The only part of the kernel I had beef with was the quadrant score. I have no doubt that the turbo aspect is only a portion of the kernel, and that if it were disabled, snap would still be a damn fine custom kernel.
Turbo is optional. It does not have to be enabled with 7.6.
And I definitely appreciate the QA effort put in. I don't question that snap overall provides a better user experience with or without turbo. It worked great for me when I tried it. I just expected a lot more because it did score so high on quadrant.
I think this is where our disconnect is. Per my PM, what exactly do you man when you say you expect a lot more? Stock, the EVO already fires applications off pretty quick. Let's say 2 seconds. If we get even a modest 20% IO boost the translation in userspace is a 1.8 second launch. How do you measure the difference if you're an end user?
But I don't think it's unreasonable to caution these end users who are flashing their high quadrant scores around that you can't judge the kernel's overall performance based solely on that benchmark, especially when that benchmark puts so much weight into the IO aspect that you have optimized.
If you spent any time in the XDA thread, you will see multiple requests by me and other members of the dev team asking users to provide feedback outside of benchmarks. I think you're just being unfair here.
My observation is that battery life is unchanged (still great, but unchanged) regardless of the presence of turbo.
That's because we're not fiddling clock, we're dialing-in IO. We're not changing bus speed either, this is all about IO. Battery isn't going to be impacted.
In the end, I'd like to see people come to their conclusions about the kernel without relying on the benchmark score, the piece about snap I feel is misleading a lot of users.
We have never misled anybody because we've never asserted anything about performance. And as I've said previously, I've done my very best to provide subjective and objective measure. The XDA thread is replete with comment after comment claiming overall better performance. You've merely stated your opinion and haven't done anything to disprove performance gains.
I have spent enough time justifying what we've done. As most of you can imagine, I am not a rookie when it comes to computing and performance measurement. My personal view on snap--independent from the fact that I tweaked/compiled it--is that the end user experience is
superior to a stock kernel. This conclusion is based on subjective assessment after interaction with the device over time, a comprehensive beta test with
devoted and exceptionally particular beta testers, a thread with multiple independent observations claiming improvements, and objective measurements from the only known comprehensive benchmark application I have access to. Per my PM, Quadrant may not be perfect, but I do believe it demonstrates
relatativity. Compared to stock, I am convinced there is a relative improvement even if it's not 100% (because IO constitutes only 25% of the result).
I've also invited you to express your views on the XDA thread where there are far more first-hand experiences--yet you haven't showed up there. I wonder why?
J