Intel Corp. (INTC) has a long way to go to be a strong competitor to the coalition of ARM Holdings Plc (LON:ARM) licensees who today dominate the smartphone and tablet industries. The company, long the dominant force in the personal computer CPU market, is working diligently to catch up. Its first smartphone effort -- Medfield was surprisingly capable, although not giving Intel a clear win over rival ARM-based offerings.
I. Tri-Gate is Crucial to Intel's Mobile Ambition
But Intel has a trick up its sleeve for the next round -- 22 nanometer tri-gate transistors. The unique 3D transistor technology employs a special fin-shaped structure to eliminate leakage at the gate, a key source of inefficiency and inaccuracy, as transistors get smaller.
Kaizad Mistry, a vice president in Intel's manufacturing group, comments to The Wall Street Journal, "There has been some press about how TriGate may not be suited for SoCs. I find that frankly quite baffling."
Intel first revealed its tri-gate transistors with Ivy Bridge, its latest personal computer CPU line, which launched earlier this year. Ivy Bridge is currently the leader in single-threaded performance in that market.
For Intel, a company who invests vast sums in process improvement, die shrinks allow the company to yield continuous improvement, while novel technologies like tri-gate allow the company to offer a little something extra. In the personal computer CPU market where Intel has long dominated, such extras are less game changing, but in the mobile market where it's struggled they're critical. Intel needs all the help it can get as its monolithic chips, while powerful, have traditionally been less power-efficient on the mobile front than ARM processors.
II. Critics Not Convinced Technology Will be Enough
Intel's critics are quick to point out that the company promised 22 nm smartphone chips in H1 2013, but has delayed the launch roughly 6 months to sometime in H2 2013.
But Intel looked to silence doubters by releasing a paper at the 2012 International Electron Devices Meeting (IEDM) in San Francisco this week. The paper details Intel's work to port the process to the mobile (Atom) chip line. The paper claims that the 3D transistors will bump processor speeds by between 22 and 65 percent, while also offering power savings over 32 nm Atom chips.
But the critics are as noisy as ever. Some say Intel's chips are too expensive to gain ground, while others call the gains of the company's novel 3D transistor technology okay, but uninspired (taking into account that part of the aforementioned performance/power gains come from the die shrink itself, not the 3D design).
Asen Asenov, a University of Glasgow electrical engineering processor and head of technology consulting firm Gold Standard Simulations Ltd., comments, "Honestly speaking, this isn't a big improvement."
Scott Thompson, who once served as an Intel research fellow and today is chief technology officer of chip technology firm SuVolta Inc. comments, "I think Intel has the best engineering team out there. The problem with where they are going is their technology is just not cost-effective for the mobile space."
Atom chips typically start at around $40 USD per unit, while ARM-based chips are priced at between $5 and $20 USD.
Source: WSJ
No comments:
Post a Comment