[help] WB 5 Serious performance problems

[Re-writer]

Hi, i have recently downloaded WB 5, it claims to be more stable and better performance.

But, running PerformanceTest 6.0 from passMark, it shows a dramatical fact:

with WB loaded the test where worst that poor (3d test with 4 balls shows 4 fps)

The same with all graphics test.

The same test with WB unloaded:

3d test 393.8 fps.

I have ati radeon mobility x700, 128 MB

Acer ferrari 4000, 1GB Ram.

64 bits processor.

So, really wb improves performance?

What can be wrong?

greetings!

PS: i post this here because i think that is an impartial site (hope so)

[firecracker6]

What are you talking about here? Performance of desktop rendering or wb for use in conjunction with 3d apps/games? Are You using a 64bit OS?

[Desides]

WB5 indeed is easier on your system than using the default WinXP theme service.

It sounds to me like you've got a problem elsewhere. Are you using the latest drivers for your system's components?

[firecracker6]

Desides reminded me.. you should disable window's theming service if you are running windowblinds. But this's probably unrelated to the results of your tests

[Paralex]

wait, if you disable the theming service, then you can't apply WB themes...

[Autumnmist]

@Paralex, that's not true. The Theme Service is Microsoft's... WB uses its own which is faster.

@Re-writer - performance, especially if you're talking about using 3d apps/games is going to be highly dependent upon the theme you're using!

If you're comparing a WB theme like VistaXP that uses transparency/alpha-blending then it's of course going to be worse for performance than using Microsoft Luna.

What you should do is pick a single, non-crazy (no animations) theme available as an msstyle. (For example, Royal Color Mod: http://www.deviantart.com/deviation/8908279/). Run your test using this theme through Windows Theme Service.

Then use Skinstudio to import the theme into Windowblinds format and apply it through Windowblinds. Run your test again.

[dreamsINdigital]

Well does it actually feel like it's running slower? Numbers and benchmarks don't mean much.

[sunb1rd]

...it claims to be more stable and better performance.

(my interpretation)

The wording here is key to the real meaning:

'Stable'-- probably just means won't crash as much and is less buggy. this doesn't seem to be the problem.

'Better performance'-- WB5 actually improves your overall system performance? Performance of what? While doing what? I think "better performance" is a direct comparison of redraw rates of similar skins via visual style, windows theming or WB4.x and lower. WB5 is faster than those methods, when applying similar skins (e.g. NO per-pix alpha).

On 'per-pix alpha' glassy skins, the way WB5 lets you do this is, is by passing off the glass operations (per pixel alpha) calculations to your video card. If you video card is crunching window translucency, then of course it's going to take SOME performance away from your overall 3D performance.

You find a way for translucent windows that doesn't use more resources or do anything extra than 'regular' windows, then you let me know

Whew! hopefully all this is accurate and makes sense. I know there are some really knowledgeable people on this board that really know how this stuff works and they'll probably follow up w/ the real scoop.

EDIT: Re-writer, you'll get no sympathy from me or my mobility 9200. Your serious 'performance hit' is still faster than what my vaio could ever hope to muster

[spclffred]

(my interpretation)

The wording here is key to the real meaning:

'Stable'-- probably just means won't crash as much and is less buggy. this doesn't seem to be the problem.

'Better performance'-- WB5 actually improves your overall system performance? Performance of what? While doing what? I think "better performance" is a direct comparison of redraw rates of similar skins via visual style, windows theming or WB4.x and lower. WB5 is faster than those methods, when applying similar skins (e.g. NO per-pix alpha).

On 'per-pix alpha' glassy skins, the way WB5 lets you do this is, is by passing off the glass operations (per pixel alpha) calculations to your video card. If you video card is crunching window translucency, then of course it's going to take SOME performance away from your overall 3D performance.

You find a way for translucent windows that doesn't use more resources or do anything extra than 'regular' windows, then you let me know

Whew! hopefully all this is accurate and makes sense. I know there are some really knowledgeable people on this board that really know how this stuff works and they'll probably follow up w/ the real scoop.

you took the words right out of my mouth, great explaination.

[Re-writer]

I have re run my test, with all suggestions above. i found that certain themes (like the great kol's vista theme) make wb extremelly slow, even with disabled translucency.

This is too bad, since y really love WB and kol's fobidden theme (especially since ms don't like it).

But, about graphics card support: if i have per pixel alpha disabled, it means no benefit from hardware?

[Septimus]

Odd, I just did some simple windowed 3D tests and all seems okay. Even Quake 4 windowed w/ a transparent theme didn't suffer.

-Edit- Just had a look at that passMark thing, doesn't seem that good imho. There is a Windows test available which tests your drawing speeds, I'll try and find it.

[firecracker6]

thanks Septimus and good luck; I look forward to hearing what you find

[Septimus]

Found it, - http://www.stardock.com/products/xpbench/ (doesn't need installing, just runs)

All it does is test drawing speeds of your gfx card etc. I'm going to test now and see what I get so you can compare. For me I'm not noticing any performance hit with WB5 + WindowFX 3. Our spec's our quite similar so hope you can gauge something from it.

-Edit-

Right, the screenshot shows my score, whatever it means *shrug*, but it was very fast and smooth.

See if it can help point to what might be wrong.

[sunb1rd]

My overall benchmark score was 6.

That's just funny. Let's see... I had open:

expose (2 pics)

avedesk (w/ a slew of widgets including vumeter)

rklauncher

nubs

miranda (adium and ichat skins)

yz shadow (this is what kills it)

objectbar

EDIT: I turned all that crap off, re-ran w/ a vista glass skin and get about 25.

[Septimus]

I was running WB5 (with transparency), AveDesk, Nubs 0.8.3, OD+, Gaim, Firefox and WindowFX3 when I did my test. A little less than you but shouldn't pull your score down that much. YzShadow makes sense that it would kill it. Unlike WFX3 it uses the CPU to create the shadows.

Running P4 3.4HT Extreme and ATI X850XT PE PCI-E (only 256MB gfx RAM).