Daschund on wheels: slow, slow, slow [SOLVED]

This is interesting. The 9200 should work. Have you tried this:

Use latest version of ATI drivers, without the Control Panel installed. If you have the Control Panel installed, uninstall, get latest drivers from ATI, download and install without the Control Panel. This has worked with some earlier WB versions and ATI 9200. I am not absolutely sure about the 9200, but the 9200se works. I switched to geForce 5500 right before WB 5 came out, while still in beta. Hopefully, someone out there with a 9200 will concur or correct me.

Try Omega drivers for ATI cards. Google it.

I'm not concerned about the pp shading at this stage. Still, it seems that paying for the full version has conferred no benefits, only, on the contrary, trashing my system.

My computer is intolerably and unuseably slow since the upgrade to the full Window Blinds product but the demo version worked properly - apart from the pp shading of course - right up until I upgraded.

I already have the 5.13 Catalyst drivers installed, but with the Control Center, not the Control Panel.

Thanks for your reply though.

I think that there is some driver contention.

Re Omega Drivers, the question is - why would the standard catalyst drivers fail in the five minutes between running the demo and running the full program? Is not the core WB code identical in both?

Did you uninstall the trial version before upgrading? You should.

This might be worth a try. Try uninstalling everything. Run the zapper utility to clean out all Stardock stuff. Then reinstall the full version.

Zapper utility.. https://www.wincustomize.com/Forums.aspx?ForumID=144&AID=64426#506598

No I didn't uninstall the trial; that was probably my big mistake - I thought that an upgrade would occur. I did though nuke all of the Stardock stuff before reinstalling it the second time.

Anyway I've just totally nuked all of the ATI software and reinstalled the drivers (without the Control Center) and all is well apart from the per pixel acceleration.

Thank you all for your responses.

I wonder if any Stardock staff can explain what happened (in technical terms) and tell us why ppa is unavailable on a card that supports it.

I was fiddling with the WB configuration and again my system started running like a dog, after checking the two boxes on the System Information screen.

I uninstalled the Radeon Catalyst drivers and unloaded WB on reboot. The system ran at the same low speed without the Catalyst drivers and without WB as it did WITH the Catalyst drivers AND WB.

It is as if WB disables the 2D acceleration of the Catalyst drivers.

Another reinstall of Catalyst fixed the problem, but I am now loathe to configure WB at all lest I need yet again reinstall Catalyst.

Further, I see in other threads that Stardock has set some kind of arbitrary reinstall limit on their software. A pity, because for some of us a reinstall is the only way to get it half working. Surely it is within the wit of Stardock to decrement the install counter when their software gets uninstalled on any particular machine. Perhaps not.

Can any Stardock staff tell me what exactly is the interaction between WindowBlinds 5 and Radeon Catalyst 5.13? Is there a fix planned for this?

And I ask again, why does per pixel acceleration not work on a card that supports it. Stardock promote ppa as one of the features of the Enhanced version. I don't think so.

Sometimes when you install Catalyst drivers you find you have no 2D HW acc at all.  I believe this happens if the smartgart part of the drivers gets upset.  I have had this happen to me after installing new drivers & then rebooting.  You will notice the first time the machine boots with new drivers, there is no HW acc as the login screen paints slowly.  Then the drivers perform tests on the pc and enables various settings.  Occasionally this does not happen & you are left with a very slow UI as the drivers are running without any acceleration.

This is nothing to do with WB and WB works perfectly with the 5.13 drivers.

Regarding the lack of per pixel on your card, it seems ATI decided to drop 2D per pixel acceleration from the Cat 5.x series of drivers for the 9200/9250 series of cards.  The last driver for your card which reported support was the 4.12 driver.  It is frustrating because your cards hardware is capable of doing per pixel borders, but the drivers are not providing a way to use it.

On the reinstalling limit, it is not possible for us to decrement the install counter during uninstall as this would be open to abuse as people could install the software and then uninstall it and then revert to a backup activated copy.

Later ATI cards all support it fine with the 5.13 drivers and if ATI fix the 9200 drivers then WB will support it on your card too.

Thank you Neil. I'll try and explain it again.

After installing WB, Windows slowed to a crawl. That was the subject of my first post.

I got WB working properly after nuking then reinstalling the ATI Catalyst drivers as scorpio-logic suggested here.

With a properly running system I then adjusted some settings in the WB panel. Immediately Windows slowed down again. This slow down did not occur after another reboot, it occured as soon as I adjusted the WB settings in my running session.

To fix the problem I now knew what to do. I yet again nuked the Catalyst drivers. After re-booting but before reinstalling Catalyst again, I unloaded WB which automatically loads at boot.

I noticed that the speed of Windows in this state, that is without any ATI drivers installed or WindowBlinds loaded, was the same unaccelerated speed that I had just experienced immediately after changing the WB settings in the properly running 2D accelerated system.

That is to say the ATI drivers seemingly got b0rked simply by my changing the WB settings whilst running a normal system.
(The system had stopped running at normal 2D accelerated speed, but was otherwise normal.)

After reinstalling Catalyst again I recovered the system.

My tentative conclusion was that WB had caused the Catalyst drivers to fail. Specifically, WB stopped the drivers' 2D acceleration.

Yes, WB did work perfectly with the Catalyst 5.13 drivers... until I adjusted the WB settings. Then either WB or the drivers partly failed.

My conclusion that WB was to blame was based on the fact that Windows slowed to a crawl immediately after my clicking 'Apply Changes' on the WB panel. Coincidence?

I'm now afraid to use any WB settings at all now.

On the subject of ppa, perhaps the Omega drivers work. I'll try them next time I have to reinstall; that time can't be far away.

WB cannot disable the drivers acceleration, and given the complete lack of other reports of problems with the cat 5.13s, I can only assume that either changing the WB skin caused a process to use lots of cpu and thats why everything felt slow, or it was just a pure chance it happened at that time.

There is one other possibility which is the ATI VPU recover kicked in, but that would have displayed a message on screen and would indicate problems with the card itself.

On the subject of ppa, I don't think the Omega drivers will support it either as they are just the ATI drivers with some registry tweaks.

Thanks, Neil, for clarifying. Have I told you how much I appreciate your efforts on WB? Excellent!

Edit: Double-post...!

As I reported earlier, cpu idle was typically 80-90%

I watch Process Explorer continuously, and cpu usage did not increase during the times that the graphical system was running slow. All other processes ran normally. The slowness was confined to the gui - screen refreshes, etc - making the system feel unusable.

There was no message about VPU recover.

Further, I run no extraneous services, and little more than the svchost stuff, explorer and procexp - and WB configuration, of course - were running at the time.

The same slowdown occurred yet again after I adjusted the WB settings.

Is there anyone at Stardock able to help me solve this problem?

Neil is Stardock.

Neil 'wrote' Windowblinds.  It's probably quite safe to say he knows more about the creature than anyone...

Also means you are getting 'the best' source of assistance....

Once this slowdown happens, is your PC still slow if you set WB to use the classic skin and then reboot?

I don't know. Next time the slowdown occurs I will try it and post here. WB is currently running properly after a fresh reinstall of WB and Catalyst following the BSOD episode detailed elsewhere.

I closed my system yesterday after running a session at the normal gui speed with WB.

On booting today, the gui system was again glacial.

I changed to the 'Classic' session and explorer crashed with an unhandled exception during the 'Please Wait' phase.

I rebooted and the gui had changed to Classic, but was still running glacially. I changed to 'Classic' again in the WB control panel to be sure and rebooted. Same thing. Unloading WB makes no difference.

Once WB starts to cause the gui to run slowly the only thing that restores the system to normal is to nuke Catalyst and reinstall it.

I did try a reinstall of Catalyst without nuking it first, but started getting the atid2vag infinite loop BSOD on reboot.

Does Stardock have any issues with ati2dvag.dll?

To be clear, once WB starts operating at glacial speed ONLY a nuke and reinstall of Catalyst restores the system to normal, but the problem then recurrs either through adjusting the WB settings or simply rebooting.

If you are getting BSOD from atid2vag then you need to contact ATI as thats very much a driver issue.  Either due to a driver bug, or a possible hardware glitch.  If you are overclocking anything (and I mean anything), then I would stop overclocking too.

It sounds like WB is pushing the hardware more, the drivers detect a problem and so turn off speed.  It may be they have changed VPU Recover to not display a message anymore.

We have lots of customers on ATI cards (Many of the test machines have ATI too), and we are not having these problems.  In fact this is the first report I have ever seen of an issue like this.

I don't overclock.

I have no confidence in ATI's software and would not buy another ATI card. I've switched to the latest Omega drivers; they are based on 5.12 so this might solve the problem.

I've only seen the ati2dvag error since installing WB, but it is not a problem unless I fiddle with the ATI Smartgart and vpu settings.

Could WB be pushing the 9200 hardware in 2D harder than a game pushes it in 3D? And where would ATI preserve the "throttle" state for its drivers between reboots?

Two other things: as I stated at first, the demo version of WB did not exhibit this behavior. Is the relevant code not the same in the demo as in the enhanced WB? And is WB SMP-safe?

I have another thought on this.  There are two separate problems.

1) The bug somewhere that is causing the go-slow.

2) The fact that once the go-slow arises a nuke of Catalyst is needed to clear it.

On point two I think that ATI could be storing a persistent flag in the registry once it gets its knickers in a twist, that only a clear-out of the registry resolves.

I should and will stay out of this but I thought I would just throw in..Have you tried a Driver Cleaner? http://www.drivercleaner.net/

I use a driver cleaner;  that's why I wrote that I 'nuked' the Catalyst drivers.  Driver Cleaner Pro is essential.

I've now found how to restore normal operation.  By retesting PCI and AGP read and writes, the system comes up working properly with the PCI settings on and the AGP settings off.

Somewhere WB is causing these settings to get screwed up, but that is not necessarily due to a fault in WB.  Any thoughts, Neil?



 

I finally solved this problem, though I have no idea why the solution worked.

The problem came down to the directx settings DirectDraw Acceleration and Direct3D Acceleration becoming disabled when WindowBlinds was running.

(AGP Texture Acceleration for some reason is always disabled.)

I reduced my AGP Aperture size to 32MB in the BIOS, and this cured the problem.

(I tried various AGP Aperture sizes and one enabled Texture Acceleration, but disabled the other two accelerations instead.)

I have been unable to find any credible documentation at all as to why this fix to the AGP Aperture size would work.