for something that triggers a drop back to implementing the draw using a lower level algorithm executed by the kernel, rather than the on-graphics board accelerator). Without a decent DIFF between versions, it's kind of hard to tell what changed and offer any useful suggestions of things that might cause problems. I would look for a new use of some diferent graphics method in drawing. >Richard W5SXD _______________________________________________ FlexRadio mailing list FlexRadio@flex-radio.biz http://mail.flex-radio.biz/mailman/listinfo/flexradio_flex-radio.biz Archive Link: http://mail.flex-radio.biz/pipermail/flexradio_flex-radio.biz/ FlexRadio Homepage: http://www.flex-radio.com