facebook rss twitter

NVIDIA reveals fifth CPU core for Kal-El

by Scott Bicheno on 20 September 2011, 18:12

Tags: NVIDIA (NASDAQ:NVDA)

Quick Link: HEXUS.net/qa7dd

Add to My Vault: x

Side-kick core

As we, presumably, get close to NVIDIA's quad-core Tegra SoC - codenamed Kal-El - finally making a public appearance, NVIDIA is starting to trickle out a few more details about what gives the chip its mojo.

The four ARM Cortex A9 CPU cores will, in theory, give Kal-El a performance advantage over its competitors, none of which are expected to hit quad-core until well into next year. How much of an advantage is hotly debated, not least by Qualcomm, with many questioning whether four cores can be properly utilized and what the power implications are.

Unperturbed, NVIDIA has now announced there is a fifth CPU core in Kal-El, which is also a Cortex A9, but is manufactured on a low power silicon process. This means it's optimised to work on simpler tasks while using very little power in so doing, in fact its maximum clock is limited to 500 MHz.

NVIDIA is calling this the companion core, and goes into much greater detail on the matter in a white-paper that you can read for yourself here. In essence it enables an ‘active standby mode', in which a device sporting this chip is still able to do background tasks, such as downloading email and alerts, while in a very low power state.

In the white-paper NVIDIA also discloses a new type of technology called Variable Symmetric Multiprocessing (vSMP), of which the companion core is central. Apparently this technology makes the companion core platform agnostic, meaning software automatically takes advantage of it without the need for special coding, which is nice.

The two images below convey the basic concept, should you not be in a very white-paper place right now. If you are, you might want to double up and read this one on why quad-core is great for mobile devices. This was no doubt produced to counter the FUD being disseminated by the likes of Qualcomm on the matter.

 

 

 



HEXUS Forums :: 6 Comments

Login with Forum Account

Don't have an account? Register today!
Bigger is obviously better, but if this is such a great idea why has it not been implemented before? My guess is it's a huge cost in actual silicon area for small benefits in speed
Only Nvidia could decide that the best way to make a chip more efficient is by making it bigger.
So, basically what they are saying is the things going to be late…
And slow. At best it will just about be as fast as the current Mali 400 from ARM.
'[GSV
Trig;2125365']So, basically what they are saying is the things going to be late…
Where does it say that - I got the impression that the only reason NVidia are releasing these details are because Kal-El is a lot closer to launch. (That, and to upset Qualcomm of course).
EzyRyder
And slow. At best it will just about be as fast as the current Mali 400 from ARM.
Not necessarily, have you seen the discussion at http://forum.xda-developers.com/showthread.php?t=977412. That said, I'm no Tegra loyalist - although the T2-based tablet I've got is fast enough for what I want to do. Which, in the end, is what matters to me - rather than benchmarks.

Got to say that this vSMP looks like an interesting idea - if they've done it “right” then it should give some noticable battery benefits. I'll reserve judgement until I actually get my hands on a tablet with that in place (can't see a quad core phone being much of an advantage - I think QC needs the larger form factor).