I'll take you on that bet. Higher base clocks and lower peak clocks reduce the jump from idle to load, not increase it.
How did you arrive to that conclusion? Idle power consumption has nothing to do with base clock speed. Higher base clock means it can spend more TDP while under full load.
Either way, you said that Clovertrail tablets won't use more than 4W under full load, after first saying they'll use 2W under tablet. You never said that the difference in idle and load consumption would be 4W, on the contrary you said it'd be under 2W.
That jump is 1.4W for Medfield in that graph, and given that Clovertrail has a sub-2W TDP, I seriously doubt that tablets based on it will jump from 2W idle to 4W+ load.
Let's try this again: the presented Medfield phone uses 3.25W under full load. What you're basically saying is that you can take that, and add:
1) A second core
2) A multicore GPU with a base core that's stronger
3) A higher base speed
4) A much bigger, and in many cases higher resolution screen
And you think all of that will take less than 750mW? The second core alone will take at least that...
Given that it was announced
back in 2010, and that
mass production was expected in 2011, my expectations weren't as loose as yours.
My expectations were based on what ARM and third party vendors were actually saying in late 2010, which was that it'll be in devices in 2013, which was later occasionally amended to late 2012. Given that release estimations are often very optimistic I'd say that expecting them to even meet them was not the least bit conservative.
I have no idea what
your expectations were based on. This wasn't at all inconsistent with lead times between when ARM announced Cortex-A8 and A9 and when they first appeared in devices.
I never said that I don't expect it in phones. I just said that low perf/W may explain the slow adoption.
Well no, you said the perf/W has already been slowing adoption, which is almost definitely not the case.
It's possible that cheap Chromebook components brought up the idle consumption of that platform, but the 4-5W jump under load is really big. 1.7GHz isn't that high, either, considering that ARM was touting 2.5GHz, so voltage probably isn't that high (and why would Google bother with trying to get 10% more perf out of it anyway when battery life is it's biggest weakness?).
ARM's 2.5GHz number was probably never something you could have expected to realize in a device with a tablet sized battery. They announced 2GHz Cortex-A9 hard macros too but I'm not aware of anyone having used them.
The load usage is big, too high, I agree with this, but it's not as if there aren't factors of the design that can affect total power consumption and not just idle, and it's not as if the SoCs used are necessarily the best of the lot.
At the moment, it looks like the 5250 will have to run at 1.0-1.3GHz to match Clovertrail's TDP, which would make their performance on par. We need more data to verify this, of course.
IMO you shouldn't compare values measured in a review with value that Intel claims. I wonder what the TDP on Medfield is.. do you think it's enough to allow for using 750mW more under load than Swift uses (when the GPU is barely used on either of course)? The core shouldn't be using more than 750mW just by itself, or at least at 1.6GHz.. if it's turboing up to 2GHz it might be a different story. But this is still pretty massive.
TDP is supposed to be an honest number since the design relies on it, but the trick Intel pulls with their Atom SoCs is that they're allowed to throttle to stay within safe thermal ranges. The reason why this is a trick is because the throttling is based on reading temperature, not power, meaning that if you design the cooling to a higher specification than what the TDP requires it'll hit higher performance margins.
Compare Cedar Trail N2600 and Clover Trail Z2760, and we see that Z2760 has a higher max clock speed (1.8GHz vs 1.6GHz), a faster GPU (533MHz vs 400MHz), and several more integrated peripherals. The only thing that uses more power for N2600 is that the memory controller supports DDR3 over LPDDR2 and the DMI to connect to an NM10. Yet it carries just over half the TDP. What does that tell you?
