News & Analysis
Comment
Tommy55
The power depends on many factors......and those same factors decide frequency. ...
jg_
Hmm... "under typical operating conditions" are the key words here, so this is ...
TSMC pushes 28-nm Cortex-A9 to 3.1-GHz
Peter Clarke
5/3/2012 5:39 AM EDT
LONDON – Foundry Taiwan Semiconductor Manufacturing Co. Ltd. has announced it has made a dual-core ARM Cortex-A9 processor test chip in its 28-nm high-performance mobile CMOS process that it has run at a 3.1-GHz clock frequency under typical operating conditions.
TSMC (Hsinchu, Taiwan) said that this achievement demonstrates that the 28-nm HPM process is ready for use in a broad range of mobile, consumer and enterprise applications.
For mobile application signoff conditions the 28HPM process delivers clock frequencies in the range 1.5-GHz to 2.0-GHz, TSMC said. But for high performance requirements it can be run up to 3.1-GHz the company said. This could either for very short term use or where power consumption is less critical because a system is connected to a main supply of electricity.
The 28HPM implementation runs twice as fast as the 40-nm counterpart made by TSMC under the same operating conditions, according to Cliff Hou, vice president of R&D at TSMC.
Related links and articles:
ARM tips availability of 'Seahawk' A15 hard macro
ARM launches processor pack for big-little on TSMC 28-nm
Amlogic offers dual-core Cortex-A9 for consumer SoCs
|
|
|
|
|
Navigate to related information
yalanand
5/3/2012 10:50 PM EDT
This is great news. So when can we expect 20nm products from TSMC? Is TSMC close to tapeout products in 20nm process ?
Sign in to Reply
resistion
5/4/2012 1:24 AM EDT
So this is withoug overclocking?
Sign in to Reply
agk
5/4/2012 5:21 AM EDT
At 3.1Ghz this consumes more power and is use full with a system connected to power or for a short time usage with batteries. So much also the heat dissipated in this chip.The cooling system got to be efficient and i feel this is more suitable to desk top items rather than handheld appliances at 3.1Ghz.
Sign in to Reply
goafrit
5/4/2012 9:43 AM EDT
Any paper that supports your case. I know that at 28nm your static power become a huge issue. The dynamic power is a function of the high freq. That is what you pay for the high speed. Unless you can explain that it is not related to freq but process, I think this product could go into mobile devices.
Sign in to Reply
TingLu
5/4/2012 9:03 AM EDT
At what voltage and what power?
Sign in to Reply
goafrit
5/4/2012 9:41 AM EDT
Does that really matter? How can you calculate the power? I think it is the freq that is news worthy. The power depends on many factors...
Sign in to Reply
Tommy55
5/6/2012 9:56 PM EDT
The power depends on many factors......and those same factors decide frequency. How can that freq news be worthy without any mention of power ???
Sign in to Reply
Robotics Developer
5/4/2012 1:50 PM EDT
Finally, there is real speed in ARM land! I can't wait to see these on the market and on Small Board Computers for use in robotics. If nothing else they would be nice to have for those cpu sucking applications like: FP or Vision.
Sign in to Reply
jg_
5/6/2012 9:16 PM EDT
Hmm... "under typical operating conditions" are the key words here, so this is really an over-clocking test, with no margins included - just bragging right stuff, not real engineering.
Sign in to Reply