Raja Koduri, Intel’s general manager for architecture, graphics and software, shows off a next-generation Xe graphics chip.
Intel
If you need a new laptop, you may want to wait until later this year when Intel’s latest chips are available. Tiger Lake chips promise faster speed, better graphics and longer battery life.
Formally called Intel’s 11th generation Core processors, the chips will make more realistic explosions in Call of Duty and allow longer PowerPoint slide decks before running out of a PC’s battery. Intel unveiled the Tiger Lake line earlier this week on its Architecture Day, along with a range of the chipmaker’s other engineering feats.
Intel has driven a lot on Tiger Lake, the company’s first major new chip design since last year. The Santa Clara, California, chip titan has struggled in the past several years. Apple has started releasing Intel chips from Macs, while Microsoft rivals like AMD and Qualcomm. Intel stumbled years late after its current chipmaking process and announced another major delay in July for its next-generation manufacturing process.
On Architecture Day, Tiger Lake shared center stage with Intel’s new Xe graphics chip technology that will accelerate basic graphics and high-end gaming. Intel would not share specific speedboost figures for Tiger Lake or Xe, one variant of which is built directly into Tiger Lake. But a performance chart sets a new tone for discussions about Intel’s products and prospects.
Tiger Lake speed boost
Intel’s performance features allowed Tiger Lake to expand its current Ice Lake predecessors through a combination of enhancements, including a new core chip design called Willow Cove and an updated manufacturing process now called SuperFin.
The Willow Cove core in the heart of Intel’s Tiger Lake chips can rotate faster and more efficiently than the Sunny Cove cores in today’s Ice Lake chips.
Intel
“I feel better about his future,” Patrick Moorhead, an analyst at Moor Insights and Research, said of the chip giant. When it comes to chip architecture, Intel gave the world reasons to believe that “it could be on top again.”
Tiger Lake chips will give PC makers the option of a chip that is faster at a certain level of energy consumption than that which has a better battery life for a certain speed, said Boyd Phelps, Intel’s vice president of silicon technology. For the highest tasks such as gaming, whereby power consumption is generally not a major constraint, Intel may also bring Tiger Lake’s clock speed even higher.
Other capacities of Tiger Lake
Tiger Lake will have built-in Thunderbolt 4 en USB 4 support for inserting fast peripherals into your PC, new security protection called CET (technology for control-stream maintenance), better AI circuit for voice control and speech-to-text conversion, and support for more and higher resolution displays.
In 2021, Intel plans to launch a successor called Alder Lake that uses a hybrid approach that marries two types of processing cores. Those cores are Golden Cove, a more powerful successor to Tiger Lake’s Willow Cove, and Gracemont, a smaller design optimized for lower power consumption.
The Alder Lake chips will be built with an improvement of the SuperFin production process with the placeholder Enhanced SuperFin. That should bring another speed boost in 2021.
Better graphics with Xe
Much of Tiger Lake’s speed boost comes from better graphics provided by Intel’s Xe design. For years, Intel built basic graphics skills into its chips, but serious gamers and others who needed more graphics power relied on separate graphics chips from Nvidia or AMD.
For next-gen PCs, Xe will be built in Tiger Lake, but it’s much faster than previously integrated graphics, judging by game demos that ran at much higher frame rates than on current Ice Lake chips. Intel will offer a variety of Xe graphics chips apart from Tiger Lake, including this DG1 shipping this year and a newly announced high-performance variant to thank gamers in 2021. That model will include features such as ray tracing that Nvidia pioneered for the gaming brand.
“It’s definitely a serious competitor to AMD and Nvidia,” said Tirias Research analyst Kevin Krewell.
Mix and Match
Making small on-off switches called transistors is key to chipmaking, but there are many areas where engineering can pay dividends. For example, Intel credits improvements in the path of metal data directly above the transistors for some of Tiger Lake’s speed improvements.
Intel has designed a variety of chip packaging techniques to allow it to mix and match different components to quickly adapt chips to specific uses.
Intel
Intel also demonstrated its capabilities in chip packaging, a rapidly evolving and highly competitive technology that allows processor makers to link two silicon chips side-by-side or on top of each other. High-speed data links allow them to function as a unit, and Intel’s have received new enhancements to those data links.
The company’s Lakefield chip exemplifies this technology, allowing Intel to store memory, data input, processing and other features in one package. But Intel will spread the idea to other chips as it improves. The Foveros technology used to stack Lakefield elements currently provides a maximum of 1,600 data links per square millimeter, a density that is expected to increase to 10,000, although Intel did not say when.
New packaging options and SuperFin manufacturing allow Intel to improve its competitiveness, said analyst Pierre Ferragu of New Street Research and colleagues in a research note Friday. A good example is Intel’s Rambo Cache – the marriage of a graphics chip with separate large and fast memory for better performance. It could “provide a breakthrough competitive advantage for Intel,” the report said.
Intel’s production gloves
Intel is stepping up its manufacturing process with products outside Tiger Lake. Intel reported $ 5 billion in profit on $ 20 billion in revenue for its most recent quarter, but it has been held back by rigid development and manufacturing.
Before, chip designs were married to the manufacturing process, a “tick-tock” cycle that improved the chip’s architecture one year and then its components the next. Now Intel is moving to a “transistor resilient design” that is no longer tied to a particular manufacturing process.
This will provide more flexibility, including a better ability to outsource production to competing chip stiffers such as Taiwan Semiconductor Manufacturing Corp. (TSMC).
The future of Intel depends much more than just on their PC chips. That future includes Xeon chips for the servers that pack data centers, run by companies like Google and Baidu, customizable processors used in network equipment, optical network sharing and memory chips used in SSDs and other storage technology.
At Architecture Day, Intel demonstrated its ability to apply a range of manufacturing tricks to the full range of its products. And in a new twist, it has added more emphasis to its software effort – specifically how it uses a software layer to bridge the programming divisions between its chip variants.
