Disclaimer: I did this test about 8-9 months ago and posted on my 'home' forums of OCForums.com. Litestep could have been streamlined since then and these test results could now be wrong.



I decided to start off the test with Explorer as the shell. I left everything at factory defaults. This is on a WinXP Pro SP2 install that is about 5 months old, so its not exactly spring fresh. I have decided to run each 3dMark in a repeat x3 test, to get an average. After watching 3dMark2001 run for about 3 hours, I realized I had set it to “loop” instead of repeat 3 times, it was repeating over and over. So I reset it. I then fell asleep in my computer chair watching the shader tests, with my wife's cat sleeping on my chest. After waking up and the test had completed, I went ahead and did the same thing for 3dMarks 03 and 05.

I then booted Into Litestep and repeated all the tests. I was initially surprised by the results, and thought they were anomolous. However, repeating them over and over provided the same results.

It turns out that the initial hypothesis that I went into this with was wrong. Just because I assumed the Alternative shell was faster, did not mean it was in all actuality. In fact, the evidence was so strong AGAINST this, that I did not have to run the battery of tests that I ended up running.

I'm not good with graphs myself, and it strained my meager mathematical abilities to set the percentages right, but I do have the results.

IMAGE

It looks like basic Explorer is the way to go. However, not having any other shells to test, I couldn't see if this was true across the board, but in general, this shocking result looks like it would repeat to a similar degree with other shells, as they would use resources differently than the basic Windows XP.

For reference, the base system is:

Athlon 64 3200+ (socket 754)
1Gig Corsair Value Select PC3200
Albatron GeForce6800 (fully unlockable to 16x6, but was at stock for these tests)
MSI K8T800 Motherboard

hgmm... looks like it wouldn't let me use the image tags.... oh well... just copy that url in between the [img] tags... it will take you to a .png of the excel graph I made.

The results are meaningless if you don't tell us what modules were loaded, how they were used... specifically, what theme was used while running Litestep. If it was Austerity, I don't think we can take the tests very seriously.

Performance can vary wildly from theme to theme.

Might also help to know which build of LiteStep you are using.

Litestep + Minimal Theme = Ass kicking performance over Explorer Shell! Period!

it was just a base Litestep install from about... last Septemberish. No extra themes of modules or plugins or anything.

Now i'll tell you it 'felt' faster than Explorer, and things generally loaded faster, but for some reason, everything related to graphics suffered a bit as you could see in that chart I made.

So, it was Austerity running on RC1.

The test was even outdated when you ran it 8-9 months ago.

I would expect Austerity and RC1 to give you those scores. Why don't you try a less ambitious theme and update to the current build and try again. I bet you will be very suprised with the changes. Heck...I think you might just see a difference between RC1 and Final.

Did my own tests, running experimental build 2006-03-30, and FBLD 1.7 (not a minimal theme). Used default benchmark settings.

Litestep

3DMark05 - 7667
3DMark06 - 4209

Explorer

3DMark05 - 7646
3DMark06 - 4210

ok.... so maybe it WAS the version I used.

I DID like Litestep when I last used it.... I just haven't been in windows much recently (I'm a Debian Man.)

there is no real way to compare LS with explorer, unless you mimicked the explorer shell in LS, but even then LS would be loading more cool stuff that is being used (should be used rather)

Who cares anyway?

Who cares anyway?

I think fractal hit the head on the nail there. 3DMark may look fancy but like all benchmark programs it doesn't tell you anything really useful. Benchmarks don't necessarily say anything about real-life performance, they're only good for comparing the performance of one system to another.

all i know is it put a SERIOUS hit on 3dPerformance. 10% slower in one of the tests!!

But if it's 10% slower doing something you only do 0.1% of the time in RL, then does it matter? Especially if LS helps you do 99% of the stuff that you really do faster?

all i know is it put a SERIOUS hit on 3dPerformance.

You're missing an important point, though. What is "it"? What put a "serious hit on 3dperformance"? It wasn't Litestep. If you used a modern build, with modern modules, and said 'this theme using this module with these settings' caused a performance hit, then you might be on to something. But, as it is, your tests are meaningless, and easily debunked.

no doubt xcal. Again, this was run last September with the build that was around then.

I'm getting ready to do another round of 3dMark tests, and I"ll probably try to include LiteStep in it again to see if that affects things again.

According to the other person in this thread, it shouldn't be a variation of more than a point or 2 by now.

You still aren't understanding. The 2 point variation was on his system with his setup. Each theme/build/module is different and will have a different level of impact on performance. As for increasing or decreasing 3D Performance...well...LS isn't 3D so it didn't do it. Explorer more than likely did something to help itself to a higher score.