Well im here again with one of those really long over informed articles about things I do in my spare time.

This time it’s all about the OS. Windows that is, I’m off the linux fad today and since I decided to reinstall for that fresh install benching, I thought I’d write up an article about it.

I’m going to show the comparison of a clogged up windows install from about 6 months ago, running SP1 and some updates, Then I’m going to compare it to a fresh install of Windows(SP2), Then using my trusty nLite app, im going to cut windows down to next near nothing, and show some comparisons to my benchmarks.

All of my scores today will be in PCMark2004, showing CPU, Memory, Graphics, and HDD scores, along side the PCmark score I get.

The reason I didn’t test the PC with other benchmarks like 3dMark or SiSoftware SANDRA, is solely on time consumption, this whole project took about three to four nights (4-7 hours a night) to complete, to reinstall windows, then install lots of benchmarking programs, would have just added days. And to think I was planning on getting it done in one night.

Hardware used:
Sempron 64 2800+
Ultra’s CL3 PC3200 512MB (single SIMM)
PNY GeForce FX 5500 128MB AGP card
WD2500JB SATA 150 HDD

First I ran all my benchmarks, starting out with stock speeds on my Sempron 2800 and running my FX 5500 at stock 270/350, and then ran it up into my OCed state (more stable than fragile) at 345FSB clocking in at 2.760GHz and running my card at 310/480 (more stable than fragile again.)

All the tests ran in PCMark04, were run over and over for best results, if I didn’t get a screenshot of the highest score, I’ll write it in somewhere that a higher score was there but only differed so many points. I ran each CPU, Memory, etc at least three times or more.

And then I took lots of screenshots, of the online result pages (im not a Futuremark ORB member) and saved the excel sheets for use in my charts.

So waiting for all those graphs to show you what’s best, here they are.



Nice thing about the nLite app, you can get rid of stuff like services that don’t need running, and it’ll disable them on the install, making tweaking very little, with SP2 and the clogged system I was running, I simply just shut down everything I didn’t need running. Show up that SP2 actually beats out the nLite version at stock, but not OCed.



So the memory show a noticeable change from clogged up, and not so clogged up. Although at my OCed state, the clogged version of XP shows up SP2, its still slower than that nice cut back nLite copy. As well as SP2 shows a lead in the stock specs market again, but wait don’t we OC computer to get away from stock, sorry windows, you need to be made to run on overclocked systems too.



Graphics now wouldn’t you know it one of the most affected areas of that clogged up windows install. A nice raise in score with the nLite copy, nice enough to show that stock speeds and my OCed speeds, aren’t much different in synthetic tests. But you can see at stock speeds the clogged copy of windows was holding back this card (3dmark might show this difference more.)



This is what really, really is boggling my mind. What happened here??

I thought the 2375 score was great after running the benchmark recently and only getting barely a 2k score. But dam once SP2 ran I was like wow this drive is good a little. Then I ran my nLite copy and Bam the score dropped. Gotta give that full blown copy of SP2 kudos here, it was made for HDD optimization, and so it wins one.


Now for PCmarks, since you’ve all seen the actual scores compare you should know why this is scored this way, beside the HDD tests SP2 only can get so close to the nLite copy, and since even at stock the nLite copy takes the only one score, it still wins by a few points. For overclocking, nLite is what you want to use, if you benchmarks like crazy one day and want a tiny overhead on your scores, try backing all the junk out of SP2 and XP. Remember that many benchmarking apps need some things to run, such as in PCmark, you need IE6 and WMP9 or better, as well as the WMencoder that isnt stock with SP2 even. For SANDRA you need the MDAC files that come with all your Windows, but you can pull them out by accident in nLite.

Well this shows the difference and I hope it helps some of you get that last point in your results, this took a lot more time than I expected but hey it was worth it, should help out a few of you striving for that last point for a record, or maybe help out some of the n00bs into getting started benchmarking the right way.

Attached Thumbnails

     

   

Took a quick look, intresting results... will have to look at in more detail tommorow after I have had some rest!

I gotta stop posting these things in the middle of the night, bump for the day grew

Thanks for going to al the effort, results look pretty much like I would have thought although i would have expected the nLite version of XP to come out another couple of hundred points ahead.

Good work Slughead - interesting results. I'd rep ya if I could. I'm surprised no-one's done this test before.

Great info

Made me decide to re-install on my primary machine, been about a year so it was due.
+ rep to ya