During a conference call with press and financial analysts, Intel boss Brian Krzanich said his company has hit a series of problems while attempting to upgrade its chip fabrication systems. As a result, the company will not focus on making its 2016 chips smaller; rather, it will seek to boost performance by different means.
The admission comes days after competitor IBM announced it had solved problem it faced that would have otherwise stopped it too from fabricating smaller components.
Intel’s goal for 2016 was to manufacture components 10nm in size, a sizeable decrease from its current roster of components, which feature components as small as 14nm. Further decreasing the size a full 4nm involves complicated upgrades to the fabrication system as a whole.
“The lithography is continuing to get more difficult as you try and scale, and the number of multi-pattern steps you have to do is increasing,” Krzanich explained.
Earlier, Intel said that it planned to phase out its 14nm components in favor of 10nm parts. Per Krzanich, this may have to wait until 2017 now. This decision delays the company’s two-step “tick tock” system it has been using for years to continuously improve processor power.
For those unfamiliar with the “tick tock” system, it’s a process Intel implemented whereby the company alternates changes to the gross layout of a chip (tock) with refinements to fabrication of individual components (tick).
It is, essentially, a working implementation of Moore’s Law, which defines steady, year-on-year improvement in processor power. Formulated by Intel co-founder Gordon Moore five decades ago, it suggests chip power could double every 12 months.
In the year’s past, this Law was proven successful time and again by shrinking the millions of transistors on a chip.
Recently, though, the Law’s estimation of how long it takes to refine production systems to make it economic to produce faster chips has become a bit stretched, with improvements now expected once every 24 months.
Per Krzanich’s admission, the Law may be close to coming to an end, though it’s not there yet — as mentioned earlier, IBM announced just a few days ago that it had figured out how to reliably produce chips outfitted with 10nm components. The company did admit, however, that these techniques had so far only been proven to work in the laboratory, and that it still had yet to see implementation within an actual large-scale chip fabrication plant.
Worth noting — Krzanich did not provide details on how, exactly, Intel would drive the improved power of its 14nm chips. Instead, this information was simply provided as the company’s temporary solution until it figured out its fabrication problems.
Via BBC
Advertisement
Learn more about Intel
