Hello everyone from toms, i was having a discussion with my roommates; i would like you guys to discuss and settle it.
Here's the underlying argument: if you have an oem built computer with the oem version of windows which included the bloatware that all OEMs include and you removed all the bloat ware and ran ccleaner or ncleaner, that computer will be slower( proven by benchmark or time difference) than if you were to just do a format and clean install of a retail version of the OS.
I believe that in my experience the computer is always more responsive with a format and clean install, no mater what i do to an out of the box oem it's still not as fast as a clean install. If i uninstall the bloatware with revo uninstaller and run ccleaner a few times and even ncleaner a few times(which is what i used to do) it will still be slower than a fresh install.
For my roommates, i used an example; if i were to get 3 identical (same company, model, etc.) computers from best buy leave one as it is out of box, uninstall all the bloat ware and run ccleaner/ncleaner a few times, and finally do a clean install on the last one. The one with a clean install will be fastest.
For the sake of the argument let's assume this the operating system is Windows XP and you buy 3 identical computers built in the same factory at the same time, and their serial numbers are sequential. You take them home open them up and prove that they are all working computers and running at the same (using 3DMark and the results show the same score or the score is off by .0X) benchmark. And then you modified the computers like such:
Computer One - Take it out of box, plug it in, (do initial test), and your done.
Computer Two - Take it out, plug it in, (do initial test), uninstall bloatware, run ncleaner, do benchmark and done.
Computer Three - Take it out, plug it in, (do initial test), pop in the "Retail Windows XP" install disc, format the harddrive, install xp, install proper drivers, do benchmark and done.
If this test proved that the retail install of XP is faster than an OEM with bloatware removed, would this be enough proof?
Roommate 1's take on this, i would need to be in the factor supervising the creation of my test computers from start to finish to prove there were no other manufacturing flaws for this to be accurate. I feel as though doing a benchmark once they are all out of the box would prove they have the same if not 99% same hardware build.
Roommate 2's take is that i would have to build the computers myself with the exact same hardware for it to be accurate. I feel that even though i get more control over the hardware going into the experiment any human error would increase infinity because a human couldn't put the same amount of thermal grease between the processor and the heatsink.
He dismisses this point because "robots mess up"; i find that true, but believe that robots are more accurate than humans, hence the reason robots are doing it.
So what is your take?
Is a de-bloatwared OEM OS slower or faster than a clean retail install?
Which review would you trust more, one using the same computer built by a robot in HP's factory or one built by a human using identical parts?
First of all the computer is built by an OEM in china or India for HP in the OEM's factory. The OEM's are the true manufacturers. Secondly the code for OEM Windows is the same as retail the difference is how they are supported. Retail version of Windows is supported by Microsoft and OEM versions are supported by the Computer company (HP). Just like Windows 7 the RTM is the same as the Retail version but Microsoft released the RTM so manufacturers could get ready for Oct 22. If there are difference in speed it's affected by hardware differences like processor speed, bus speed and graphics card to name a few.
Your experiment would be flawed if you ran it - you can't change the machine and run the test three times, if indeed they are the same. That won't show any correlation - who's to say one of the machines got the lower-quality parts put in it?
If you were really to do this, it's a much better idea to use a single machine. Take it out of the box, test it, then remove the bloatware, test it again, and then reinstall windows, and do your final test. This will remove the possibility of hardware differences.
It's also important you use the same drivers for all the hardware throughout the tests.
Anyway, as to the actual results...depending on the type of computer you got (if it was a budget machine and already low on memory, it will severely impact results, but a machine with 4+GB of memory would hardly flinch even with bloatware), you'll probably only see a minor performance jump from the bloated machine to the cleaned and fresh machines, and a negligible one from the cleaned to the fresh installed one. I'd say <5% and <2%, respectively.
Any creation of your own computer automatically invalidates the test. You're testing how fast the software runs in any given configuration - for that to happen, you cannot, under any condition, change the hardware.