NVidia Tegra 3 power consumption (AT article)

[sontin]

Originally Posted by Skurge

I'm sorry, but that logic is poor. When are comparing the architectures you use the same OS.

Tegra 3 is powering Windows RT and Android. So i guess that makes Tegra 3 or better ARM the superior CPU architecture?!

I could say it's not Intel's fault that nVidia couldn't come up with a more elegant Solution for bringing idle power use down. Samsung brought down idle power use without a 5th core.

The 5th core is the "elegant Solution". It's working since the Prime from last year.

There is really no point in comparing an Android tablet to a Windows tablet when you are looking at which is the more efficient architecture.

Sure when the OS is limiting the potential of the architecture.

For all we know that advantage will carry over to the Android if there was such a tablet. How much does that 5th core really help idle power use compared to smaller 32nm process.

So Android is the better OS than Windows RT/8?

EDIT: I'm talking about the SOC as a whole, include the process advantage, not just the architecture.

Tegra 3 uses two different process technologies: LP and HP. Without the 5th core the idea of this integration is not there so the test says more about the process technolgy than about the CPU architectures.

[BenSkywalker]

You guys are aware that the fundamental Atom architecture in "Clover Trail" is, what? 4 years old now? And it's still competitive with Krait and A6?

64bit ARM spec(v8) was finalized in 2011, that's the generation after A15. What is your point exactly? The SoC market is *nothing* like the desktop market. If it were, Intel may stand a chance.

But when has Intel not eventually dominated a market that it felt truly threatened in?

XScale, Itanic and Larrabee off the top of my head. Those represent more then the entire net worth of nVidia and AMD combined. Intel fails, and fails *HUGE* on a not terribly infrequent basis.

Edit: Rofl, Ben really has an axe to grind.

You know how long from now I'm still going to have to explain how utterly worthless that article is? Meh.

To be fair, all the numbers seem discrepant from review to review and I have no idea what kind of Atom that is in the graph

2GHZ Medfield. Browser benchmarks are only useful for comparing browsers, never CPUs(honestly the main point I was getting at linking up those numbers, 100% is never an acceptable margin of error for anything that I am aware of).

[lopri]

Originally Posted by BenSkywalker

what's worse is he admits that he knows T3 isn't running properly under WinRT

That is really what makes the whole endeavor a sad waste of time, IMO. Tegra 3 is not a cutting-edge today, but we know its design and what its 5th core does. You could say that Surface RT is a sub-optimal product compared to Surface Atom, from performance/watt point of view, because of MS' gimped implementation of Tegra 3. But turning that around to use as an evidence of power [in]efficiency of a CPU architecture is kind of an oxymoron, as the reviewer is evidently aware. Why spend so much time (and bandwidth) on an investigation based on a false premise, I will never know.

A similar analogy would be something like this: "Radeon 6870 consumes more power than Radeon 5830, thus Bart is less efficient than Evergreen XT. Oh, BTW, we don't know which one is faster."

[tvdang7]

I never understood why tegra is used at all anyway..... I can understand tegra 2 with the initial android tablets coming out but why did android tablets still continue to use tegra 3 while android phones mostly used qualcomm? Is there something about driving a big display that a tegra 3 can do better than a s4 could?

[gizbug]

Great read on this article

[tviceman]

Originally Posted by tvdang7

I never understood why tegra is used at all anyway..... I can understand tegra 2 with the initial android tablets coming out but why did android tablets still continue to use tegra 3 while android phones mostly used qualcomm? Is there something about driving a big display that a tegra 3 can do better than a s4 could?

Tegra 3 came out 5 months before the S4 did which was a significant lead time. Also, Nvidia is known to have among the best driver / software / developer support, which goes a long ways in helping to acquire adoptions. S4 is clearly superior to Tegra 3, but it seems like Qualcomm didn't aggressively go after tablets OR many of the design wins Tegra 3 secured came before S4 was ready to go. In the case of the Surface RT, it simply boiled down to Nvidia having significantly more robust drivers than any other ARM vendor (which makes sense, since Nvidia has been developing windows drivers for over 10 years).

[Alaa]

How about throwing some Nexus 10 benches in the middle so we can get a bigger picture of power usage in the tablet arena?

[Skurge]

Originally Posted by tviceman

Tegra 3 came out 5 months before the S4 did which was a significant lead time. Also, Nvidia is known to have among the best driver / software / developer support, which goes a long ways in helping to acquire adoptions. S4 is clearly superior to Tegra 3, but it seems like Qualcomm didn't aggressively go after tablets OR many of the design wins Tegra 3 secured came before S4 was ready to go. In the case of the Surface RT, it simply boiled down to Nvidia having significantly more robust drivers than any other ARM vendor (which makes sense, since Nvidia has been developing windows drivers for over 10 years).

I think his question is why is Tegra 3 in the surface, but S4 is Windows Phone 8 devices?

[tviceman]

Originally Posted by Skurge

I think his question is why is Tegra 3 in the surface, but S4 is Windows Phone 8 devices?

Integrated LTE modem.