(06-20-2012, 03:35 PM)gjfklhg Wrote: [ -> ] (06-20-2012, 02:31 PM)lionheart5656 Wrote: [ -> ] (06-19-2012, 10:22 AM)AnyOldName3 Wrote: [ -> ]In theory it means that they are faster. In practice, it's probably barely noticeable, but the placebo effect gives a good impression.
I disagree.
Disagree all you want, but it doesn't really mean anything. Every single benchmark that has been posted has shown them to be almost exactly the same. You would think if so many people are getting as large speedups as they claim, at least one of them would be able to post a benchmark proving it. Apparently not though.
You may want to read over http://en.wikipedia.org/wiki/Confirmation_bias to see why any results you observe without measuring them are completely meaningless.
To be fair I've not seen any definitive benchmarks on either side of the argument. At the end of the day, these are unofficial builds and clearly marked as such. If they work better for
some people in
some games under
some circumstances, why shouldn't they use them? And why is the onus on them to prove it?
If these builds work for some people, do you think that they would really want to spend time and effort to produce a serious benchmark up to your standards given the hostility shown in this thread? I don't think so.
And just for the record, I do occasionally use these builds but I am currently not. I just don't think the atmosphere is very becoming in this thread.
No, sorry.
If I posted benchmarks like that ''proving" that these builds were faster, I would be laughed at. At the very least I think people need to see screenshots. Check out the thread with the guy trying to prove his Bulldozer chip is underrated. Anybody could write down any frame rates they liked to bolster their position. There is no 'proof' there, and you seem to be confirming that you are also susceptible to confirmation bias.
Don't get me wrong, I am not arguing with you that these builds are faster or not, simply that people need to be objective and the rules need to be the same for both sides.
I never said that ICC builds are better on any PC, there are just too many factors and settings that influence overall performance. Anyone can use the standard builds if they provide sufficient performance or make no difference on the specific setup.
On my individual system setup however (Dell XPS 17 Mobile 4-Core+HT i7 CPU (8 Threads) + weak GT 555M GPU), the speed gain is no placebo effect. The difference is dramatic, reproducable and statistically significant in my specific configuration - I do not need to prove that to anyone since this is no scientific discussion and I do not care if other people believe it or not. A clear fact is that overall Gameplay in Zelda SS and SMG is much smoother using Lectrode or DefenderX ICC Builds on my configuration compared to non-ICC standard builds.
Please provide a proof that ICC builds are not faster in SMG / Zelda SS on a comparable configuration and comparable settings to mine. You will not be able to.
I'm tempted to try and see how much of a difference the SSE instruction sets make. If someone else is willing to do some objective benchmarks then that would be swell, as I'm not about to switch off the official builds until I see the numbers.
I have an i7 2600k + GT 570, so you wouldn't think speed is imperative. But I like to attempt to run games as accurately as possible at as high a resolution as possible (up to Full HD) while still maintaining a steady 60fps. So, any speedups will help allow me to figure out what's preventing the SRAM framebuffer copy to render particle alphas correctly in a specific game, or whatever it is I'm working on at the moment, and then promptly reading the documentation on it*.
If no one's up to the task, and next time I lurk the forums none have taken the challenge, then I might have to do some thorough benchmark testing myself. I implore anyone else who's interested to give it a shot, as I've got other projects on my hands right now.
*Yes, I'm mulling over contributing to the dolphin project, but am waiting for the proper time. In the meanwhile, practice...
I tried this, and I don't really see any difference between the regular 3.0 710 build and DefenderX's ICC version. The FPS and Speed are about the same, with subtle differences (+-1 fps). I'm running on a Core 2 Quad Q9550@Stock.
Quote:there are just too many factors and settings that influence overall performance.
This is the case with you
The result depends on many factors.
this .710 ICC has been the best for me for XenobladeChronicles, no freezing, no sound cracking (no patching), HD textures and about 4~5 FPS improvement. Great job! and thanks!
you're welcome. Always good to hear, that my builds work for you
Just be patient until i get my Macbook...