Active Topics

 


Reply
Thread Tools
Capt'n Corrupt's Avatar
Posts: 3,524 | Thanked: 2,958 times | Joined on Oct 2007 @ Delta Quadrant
#71
This is a very interesting technology Zii Labs zms20.

http://www.ziilabs.com/products/processors/zms20.aspx
VIDEO: http://www.youtube.com/watch?v=CvZwSdQHGZQ

What's most interesting about this chip is that it's the first mobile chip (so far as I can tell) that's fully programmable by Open CL. This should enable developers to offload much work onto the GPU for computation for lower power and faster computation.

I would be interested in seeing some benchmarks of this SoC.
 

The Following User Says Thank You to Capt'n Corrupt For This Useful Post:
Capt'n Corrupt's Avatar
Posts: 3,524 | Thanked: 2,958 times | Joined on Oct 2007 @ Delta Quadrant
#72
Intel is not going down without a fight! The new ATOM detailed:

http://www.anandtech.com/show/4333/i...m-architecture

It seems that Intel is looking to matching the performance of the upcoming A15, and with a GPU to boot.

It's heating up!
 
Capt'n Corrupt's Avatar
Posts: 3,524 | Thanked: 2,958 times | Joined on Oct 2007 @ Delta Quadrant
#73
If there's one rule of thumb, it's don't count intel out of the game early:

Intel is shrinking it's process to an anorexic 14nm, and heavily targeting Atom:

http://www.anandtech.com/show/4345/i...t-atom-in-2014

What they're expecting is

I have an idea, why not drop x86 for something more space efficient before it's too late? Even if only a contingency, a more efficient instruction set would put serious heat on ARM. I expect the trick would be to fill in the software gap by providing compilers for linux.
 
Posts: 1,746 | Thanked: 2,100 times | Joined on Sep 2009
#74
Originally Posted by Capt'n Corrupt View Post
I have an idea, why not drop x86 for something more space efficient before it's too late? Even if only a contingency, a more efficient instruction set would put serious heat on ARM. I expect the trick would be to fill in the software gap by providing compilers for linux.
Is this you saying as much, or a broken quote?

The problem with dropping x86 and moving on to something else is scaring the hell out of all the vendors suddenly faced with an entirely new ISA. Even for Intel that would be so extremely risky they'd probably be better off (ROI wise) just re-licensing ARM.

Not that I don't think it'd be cool to see Intel charge out of the gate with a new ISA designed around their newer processes (though they might at the lower levels of x86 these days,) but I think that even Intel would be crazy to try.
 

The Following User Says Thank You to wmarone For This Useful Post:
Capt'n Corrupt's Avatar
Posts: 3,524 | Thanked: 2,958 times | Joined on Oct 2007 @ Delta Quadrant
#75
Originally Posted by wmarone View Post
Is this you saying as much, or a broken quote?

The problem with dropping x86 and moving on to something else is scaring the hell out of all the vendors suddenly faced with an entirely new ISA. Even for Intel that would be so extremely risky they'd probably be better off (ROI wise) just re-licensing ARM.

Not that I don't think it'd be cool to see Intel charge out of the gate with a new ISA designed around their newer processes (though they might at the lower levels of x86 these days,) but I think that even Intel would be crazy to try.
It's me, and I completely agree. For Intel to push a new architecture would be extremely nerve racking, unless positioned it in a way that didn't make vendors nervous. This is less a computer science/engineering problem and more a marketing problem.

However, I feel that they're resisting change at the peril of a swath of their contracts. Now that the market is paying more attention to ARM SoCs in the mobile space, the laptop market is taking a hit, and will increasingly do so when cortex A15 launches. Apple, is rumored to switch its macbook line to ARM SoCs in light of their performance/power characteristics.

Some (probably terrible) ideas:
- include an x86 translator in code on an instruction set that will run it well
- strip x86 down to a 'fundamental' subset of instructions
- support the new arch as part of an augmented instruction set for upcoming high-power CPUs.
- leave x86, but focus more on GPGPU hardware, and more importantly software, as a means of evening the playing field.

The market is in flux right now with the proliferation of new devices, and as such there is a window to be creative. It could be that intel sees the futility of a new instruction set with the shrinking of dies from a performance-to-power standpoint. Of course, ARM is about to put serious pressure on intel for the low-end market next year and if intel isn't ready, they will lose a bunch of the market.
 
Kangal's Avatar
Posts: 1,789 | Thanked: 1,699 times | Joined on Mar 2010
#76
Originally Posted by Capt'n Corrupt View Post
If there's one rule of thumb, it's don't count intel out of the game early:

Intel is shrinking it's process to an anorexic 14nm, and heavily targeting Atom:

http://www.anandtech.com/show/4345/i...t-atom-in-2014

What they're expecting is

I have an idea, why not drop x86 for something more space efficient before it's too late? Even if only a contingency, a more efficient instruction set would put serious heat on ARM. I expect the trick would be to fill in the software gap by providing compilers for linux.
Interesting read. But what really makes me feel bad is the speed at which this innovation is coming. Back in sometime like 2004, there were many limits to the processor architecture which were bypassed and allowed us to come this far.

But it is true when I tell you 14nm processors are available today, its just waiting in the labs for Microsoft to make Windows NEXT/8 and the market to adopt the 'Tablet' as the new personal computer. And then there's the investors who want slow progression so they can squeeze out each dime.

What really will be interesting is when they hit 10nm and 6nm, when the laws of physics will take a huge toll on the microprocessor. In fact, 6nm would be really unstable and hard to manufacture. I'm guessing Intel is looking into new methods, probably laser-based, to target the same energy consumption but speed up the process.

But let's brainstorm what such technology could be used for?
qHD-3D in your pocket with 48hours playback? or perhaps they're building the xPhone (just youtube it, heh).
 

The Following User Says Thank You to Kangal For This Useful Post:
Capt'n Corrupt's Avatar
Posts: 3,524 | Thanked: 2,958 times | Joined on Oct 2007 @ Delta Quadrant
#77
Very, very interesting.

It seems that Windows 8 ARM will not be backwards compatible with non-ARM variants.


http://www.engadget.com/2011/05/18/i...ffer-no-compat

In the meantime Intel is ramping up to flood the market with some 35 tablets in the coming year!
http://www.zdnet.com/blog/computers/...the-works/5488

It's a crowded market, and it's under flux. It will be interesting to see what happens.
 

The Following User Says Thank You to Capt'n Corrupt For This Useful Post:
Kangal's Avatar
Posts: 1,789 | Thanked: 1,699 times | Joined on Mar 2010
#78
Originally Posted by Capt'n Corrupt View Post
Very, very interesting.

It seems that Windows 8 ARM will not be backwards compatible with non-ARM variants.


http://www.engadget.com/2011/05/18/i...ffer-no-compat

In the meantime Intel is ramping up to flood the market with some 35 tablets in the coming year!
http://www.zdnet.com/blog/computers/...the-works/5488

It's a crowded market, and it's under flux. It will be interesting to see what happens.
Hey my article got approved in engdaget, yay!
Thanks CC, your the one who pointed it out to me.
I'm not really reading engadget as often since it's jumped the shark. The old crew of engadget got sacked by AOL, so they made their own new gadget-blog its called thisismynext.com ... its quite good.

Back on topic:
ARM-based Windows 8 will support Windows 8 programs.
x86-based Windows 8 will support Vista, Windows 7, and Windows 8 programs.
I expect the "Ultimate" version of x86-based Windows 8 to also support Windows XP for legacy code, as history shows us.
 

The Following User Says Thank You to Kangal For This Useful Post:
Posts: 1,746 | Thanked: 2,100 times | Joined on Sep 2009
#79
Originally Posted by Kangal View Post
ARM-based Windows 8 will support Windows 8 programs.
x86-based Windows 8 will support Vista, Windows 7, and Windows 8 programs.
Looks like Microsoft will be pushing .NET based languages and the CLR very, very hard. I wonder if they'll eliminate Visual C++ and the public face of their Intel and ARM compilers entirely.
 

The Following User Says Thank You to wmarone For This Useful Post:
Capt'n Corrupt's Avatar
Posts: 3,524 | Thanked: 2,958 times | Joined on Oct 2007 @ Delta Quadrant
#80
Originally Posted by wmarone View Post
Looks like Microsoft will be pushing .NET based languages and the CLR very, very hard. I wonder if they'll eliminate Visual C++ and the public face of their Intel and ARM compilers entirely.
That would certainly be an interesting strategy, and a solution.

I wonder what the metrics are like on C# vs Visual C++ proliferation?

I bet that MS is going to open up an 'App Store' in the same vane as Apple's OSX App Store, or the Chrome Web store, to draw attention to apps that can run on the new system.

I also forsee them pushing Silverlight to high levels of performance and using this as a vehicle to push new titles to the platform and to port existing ones.

All speculation.
 
Reply

Tags
mobile, mobile soc, rainbows, revolutionizing, unicorns, zmobility


 
Forum Jump


All times are GMT. The time now is 11:06.