TSMC 28 nm Technology in Volume Production

[btarunr]

TSMC today announced that its 28nm process is in volume production and production wafers have been shipped to customers. TSMC leads the foundry segment to achieve volume production at 28nm node.

TSMC’s 28nm process offering includes 28nm High Performance (28HP), 28nm High Performance Low Power (28HPL), 28nm Low Power (28LP), and 28nm High Performance Mobile Computing (28HPM). Among these technology offerings, 28HP, 28HPL and 28LP are all in volume production and 28HPM will be ready for production by the end of this year. The production-version design collateral of 28HPM has been distributed to most mobile computing customers for their product-design use.

The number of customer 28nm production tape outs has more than doubled as compared with that of 40nm. At 28nm, there are currently more than 80 customer product tape-outs. The TSMC 28nm process has surpassed the previous generation’s production ramps and product yield at the same point in time due to closer and earlier collaboration with customers. TSMC’s 28nm design ecosystem is available through its Open Innovation Platform®, with qualified EDA design tools and third-party IP ready for customer designs.

“Building on TSMC and Altera’s 18 years of established technology partnership, TSMC’s comprehensive 28-nm process offerings and Altera’s leading-edge FPGA technology complement each other perfectly, enabling us to uniquely tailor our 28-nm product portfolio to best meet our customers’ diverse design requirements,” said Vince Hu, Vice President of Product and Corporate Marketing at Altera Corporation. “In our 28nm generation, TSMC’s 28LP process fits the requirement of Cyclone V and Arria V families with the lowest power and costs, and we have utilized the 28HP process for the industry’s first delivered high-end 28nm FPGA, Stratix V with the highest performance and the lowest power in high-performance systems.”

"We applaud TSMC’s success bringing a robust 28nm process to market, and we look forward to leveraging the benefits of this new process when we ship our next-generation discrete graphics products,” said Matt Skynner, Corporate Vice President and General Manager, GPU Division, AMD. “The combination of AMD’s industry-leading graphics IP and TSMC’s manufacturing prowess will enable the next big leap in graphics performance with the parallel compute horsepower and power efficiency designed to meet the needs of even the most demanding gamer.”

“NVIDIA and TSMC have a history of delivering the most complex GPU architectures on state-of-the-art process nodes. This partnership has been among the industry’s most prolific, resulting in more than one billion GPUs shipped. Our close collaboration in developing 28nm processors will once again deliver the most energy-efficient GPUs and highest-performance graphics processors on the market,” said Jeff Fisher, Senior Vice President, GeForce Business Unit, NVIDIA.

“Qualcomm and TSMC have a long history of collaboration to bring to market the latest in mobile semiconductor technology on the most advanced silicon manufacturing processes, and we are excited to be introducing the first integrated smartphone processors at the 28nm node,” said Jim Clifford, senior vice president and general manager of operations at Qualcomm. “Most recently, Qualcomm’s work with TSMC yielded our Snapdragon™ S4 class of processors, including the Snapdragon S4 MSM8960™, a highly-integrated, dual-core SoC designed to reduce power in cutting-edge smartphones and tablets. The Snapdragon S4 class of processors are manufactured in TSMC’s highly sophisticated 28LP process, enabling Qualcomm to deliver the breakthrough combination of high performance and ultra low power to mobile devices.”

“Building our 7 Series FPGA and processing families on the 28nm HPL process in collaboration with TSMC is enabling Xilinx to lower static power by 50% while also increasing both raw performance and usable performance ,” said Vincent Tong, Senior Vice President, Worldwide Quality and New Product Introductions at Xilinx. “Xilinx has delivered several industry-firsts to our customers, including the first 28nm FPGA to begin shipments and the industry’s highest capacity FPGAs with the lowest power. To date, we’ve passed preliminary product qualification, our yields are on track and we’ve shipped 7 Series FPGAs to customers months ahead of our competition thanks to TSMC’s 28nm HPL process.”

“Being the first to 28nm volume production demonstrates TSMC’s leadership in technology and brings great value to our customers through design wins with competitive products,” said Jason Chen, TSMC Senior Vice President, Worldwide Sales and Marketing.

[n-ster]

I'm looking forward to 28nm GPUs and 22nm CPUs

[Riotpump]

Looking forward to those new Nvidia chips and Snapdragon products.

[Onenutleft]

ok, continues to save ...

[NC37]

Alright, lets see how many times TSMC will botch production on these new goodies!

[BorgOvermind]

Epic-lol:
"lowest power vs. power efficiency vs. the most energy-efficient vs. ultra low power"
So which one is the 'one' actually ?

nV is probably last place of all 4 since their latest and greatest were actually the most power-inefficient GPUs since quite a while.

[shb-]

This is all nice and all, but where is tech events with demos of 7970 and 680 : ] ? Seeing there is none, i guess those cards aint so near after all..

[Recus]

Originally Posted by BorgOvermind

Epic-lol:
"lowest power vs. power efficiency vs. the most energy-efficient vs. ultra low power"
So which one is the 'one' actually ?

What one? Low power for Altera, Qualcomm and Xilinx.

nV is probably last place of all 4 since their latest and greatest were actually the most power-inefficient GPUs since quite a while.

Go back to semiaccurate and cry.

[de.das.dude]

was this the one that AMD went to for new chips after GF failed them?

[Thefumigator]

Originally Posted by de.das.dude

was this the one that AMD went to for new chips after GF failed them?

Most likely AMD will have no other choice. You can thank GloFo for part of AMD failure.

[eidairaman1]

AMD stated they were going to switch to TSMC after GF messed up.

[de.das.dude]

Originally Posted by Thefumigator

Most likely AMD will have no other choice. You can thank GloFo for part of AMD failure.

we can say GloFo can GO FO? haha?? anyone?? .. okay...

[Thefumigator]

Originally Posted by de.das.dude

we can say GloFo can GO FO? haha?? anyone?? .. okay...

you won't believe this but I actually laughed

[Winston_008]

Originally Posted by Recus

What one? Low power for Altera, Qualcomm and Xilinx.



http://s3.postimage.org/p2p6os9wj/Capture.png

Go back to semiaccurate and cry. http://i56.tinypic.com/2e1sp3m.gif

Lol dont know where you got those results from,
During furmark- Hd 6970= 287w
gtx 580= 317 w
During 3dmark 03 nature average power = hd 6970 = 157w
gtx 580 = 197w
http://www.techpowerup.com/reviews/M...htning/21.html


Id say thats more accurate.

[xenocide]

Isn't the 580 more powerful though? Wouldn't that justify the increased power usage?

[n-ster]

Originally Posted by Winston_008

Lol dont know where you got those results from,
During furmark- Hd 6970= 287w
gtx 580= 317 w
During 3dmark 03 nature average power = hd 6970 = 157w
gtx 580 = 197w
http://www.techpowerup.com/reviews/M...htning/21.html

Id say thats more accurate.

http://www.techpowerup.com/reviews/M...htning/24.html

Check the performance per Watt. The GTX 580 is at 98% of the 6970 (2% worse), at 1200p it is 4% worse and at 2560x1600 it is 9% worse. Yes at 2560x1600 the numbers are less impressive, but for the rest it was BETTER than the 6970 IMO, as you have to usually sacr9ce a bit of perf/Watt when the card is more powerful (ie: 6950 vs 6970). Not to mention that the GTX 580 is the most powerful single GPU

Now I would say that the power efficiency of the GTX 580 is around the same as the 6970. I'd say THAT is more accurate

Originally Posted by xenocide

Isn't the 580 more powerful though? Wouldn't that justify the increased power usage?

Yes, look above

[BorgOvermind]

How about performance per die size ?
If you take the 5870 die and just make the chip it as big as the 580 just by adding more shaders you will beat a 580 by very far. But then, that's an engineering aberration and that's why it wasn't made that way. The very big die size is a recognized fail (as it was for Radeons some series ago and as it's now days for nV).

[Thefumigator]

Originally Posted by BorgOvermind

How about performance per die size ?
If you take the 5870 die and just make the chip it as big as the 580 just by adding more shaders you will beat a 580 by very far. But then, that's an engineering aberration and that's why it wasn't made that way. The very big die size is a recognized fail (as it was for Radeons some series ago and as it's now days for nV).

I recall the ATI 2000 series, the 2900XT power consumption being out of this world, it wasn't a bad card in performance but hopefully the 3000 series did improve in power consumption, not mentioning the great success of the 4000 series that came later.

Taking a look at how radeons improved from the 2000 to the 6000, the AMD graphics division has to be in the top of the "% amount of improvement per year" chart. They were really deep, in some stage everything looked... insurmountable... but finally got everything going fine, including drivers.

[Recus]

Originally Posted by BorgOvermind

adding more shaders you will beat a 580 by very far.

Do you know that HD 5870 has more than 1000 more shaders and still can't beat GTX580?

[theoneandonlymrk]

Originally Posted by Recus

Do you know that HD 5870 has more than 1000 more shaders and still can't beat GTX580?

their different types of shader used in slightly different ways and nvidia runs them twice as fast to get their performance, im not slateing either way tho i dont care what they do so long as it improves performance and both companys are doing that year on year, go competition, winner = us

[Thefumigator]

Originally Posted by theoneandonlymrk

their different types of shader used in slightly different ways and nvidia runs them twice as fast to get their performance, im not slateing either way tho i dont care what they do so long as it improves performance and both companys are doing that year on year, go competition, winner = us

I wish the difference between Intel and AMD was as balanced as the differences between AMD and Nvidia. Specially in time, because I don't feel AMD is crippled technologically, I find it is behind in time. Whenever Intel releases something, AMD has to catch up, and many times they are late. Lately it was like this

[eidairaman1]

I see a Stepping of BD coming out with this company and AMD Radeons produced aswell, GF needs to get their act together or they will be shut down