backtop


Print 48 comment(s) - last by someguy123.. on Dec 22 at 6:52 PM

Details on the next generation Tegra 4 processor, codenamed "Wayne", leak

NVIDIA has done pretty well on the mobile market with its Tegra 3 processor, and it should be of no surprise to anyone that NVIDIA has been working on the next generation Tegra processor. The company even provided some rough details early last year.
 
However, some detailed information on the next-generation 28nm Tegra 4 leaked today, and it promises six times the power of Tegra 3. Tegra 4 appears to be, according to a leaked slide, a 4+1 quad-core design similar to that of the current Tegra 3. The Cortex-A15-based Tegra 4 has 72 graphics cores and supports dual channel memory. Tegra 4 will also support resolutions of 2560 x 1440 for encode and decode and promises very low power consumption.

Other supported features will include the USB 3.0, making this the first NVIDIA chipset to support the new and faster USB standard.

With CES 2013 kicking off in mere weeks, we should have significantly more details coming in the not-too-distant future. 

Source: Engadget



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

RE: Still 32-bit
By Shadowself on 12/18/2012 1:18:37 PM , Rating: 3
quote:
as long as performance doesn't suffer
This is the crux of the problem. Running the current versions of MS Office applications on 5+ year old systems can be a royal pain.

ARM and similar SOCs are not equivalent to *today's* mid to high end desktop processors. Within five years these SOCs certainly will be equivalent to today's mid to high end (maybe not the very high end Xeons, e.g.). However, software will get more compute hungry in that time. Both users want more functionality and developers want to insert more functionality -- not to mention marketing wanting to scream about the latest and greatest new features.

Eventually (my guess is within 10 years, 15 years AT MOST) the SOCs like ARM will be at a state that very few people will have dedicated boxes sitting on (or under) their desks. It's just not the near future. Certainly not within the next five years.

The transition has started. It will be taking a bigger step next year with the new Surface tablet ships with a "full" version of Office and a "full" version ships for iOS. It's just that the transition will take at least another decade to take hold.

This in not unlike the very late 1970s leading into the early 1990s with the switch from "big iron" to networks of "desktops". (For the sake of simplifying this discussion, I'm lumping large mainframes through what were called "minicomputers" like the VAXes into the same category.) "Desktops" got their nose under the tent in the late 1970s just like the current "bring your own device" trend is getting tablets into enterprises today. It was well more than a decade (into the 1990s) before big iron really, really started to die -- as was shown by consolidation of such entities as DEC, Data General, Wang, etc. into other companies and IBM transforming into a services company. Personal usage in the home moved a bit faster om the 80s and early 90s simply because people didn't have mainframes in their homes and therefore had nothing to replace. Now desktops and laptops are firmly entrenched in most homes.

Sure, they will be replaced by a combination of smartphones, tablets and ultrabooks. But that entrenchment will take time to replace especially when you consider evolving/expanding requirements/needs as mentioned above.

My best guess (100% pure speculation) will be that by 2025 the "standard" will be high performance tablets (of various sizes) with "smart" TVs in the home and maybe even office. Moderate resolution (I'm avoiding the 'retina' label) on the tablet when not connected to the smart TV, but with the tablet supporting (with enough performance capability) to drive a 4K smart TV when connected -- and that connection will be optionally either wired or wireless.


RE: Still 32-bit
By othercents on 12/18/2012 2:50:31 PM , Rating: 2
quote:
ARM and similar SOCs are not equivalent to *today's* mid to high end desktop processors.
Yes, but we were not talking about today's ARM processors. Plus the speculation is that ARM x64 will be a significant jump in performance that hopefully fit into the low-end PC arena.

The market is definitely being pushed from both ends requiring more performance with less heat and power costs. There hasn't been processor to fit that mark yet, however the ARM processor is close without having all the legacy code. I speculate that we will see a performance jump next year in ARM that will make it the perfect processor for Windows RT devices and will meet the demands for most people.

Face it.. now almost everyone and their dogs has a tablet (my dog has one, she watches snoopy) and the number of cycles coming from the "legacy" desktops and laptops for those families has dropped drastically. If I could get the web apps my daughter uses for school to work on android she wouldn't ever need to use her desktop (however parental controls suck, so she is banned from using the tablet).

Other


RE: Still 32-bit
By NellyFromMA on 12/19/2012 1:04:47 PM , Rating: 2
Intel will likely become sufficiently power efficient before ARM meets any current gen Intel performance benchmark. 64-bit isn't going to boost performance that much, unless I missed something when Intel did the same.

Yes, mobile devices are out and about certianly not going anywhere. Arm is great for light workloads, especially battery powered. ARM would have to revolutionize at a pace faster than Intel's slowest pace to win out in the server space though.

As for PC replacements, ya ARM could have a place. It's so cheap. But, there's a reason for that too..


RE: Still 32-bit
By someguy123 on 12/22/2012 6:52:45 PM , Rating: 2
If you're not talking about today's processors, why do you ignore advancements by other companies? AMD is getting its APU package down to less than 20 watts, and intel has already produced an Atom chip running similarly to an snapdragon core (though core count still limited) at similar draw without finfet.

The claims of gloom and doom over conventional x86 parts are highly exaggerated. The industry isn't some kind of bubble where only one standard can succeed without the entire business popping. 10 years ago the death of x86 would've made significantly more sense considering the node sizes, but now its becoming less and less influential to draw.


"This is from the DailyTech.com. It's a science website." -- Rush Limbaugh














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki