Google Pixel C: "Quad Core Tegra X1"

[Deleted member 13524]

The Pixel C tablet has just been announced as a "quad-core" Tegra X1 device.

Was it a mistake or could this be a Denver-based Tegra like the Nexus 9 before it?

 

[eastmen]

its an interesting design take , looks very much like the ipad pro. This could be smarter since android supports pointer devices like mice and track pads.

 

[willardjuice]

I don't think it was a mistake. The Nexus 9 was K1 (kepler and denver) and the Pixel is X1 (maxwell and 4 A57 + 4 A53). I wonder if the A57 cores and the A53 cores are allowed to run simultaneously (previous tegra devices it was one set or the other). It could be why they only label it a "quad-core".

 

[Deleted member 13524]

Anandtech is saying it's just the "regular" version of Tegra X1, so the big.LITTLE arrangement that is already present in the Shield TV.
Indeed the "quad-core" part is probably referring to the fact that nVidia is using cluster migration instead of global task scheduling, so there are only 4 cores available at each time.


As for what segment the device is aiming, I think price and size-wise this is going for the Surface 3 and regular ipad air.
The ipad pro is much larger and it seems to be going against the Surface 3 Pro, and both of them cost a lot more than $600.

 

[BRiT]

The Surface Pro 3 is also much larger at 12" screen.

Also, in one week I suspect the Surface 3 and Surface Pro 3 will see discounts as MS releases the Surface 4 / Surface 4 Pro line.

 

[BadTB25]

I just saw a hands-on video of this...seems clunky and inelegant with the way it attaches to the keyboard.

 

[Silent_Buddha]

The Surface Pro 3 is also much larger at 12" screen.

Also, in one week I suspect the Surface 3 and Surface Pro 3 will see discounts as MS releases the Surface 4 / Surface 4 Pro line.

I'm actually not expecting a Surface 4 release as there has been no update to the Atom lineup since Cherry Trail. The Surface 4 Pro is quite likely, however. It'd be nice if they aligned the launches of the consumer and professional lines, but we may see them remain separated due to Atom and Core CPU releases not being aligned.

Regards,
SB

 

[Deleted member 13524]

Yup. The Surface 3 is only 5 months old and Cherry Trail is still pretty recent. Furthermore, I've started to see some tablets with the lower-end X5 but the Surface 3 is still the only X7 Cherry Trail I've seen so far.

Intel is still sticking to annual tick-tock releases on Atom, so I'm guessing Surface 4 won't appear before mid-2016 with the what's-his-name-Trail carrying Goldmont cores.

 

[Rys]

I don't think it was a mistake. The Nexus 9 was K1 (kepler and denver) and the Pixel is X1 (maxwell and 4 A57 + 4 A53). I wonder if the A57 cores and the A53 cores are allowed to run simultaneously (previous tegra devices it was one set or the other). It could be why they only label it a "quad-core".

The Shield Android TV's X1 unplugs the A53s after boot, so it's quad-core A57-only for all intents and purposes. Probably the same for the Pixel C.

 

[liolio]

It makes me happy that Google is one step closer to realize that Android is ripe for touch enabled laptops.
It can happily coexist with ChromeOS which offer a different experience.

 

[Turbotab]

I noticed that gfxbench is listing OpenGL ES 3.2 support for the Tegra X1 - Shield TV, so perhaps the Pixel C will also ship with ES 3.2 drivers.

 

[mczak]

The Shield Android TV's X1 unplugs the A53s after boot, so it's quad-core A57-only for all intents and purposes. Probably the same for the Pixel C.

The Shield TV doesn't care much about a watt (or fractions of it) here or there more, but wouldn't that have quite a bad influence on battery life on the pixel c? Unless you're telling me something migration related is so broken that there's really no other choice .

 

[Rys]

I make no comment on why they run the hardware the way they do. Also worth noting is that at idle, Shield Android TV also unplugs 3 of the 4 A57s too, so when it's doing close to nothing it's just a single core in effect. Cores come online based on demand.

 

[Nebuchadnezzar]

Also worth noting is that at idle, Shield Android TV also unplugs 3 of the 4 A57s too, so when it's doing close to nothing it's just a single core in effect. Cores come online based on demand.

As far as I know their CPUQuiet driver is more efficient than hotplugging.

 

[swaaye]

They should have put a touchpad on it. The pretty screen is going to be a greasy mess.

 

[Ailuros]

If memory serves well they don't support global task scheduling for their big.LITTLE implementation. Now for the probable question why they implemented a 4+4 config in the first place and didn't use just a quad A57 from the get go, they probably were aiming at smaller tablets too when they designed the X1. It's not like the quad A53 killed their transistor budget either within the SoC die estate.

 

[RecessionCone]

If memory serves well they don't support global task scheduling for their big.LITTLE implementation. Now for the probable question why they implemented a 4+4 config in the first place and didn't use just a quad A57 from the get go, they probably were aiming at smaller tablets too when they designed the X1. It's not like the quad A53 killed their transistor budget either within the SoC die estate.

No one has been able to show that global task scheduling is useful. There are no benchmarks illustrating it saves power or improved performance.

Marketing.

 

[Nebuchadnezzar]

No one has been able to show that global task scheduling is useful. There are no benchmarks illustrating it saves power or improved performance.

Marketing.

Oh god not this discussion again. When basically every ARM (and MIPS for that matter) vendor in the industry besides Apple is embracing bL maybe it's time to start to think that it's maybe not marketing?

http://anandtech.com/show/9518/the-mobile-cpu-corecount-debate/18
http://anandtech.com/show/9330/exynos-7420-deep-dive/6

If perfect use-cases like this one don't convince you then nothing will.
http://anandtech.com/show/9518/the-mobile-cpu-corecount-debate/16

The only reason Nvidia doesn't have it is because their interconnect just can't do it.

 

[willardjuice]

When basically every ARM (and MIPS for that matter) vendor in the industry besides Apple is embracing bL maybe it's time to start to think that it's maybe not marketing?

But will Qualcomm's Kyro cores do it? That would be a true tell. Basically so far the only people to do big.little are those who use vanilla ARM cores. Perhaps it makes sense for ARM cores, but for the general case that's not very convincing. Why is it that two of the top performing SoCs (Intel and Apple) ignore it? And not too long ago (and maybe once again very soon) it was three (Intel, Apple, and Qualcomm).

 

[Nebuchadnezzar]

But will Qualcomm's Kyro cores do it? That would be a true tell. Basically so far the only people to do big.little are those who use vanilla ARM cores. Perhaps it makes sense for ARM cores, but for the general case that's not very convincing. Why is it that two of the top performing SoCs (Intel and Apple) ignore it? And not too long ago (and maybe once again very soon) it was three (Intel, Apple, and Qualcomm).

Yes they'll do it (in terms of basic principle of low-power + high-perf clusters).

Keep in mind this isn't a design decision that you simply switch to at the snap of a finger, ARM has quite the lead here. The Zenfone 2 doesn't really have what one could consider good battery life and Intel is otherwise quite less aggressive than other vendors in the space. Apple is the only outlier here but in my opinion that's mainly due to just having a vastly different and more efficient OS. I'm sure if I were to lay out the perf/W curves it'd end quite up below current ARM cores.