What Will Tegra 4 Mean to Windows RT?

windows phone 7 sets were orphaned because the new kernel wasn't getting ported. surface 1 sets won't be orphaned because they are just windows computers and there won't be a drastic kernel change happening for decades by which time even surface 2 devices will be obsolete.

true surface isn't as fast as I'd like but it is fast enough for the limited capabilities of windows RT. Ultimately tegra 5 will leave tegra 4 in the dust meaning that if being obsolete worries you, stay out of ARM chips.

Don't understand why you use the word 'orphaned'. AFAK, T4 is just another more powerful chip and is compatible with T3. All apps run in future T4 Win RT tablet will still run in current Surface RT. It is like we won't call PCs with Intel 386 orphaned when 486 is out.

Tegra 4 CPU cores are Cortex A15 based which still uses ARMV7 instructions. Do note that Tegra 3 used A9, thus there's a significant per-clock performance improvement, in addition to the clock speed boost. The net result should be 2x performance, up to 2.6x as Nvidia claims. There's no doubt that Windows RT was designed to target Cortex A15 SoCs, which were back in 2010 announced to release in 2012. However, except Samsung's Exynos 5250 all got delayed to 2013, and hence Windows RT was forced onto weaker Tegra 3. (Note: Qualcomm's S4 uses a custom core Krait, which is close A15, hence why Ativ Tab and XPS 10 perform much better than Surface RT running Tegra 3)

64-bit ARM comes with ARMV8 instruction set, the first of which should be here in 2014/15 with Cortex A57 CPU cores. This has to be the long term goal for Windows RT. WP7-WP8 might happen again - ARMV7 SoCs might not support the next-gen 64-bit ARMV8 ready Windows RT. They could do what they did with x86, keeping x86 and x86-64 lines going parallel, but they had a massive market share to cater to then.

I have to say though, x86 is looking very strong now. Anandtech has a pretty comprehensive article which shows clearly Atom has better performance/power ratio than even the latest ARM A15 SoC, while as you have seen in the Ativ Smart PC review, blowing Tegra 3 away. And Atom isn't really very good either.

My SoC to watch out for 2013 is definitely AMD's Temash. Remember when in 2010 AMD's Brazos came in and obsoleted the entire Atom line? They were late, but brilliant. It's going to happen again. Temash is going to be the one to watch, especially as Intel's next Atom is delayed to 2014.

As a side note, even today's aging 40nm AMD Z-60, another time-to-market play, has the most powerful GPU in the mobile land. It is a bit slower than Clover Trail on the CPU side and can only manage 6-8 hours battery though.

I think you, or Nvidia, meant to say the first quad core A15 ARM SoC. Without a Surface, I can only go by what those of you who have one say, but does it feel sluggish? By saying that the Tegra 4 is the target, if it doesn't feel sluggish, then Tegra 3 was probably the target.

It might be true to say that Microsoft can see the way the industry is going, and does have relationships with Nvidia, so they decided to port to ARM thanks to their friends their, but that doesn't mean they were targeting the Tegra 4. The first example of WindowsRT was shown at CES last year on Tegra 3.

Also, does the current Visual Studio target the Cortex A15 specific instructions, or is it geared toward the Cortex A8?

Anyway, I enjoy reading and watching you.

Paul, ARM is here to stay. ARM has gone from single-core low-end processors to quad-core in a very short amount of time compared to Intel. ARM's goal is low-power consumption, but there will come a time when ARM will be "good enough" for any consumer-level functionality that improvements in performance will be as much unnoticeable as they will be unmanageable with current manufacturing technologies. ARM is also very consumer-centric in their approach to chip design. They design chips with 3D and HD video support incorporated into them from the start. I think AMD's approach to heterogeneous system architecture will benefit ARM. Intel still thinks that x86 is the only path forward but they lack consumer multimedia focus, so their graphics will always suffer. They even thought that x86 would be usable for graphics, but it failed in development. AMD has far more usable compute power with their GPU cores in their APU processor line, and that will benefit the low-power integration market. Intel still needs discrete GPU's to get the same level of compute power, and that raises the total TDP and power consumption. The next step for AMD will be the completely heterogeneous processor core that handles integer, floating point, and SIMD instructions that can handle anything (including graphics) along a single pipeline, and they're almost there. The benefit is (obviously) power consumption, but also scalability. If you have 80+ GPU cores (which they do already in even their low-end APU's) that can also process x86 instructions, then you're talking some serious computing in a small chip. GPU Compute is no longer a separate "thing" anymore - it's just processing, regardless of whether the output is video or data-centric. Think about that for a minute. Currently your system outputs data separate from visualization of said data. Why? It makes more sense for visualization to be part of the main computational procedures of a computer. AMD and ARM are getting closer to this goal. Intel, not so much. This is why I think Intel doesn't innovate anymore. Intel only brings out technologies (3D gate) to increase clock speeds of x86 with no real focus on why.

BTW: I wrote a big comment on the other article about Windows 8 vs. Windows RT and it got nuked, so I'll summarize it here:

The best thing Microsoft should do is put more focus into refining WinRT. They shouldn't open up ARM development for the desktop because it's dead end. What they should do is open up the sideloading abilities of third-party software publishers. Forget about desktop refinements or combining WinRT with Desktop like you say, since this has major security implications. What they need to do is appease other publishers over their complaints about the Windows Store. Allow them to open their own app store but also require that these alt-loaded apps to be checked and code-signed by Microsoft to maintain stability and security within the system. Microsoft should absorb the cost of a minimal software inspection and code-signing in the price of an MSDN subscription or purchase of Visual Studio. Because they aren't processing credit cards or publishing them, they shouldn't charge their normal 20-30% processing fees. Updates should be handled by the software vendor as they always have before too. I don't think Microsoft should feel obligated to offer software updates, but they should at least dictate how third-party app stores work. If anything, they should make third-party app stores available through the Windows Store so that they can control the experience of those app stores, but do away with in-app purchases for third-party app stores. This way, the code-signing chain can be propagated from the Windows Store, to the third-party app store, to those third-party apps. Problem solved.

Tegra 4!
Definitely should be in the next Nokia Windows 8 Phone! Hopefully, it will also come with 64GB of storage.

And this in a nutshell is why I've chosen to sit out this 1st generation of Surface tablets. I do want to invest in one at some point. But with the current influx of new hardware & the ever evolving question of what Windows 8 in a tablet should be, the best approach right now is to see what develops by the late summer or fall of this year. It's these same reasons that are keeping me from upgrading my laptop. I've already put Windows 8 Pro on it & I'm thoroughly enjoying it & liking its more streamlined interface. New hardware just isn't that interesting or fitting to Windows 8 right now.

For now my Nexus 7 & upgraded HP laptop will have to suffice.