The big problem with Intel’s processor design

Intel recently unveiled its ninth generation of Intel Core desktop processors, which offer increased performance over 8th-gen chips in addition to increased core counts and clock speeds.

This line-up of processors includes the manufacturer’s new Core i9-9900K, which boasts eight physical cores and sixteen threads, and is labelled by Intel as the “best processor for gaming” on the market.

While these chips are becoming increasingly more powerful, Intel has been forced to change its development strategy due to difficulties with its previous method of regularly reducing transistor size.

Intel’s 9th-generation Core chips may be faster than the previous generation, but they are still built on the same 14nm manufacturing process introduced in its 2014 Broadwell processors.

The difficulty of moving to a 10nm process node has forced the company to look in other directions for performance improvements.

Reaching for 10nm

Historically, Intel used a tick-tock model to plan out the development of its CPU technology.

Using this model, the company alternated every year between improving the architecture of the previous generation and moving to a smaller manufacturing node.

This model ceased to be feasible once Intel moved to the 14nm manufacturing process, however, as it becomes more difficult to reduce the size of the transistors in a chip while maintaining stable function.

For this reason, Intel began to conduct extra refinements of its 14nm technology; although, without increasing the density of transistors per square centimetre, there is only so much that can be done to boost performance.

Intel originally meant to release its Cannon Lake microarchitecture based on a 10nm process node in 2016, but the technology has constantly been pushed back year after year, and there is still no confirmed release date for a stable 10nm Core desktop processor with an integrated GPU.

In the meantime, Intel has been forced to deliver performance improvements on an old manufacturing process without shrinking the die size.

Refinement of 14nm

Intel’s latest Core desktop processors are built on the manufacturer’s 14nm CPU microarchitecture, which has seen countless revisions and is heavily optimised to deliver expected performance improvements.

The manufacturer went in another direction to provide acceptable performance increases with its latest chips, however, cramming more cores into the CPU itself in an effort to boost chip performance.

Compared to the previous generation of Core processors, Intel’s 9th-gen chips boast an increased number of physical cores and threads.

However, their power consumption is also slightly higher, despite the optimisations made to the 14nm process node.

The biggest improvements that stem from moving to a smaller process node are increased power efficiency and transistor density, allowing for increased performance at the same power consumption as the previous generation, or lower power usage at the same performance level.

As Intel cannot take advantage of these options, its new chips show a slight increase in power consumption in addition to performance increases of around 10% compared to the previous generation.

For example, below is the listed TDP of equivalent 14nm chips as far back as Intel’s 6th-generation CPUs:

  • Core i5-6600K – 91W
  • Core i5-7600K – 91W
  • Core i5-8600K – 95W
  • Core i5-9600K – 95W

The increased number of cores does make them more suitable for certain applications, however, although most gamers would be fine with a quad-core processor instead of a hexa-core option.

Intel has managed to maintain the year-over-year performance increases expected from its generational improvements, but the question of whether it can continue to optimise its chips on the 14nm process node is unclear.

Now read: New Intel 9th-gen Core processors – South African pricing and details

Latest news

Partner Content

Show comments

Recommended

Share this article
The big problem with Intel’s processor design