UPDATE (3/31/15): Thanks to another tip we can confirm that the new SHIELD P2523 will have the Tegra X1 SoC in it. From this manifest document you'll see the Tegra T210 listed (the same part marketed as X1) as well as the code name "Loki." Remember that the first SHIELD Portable device was code named Thor. Oh, so clever, NVIDIA.
Based on a rumor posted by Brad over at Lilliputing, it appears we can expect an updated NVIDIA SHIELD Portable device sometime later in 2015. According to both the Bluetooth and Wi-Fi certification websites, a device going by the name "NVIDIA Shield Portable P2523" has been submitted. There isn't a lot of detail though:
- 802.11a/b/g/n/ac dual-band 2.4 GHz and 5 GHz WiFi
- Bluetooth 4.1
- Android 5.0
- Firmware version 3.10.61
We definitely have a new device here as the initial SHIELD Portable did not includ 802.11ac support at all. And though no data is there to support it, you have to assume that NVIDIA would be using the new Tegra X1 processor in any new SHIELD devices coming out this year. I already previewd the new SHIELD console from GDC that utilizes that same SoC, but it might require a lower clocked, lower power version of the processor to help with heat and battery life on a portable unit.
There’s no information about the processor, screen, or other hardware. But if the new Shield portable is anything like the original, it’ll probably consist of what looks like an Xbox-style game controller with an attached 5 inch display which you can fold up to play games on the go.
And if it’s anything like the new NVIDIA Shield console, it could have a shiny new NVIDIA Tegra X1 processor to replace the aging Tegra 4 chip found in the original Shield Portable.
I wouldn’t be surprised if it also had a higher-resolution display, more memory, or other improvements.
Keep an eye out – NVIDIA may be making a push for even more SHIELD hardware this summer.
I think this is cool, but I
I think this is cool, but I would really be excited about a new Shield Tablet, with a larger (about 10 inch) screen, with the added benefits of the Tegra X1.
That would make for one heck of a Andriod gaming tablet.
Except for this iteration the
Except for this iteration the Denver cores are absent, hopefully Nvidia will bring Denver back on a smaller process node. It takes 2(and then some) of the ARM reference cores to equal 1 Denver core in IPC, and the Denver cores had a novel way of dealing with out of order execution in micro-ops/hardware with the reorder buffer/memory, via dynamic code optimization and microcode caching for reuse of decoded instructions. Dynamic Code Optimization Breaks the out-of-order window physical limitation Improves efficiency by optimizing once and using many times. The Denver core with an 8 decode/pre-decode and 7+ IPC. Maybe Nvidia can tweak the Denver’s design libraries in the same manner as AMD is doing with Carrizo and get a little better power profile, or maybe a process node shrink will be enough. ARM holdings’ reference design A53/A57 cores are just not as wide order superscalar as the custom Apple A7/A8, or the Denver custom cores. AMDs K12 will bring SMT to its custom ARMv8 ISA based designs, so things will heat up for the custom ARM designs come 2016.
Well, since the rumors (like
Well, since the rumors (like this post) are that the next SHIELD portable will have the Tegra X1, maybe the Tablet will have Tegra X1 with Denver CPU?
But who knows, last I heard was that Denver was powerful, but NVIDIA was having problems shrinking it to 14nm, and/or getting much more performance out of it.
And Intel did not have
And Intel did not have problems with a process node shrink, or AMD and Nvidia and their fab partners. Nvidia did not have a 14nm process to work with this time around, and Nvidia could have done what AMD had done with AMD’s GPU design libraries applied to its Carrizo CPU core, it’s not that Nvidia itself does not have GPU design libraries it could have applied to its Denver cores. The Denver core is very a powerful custom implementation of a custom microarchitecture designed to run the ARMv8a ISA, with execution resources and an IPC that beat Apple’s A7/A8(?) IPC. Just go read Nvidia’s presentation at the Hot Chips Symposium, and see for yourself. Hopefully Nvidia will bring back the Denver CPU core and some HBM, because It will have to compete with Jim Keller’s K12 custom ARMv8 APU design with SMT, and HSA compliant Graphics.
With the Khronos group’s Vulkan having much better multiprocessor/multi processor thread capabilities to go along with better HSA capable versions of CPU hardware, and the gaming/other software stack, to better take advantage of both graphics and GPGPU, the android, and hopefully full Linux based tablet market will have a very powerful graphics/GPGPU API/LLVM to enable tablets to be perform more work with the resources that they have. The Vulkan API/LLVM will sure come in handy for any tablet with multi CPU cores, and AMD’s K12 will be something that Both Nvidia, and Apple will have to look out for in the future. The Vulkan API itself is very HSA in the way it can utilize a GPU for both graphics, and general purpose compute on any SOC that makes use of the Vulkan API/LLVM.
Intel didn’t have problems
Intel didn’t have problems with the node shrink because their desktop processors are only 1.4 billion transistors. Compare that with my GTX 780ti for example, which has over 7 billion. Graphics chips are far more complex and difficult to manufacture, and since cutting-edge fabrication processes initially have a much higher cost per wafer due to low yields… it’s just not economically feasible to manufacture such complex chips right when the fab starts taking orders. Both AMD and nVidia made this mistake in the past and it caused high prices, low availability, and chips that ran a lot hotter than a later refresh on the same node did.
Anonymous- K12 wont be available till 2016 and if history is to be any indicator.. it will be more power-hungry than nVidia’s chips.
Has there been any further
Has there been any further news or leaks regarding an X1 tablet?