Tiny, rlowens, Ed, et al. - Please set your Microwave's to '7g Defrost'

..... Solved!

Beta 7g does correct this problem on my system. Thanks!

Unfortunately, the 7G version introduced a bug here. While I never did duplicate the "frozen fish" bug, I downloaded and tried the latest beta this morning and immediately ran into a screen refresh problem. When switching from full screen to windowed (via the F key), the desktop backgrounds on the non-main screens are not refreshed. In fact, even parts of open applications are not (though other parts strangely enough are). I'll pop a screencap here:

http://webpages.charter.net/rlsimon/screencap.jpg

I'm running a 3 screen setup with an ATI 9550 driving the center (#1) and righthand (#2) screens and an NVida FX5200 driving the lefthand (#3) monitor. In full screen mode (or windowed mode as long as the window does NOT cross onto another monitor) I get decent frame rates in the 33-35fps range. Any time any part of the window is displayed in either the lefthand or righthand screens (#2 or #3), the frame rates drop to ~5fps. This includes if I shrink the window down and move it so that it is entirely within the #2 or #3 monitor. I only get full frame rates when the window is completely enclosed within the #1 (center) monitor. Here's another screencap showing that:

http://webpages.charter.net/rlsimon/screencap2.jpg

You can note by the Netstat util running in the LR corner that the problem doesn't appear to be one of running out of CPU cycles. While the CPU usage spikes during the initial startup and switch-over to windowed mode, it then drops well down to within normal ranges.

Let me know if there's any other info I can provide to help track this one down.

Welcome to the forum Rick, nice detective work. ;)

Rick, - Welcome! :) - The screen refresh of the secondary monitor on my setup, is something that usually clears itself up after a few moments, and has always been there with my current video card and drivers. But I must admit that it was not as pronounced with my previous card and driver. - I have not noticed any change introduced after downloading and running Beta 7g.

I do think you should try changing over the 'Primary Monitor' to one of your other screens, to see what if any difference it makes. It can and does affect the refresh of the backgrounds on my system, though not always! - I have assumed that this is a video card/driver problem, more than a MA3 bug.

The large drop in FPS for a window covering a part of more than one monitor has been present since MA3 went to multi screens. I've reported on it a number of times in the past. - Most recently here and here for example, and also numerous time during previous testing - I too hope that it can eventually be solved.

Unfortunately, that is not the case here. The affected areas will stay that way until/unless I take a window and "swipe" across it, forcing a refresh.

I went ahead and tried it here but it makes no difference. Regardless of which monitor is set to primary, when MA3 switches to windowed mode it shows up on the primary monitor whose background is properly refreshed while the other two monitor backgrounds are not.


In reviewing my earlier post, I didn't make it as clear as I should have. The point I was trying to make was that prior to the 7G version, the frame rates stayed up in the 30's as long as the window did not touch more than one screen at a time. It didn't matter whether the window was on screen #1, 2 or 3. So long as it did not span more than one screen, the frame rates stayed up in the 30's. Now with 7G, it only exhibits that behavior on the primary monitor on my system. If the window is on either of the other screens, even if it is not spanning multiple screens, the frame rate drops to and stays at ~5-6fps. Provided that this can be duplicated on other systems, perhaps that info along with knowing just what parts of the coding were changed to fix the "frozen fish" bug will help narrow down the cause of the low frame rates when spanning screens with a window.

Your first point clearly shows that our two systems are acting somewhat differently, which is not surprising, as we are using different hardware and therefore different drivers.

Your second point surprises me a little, as it is the setting of Primary Monitor that could cause the fish to freeze in a window. - I feel sure that if you run Beta 7e or earlier, with a Primary Monitor change, you will find one situation that produces 'frozen fish'. - I know that that is not something you were testing for on this occasion, but I mention it only to illustrate the difference that the choice of primary can make.

You said that you never did duplicate the "frozen fish" bug. - The occurrence of that bug appears to depend solely on which of your screens you choose as the Primary Monitor. - Which in turn is determined by which socket you connect each screen to on your video card, and/or the position of the screens relative to each other.

As regard your last point, my FPS for the Windowed mode, prior to and including 7g, have always fallen to an unacceptable level if the whole window was not on the screen chosen as primary. In one of my earlier posts I made a mistake, - I said "Once the window is moved onto only one monitor, everything returns to normal." - That should be: "Once the window is moved onto the Primary monitor, everything returns to normal."

I do think we should be careful about laying the blame at MA3 when video cards and/or drivers can make such a difference, and act differently on different systems. - I never fail to be amazed at the effect that re-installing Directx again, can have after some visual glitch suddenly appears as if from nowhere!

"When switching from full screen to windowed (via the F key), the desktop backgrounds on the non-main screens are not refreshed"

I have a fix for this bug. When Jim sends out the next update, let's hope my fix gets included.

It will be in Beta7h. I'm also trying to include a fix for the "120 DPI" Fishpicker problem, but this is turning out to be a major stumbling block. Researching it on the Internet, I've found that it's a problem that's plagued developers for a long time.

Cliff,
Great! Awesome detective work on this one.

Jim,
Any particular reason we're on instances of 7 on the beta now? Not really an issue, but why not continue with #8?

All the changes lately have been incremental. No major new additions.

Ok, thanks.