Affiliate Disclosure
If you buy through our links, we may get a commission. Read our ethics policy.

Apple's A5 processor could pave way for larger chip trend

Despite sharing similar specifications, Apple's bigger A5 System on a Chip processor used in the iPad 2 has posted better performance than Nvidia's Tegra 2 chip, causing one analyst to believe that Apple is leading a trend toward larger, faster mobile processors.

Responding to another analyst's claims that, from a semiconductor perspective, "Apple is falling behind - way behind," analyst Didier Scemama of RBS issued a report on Friday defending the A5's larger footprint, Barrons reports.

Given that the A5 is 122 square millimeters in dimension, compared to the Tegra 2's 49 square millimeters, Scemama finds the difference "surprising" because both processors have similar specifications: "dual core Cortex-A9 ARM CPU, high-end GPU, HD video, high-end audio and imaging."

That size difference is partly attributed to the 45-nanometer process used by Samsung to build the A5, as Nvidia used Taiwan Semiconductor's smaller 40-nanometer process.

According to Scemama, "the major difference in specification between the chips is that Apple uses Imagination Technologies‘ PowerVR SGX543 dual-core GPU in the A5 and Nvidia uses its own GPU called GeForce.”

The PowerVR SGX543, which AppleInsider was first to report would be included in the iPad 2, provides a dramatic boost in graphics power. Though Apple claims a 9x graphics speed bump from the original iPad to the iPad 2, real world tests suggest the second-generation tablet has roughly 5x the graphics power of its predecessor.

The analyst cited OpenGL benchmark tests that show "that the A5 beat Tegra 2 handsomely." When tested against the Tegra 2-powered Motorola Xoom tablet, the iPad 2 performed 3.7x better than the Xoom.

The Tegra 2's slower performance may be a result of cost limitations that Nvidia faces with its customers, the report noted. Scemama believes that Nvidia needs to keep the Tegra 2 chip in the $15 to $20 range. In contrast, UBM TechInsight estimates that Apple's new chip costs $25 to make, though the research firm speculates that high volume, generated by building A5 into a range of products such as the iPhone 5 and next-generation iPod touch, could bring the costs down to $15.

"Apple can afford to use a larger chip, which delivers substantially better performance, as it pays a fraction of the price it would pay if it used an off-the-shelf solution with similar performance," said Scemama.

Scemama speculates that ARM and Imagination Technologies would benefit from an ensuing spec race as competitors attempt to match Apple in terms of performance.

"Given the current gap in performance between the A5 from Apple and other competing chips, we believe it is possible higher performance (and therefore potentially larger die size) processors may be required by OEMs. This may explain the current race among ARM partners to launch higher performance processors in terms of CPU but also, crucially, GPU," he noted.

Nvidia's price limits could stem from struggles by rival tablet makers to keep costs down while remaining competitive with the iPad on features and performance.

Last year, Apple CEO Steve Jobs noted that the company's "potential competitors [in tablets] are having a tough time coming close to iPad's pricing. iPad incorporates everything we've learned about building high value products. We create our own A4 chip, software, battery chemistry, enclosure, everything. This results in an incredible product at a great price. The proof will be in the pricing of our competitors' products, which will offer less for more. "

Jobs' words rang true this spring when Motorola revealed a $799 price tag for the entry level Xoom, despite the company's use of the cheaper Tegra 2 chip. Shortly after Apple unveiled the iPad 2, Motorola announced a $599 Wi-Fi only version of the Xoom.



120 Comments

tipoo 15 Years · 1124 comments

Yep. Remember, to get the surface area of a transistor, figures like 45nm are squared, so a difference of 5nm is larger than it seems. Plus, the A5 uses the MP2 version of the 543 GPU, which obviously would take more room, seeing as its a whole 'nother graphics core. Not really surprising that its bigger, nor that it does better in raw graphics benchmarks.

I know the 543 was built from the ground up to be scalable (up to 16, I think, and the PSP2/NGP will use the 4 core version), I wonder if Nvidia's equivalent GPU is like that? If so, they could catch up soon.

ascii 20 Years · 5930 comments

The important thing is how flat/thin it is, right? Otherwise if you can make it cheaper by spreading it out, and you have the room anyway, go for it.

sheff 16 Years · 1407 comments

Apple made a great move by building their own chip on top of an already fast arm architecture. This means the performance will stay competitive, but competition will gain little in adopting newer versions as direct comparison based on chip number or even gigahertz impossible.

sheff 16 Years · 1407 comments

Quote:
Originally Posted by ascii

The important thing is how flat/thin it is, right? Otherwise if you can make it cheaper by spreading it out, and you have the room anyway, go for it.

Agreed. Not a big deal for iPad, which at least in version 1 had room to spare, but an iPod touch might have a problem with larger design. Also heat and ventilation are a concern though a bigger surface area may make cooling easier.

rhyde 16 Years · 293 comments

Quote:
Originally Posted by tipoo

Yep. Remember, to get the surface area of a transistor, figures like 45nm are squared, so a difference of 5nm is larger than it seems. Plus, the A5 uses the MP2 version of the 543 GPU, which obviously would take more room, seeing as its a whole 'nother graphics core. Not really surprising that its bigger, nor that it does better in raw graphics benchmarks.

I know the 543 was built from the ground up to be scalable (up to 16, I think, and the PSP2/NGP will use the 4 core version), I wonder if Nvidia's equivalent GPU is like that? If so, they could catch up soon.

Don't forget the layers. These chips are also three-dimensional.
Does the A5 have more layers than the Nvidia chip?