Buyer’s Guide

Computer Processors

The Central Processing Unit, or processor, defines the capabilities of a computer to execute instructions, however many technology professionals struggle to decipher the marketing terms use to define processor capability.

The objective of this buyer’s guide is to enable readers to make informed decisions when evaluating processor options. In particular, the guide has a key focus on unravelling the marketing terms and codenames used to promote computer processors.

This guide covers x-86 Intel and ARM processors for desktop and laptop computers in a business environment.

The guide concludes with purchasing recommendations.

Scope

This technical primer will focus on Intel and AMD who together serve 99% of the x-86 computer processor market, with Intel capturing 80-85% of Desktop and Mobile processors and AMD around 15-20%.

Processor manufactures overwhelmingly include their own integrated graphics processors with business laptops and desktops, as such this primer will not cover discrete graphics processors (a market dominated by Nvidia and AMD).

Processor capabilities

This primer focussed on marketing terms, however these are clearly linked to technical terms, as such some key terms and their relevance are defined below.

  • Cores – A Core is a processing unit within a CPU. Each Core can process instructions independently which improves the processors efficiency. CPUs have between 2 and 18 cores, while GPUs which are optimised to parallel processing for Graphics will have hundreds of cores.
  • Thread – A thread is a virtual version of a Core. Intel’s implementation is called hyper-threading. AMD’s implementation is called simultaneous multithreading (SMT).
  • Clock Speed – Clock speed is how many operations a processor can do in one second, measured in hertz (Hz). The greater the clock speed, the more the more powerful the processor in executing operations for a constant Instructions Per Second (IPS) value. IPS is continually improving due to innovations in processor design, thus a new generation processor may have a marginally better clock speed, but if the IPS has dramatically increased, the effective power of the processor has also dramatically increased.
  • Overclocking – Overclocking is a process of tuning a processor to operate at a higher Clock Speed than the default design value. Overclocking requires specific attention to power and heat dissipation and is only recommended for advanced users.

Intel processors

Android based Microsoft Surface Duo. Source: Microsoft

Intel markets its processors under various brands (listed in order of improving capability); Atom, Celeron, Pentium, Core and Xeon.

The majority of business computers have Core processors for both desktops and laptops, and as such this guide will focus on the Core brand.

Intel development history

A summary of the development history of Intel processors provides some context to changes in production models, delays in new micro-architectures and even potential changes in who manufactures their processors.

From 2008 to 2015 Intel processor development was enabled by a production model called, “Tick Tock”. Each Tick represents a die shrink, whereby the fabrication process to manufacture a processor is physically reduced without compromising processor capabilities. The result is a processor that needs less power, less cooling and is cheaper to manufacture. Each Tock provided a microarchitecture change where the processor hardware design is evolved (with an improvement to an existing microarchitecture in 2014 called a Refresh).

From 2017 the Tick Tock production model was replaced by a three-year POA model, “Process”, “Architecture”, “Optimisation”. The Process change reflects a die shrink (formerly Tick), followed by an Architecture change (formerly Tock), and then one or more Optimisations (similar to the Refresh seen in 2014 where an existing Microarchitecture is improved).

The below table shows the history of the Tick Tock and POA processor development.

The POA model reflects Intel’s slowing pace of achieving regular die shrinks, in particular in moving to 10nm. As a result, Intel released several optimisations to the 14nm Skylake micro-architectures as a compromise in achieving a 10 nm die shrink. Intel continues to be challenged in reducing die size which is critical to increasing the processing density for a given size processor without increasing cooling or power requirements. Intel has delayed the release of their 7nm processors multiple times with latest estimates sometime in late 2022 or early 2023.

Die Shrink benefits

Processors are manufactured on a semiconductor wafer, with each unit of electronic circuit called a die, measured in millionth of a meter, or nanometres. Reducing the die size, a “die shrink” results in multiple benefits for a given processing power; less cost, less power required and thus less heat generated making heat dissipation easier to manage.

Samsung Galaxy Book S with Lakefield processor combining both Core and Atom processors. Image: Samsung

The POA model also includes an exception in 2020, where a hybrid processor, Lakefield, was released. The Lakefield processor contains a Sunny Cove core, along with four low power Atom cores. This hybrid approach aims to bring a mix of low power performance with high power aimed at always connected laptops. The result is a laptop claiming up to 17 hours of battery. Intel does not identify this processor with any particular processor generation. The hybrid approach used for Lakefield is also expected to be used for Alder Lake processors forecast for 2021.

Within each microarchitecture, Intel uses codenames to reflect CPU releases. The below table provides codename and respective releases for Desktop and Mobile. The sections that follow will provide some supporting detail to the table content.

Unless specified otherwise, Intel provides graphics processing functionality that is integrated with the CPU. Intel markets these, in order of increasing capability, as HD, UHD, Iris, Iris Plus and Iris Xe. Each graphics model will come with an identifying tier such indicated with a G or a GT, with the trailing digit, the larger the more capable the graphics.

The number of cores and related threads are not apparent from either the processor generation or processor name. This information is only available from specification sheet with Intel processors having anywhere between two and ten cores and up to 20 threads.

Intel processor naming convention

Intel uses the above processor naming convention. The brand modifier options, in order of increasing capability are i3, i5, i7 and i9 (with i9 also called X-Series and focused on gaming and video editing). Gen indicator aligns to the generation of processor, with later generations more powerful for a given brand modifier, i.e. a 10th Generation i5 should be more capable than a 9th Generation i5. SKU allows differentiation between otherwise identical processors; in general, as the SKU number increases it reflects improvements in capability (SKU, pronounced “skew”, short for stock keeping unit) The Product Line Suffix definitions are shown in the following table.

As of October 2020, devices are available from manufacturers for 8th, 9th, 10th and 11th Generations processors. Examples for each are described in the pages that follow.

8th Generation: Cannon Lake and 10nm

Dell Latitude 7300 with Cannon Lake processor. Image: Dell

Eight Generation Cannon Lake processors, released in 2018, are a 10 nm die shrink of the 7th Generation 14 nm Kaby Lake processors. Cannon Lake was released after 9th and 10th Generation 14 nm processors which illustrates the challenges Intel faced in releasing 10 nm processors. The Dell Latitude 7300 is an example device with a Cannon Lake processor, i7-8665U, i7 quad core, 8th generation, SKU 665, and mobile efficient (U).

9th Generation: Coffee Lake Refresh

The 9th generation Coffee Lake Refresh processor was released in 2018 as an optimisation to the Skylike architecture based on 14nm die. Many of the processors are unlocked to allow overclocking or require discrete graphics; both characteristics of graphic intensive use cases which is why 9th Generation processors are often seen in gaming computers.

10th Generation: Comet Lake and Ice lake

10th Generation branding is applied to both Comet Lake and Ice Lake processors. The former based on Skylake microarchitecture with 14nm die and the later on Sunny Cove microarchitecture with 10nm die. Thus “10th Generation” has two different meanings.

The 14nm based Comet Lake processors have better performance, in terms of clock speed, than 10nm Ice Lake processors, perhaps highlighting the challenges Intel faced in executing a die shrink. With that said, all 14nm Comet Lake processors have integrated UHD graphics, while 10nm Ice Lake processors include options for more capable Iris Plus graphics (G1 = UHD Graphics, G4 = Iris Plus, and G7 = Iris Plus with improved performance).

Left: Lenovo ThinkCentre M90t, 10th Generation Comet Lake processor, Image: Lenovo. Right: Microsoft Surface Pro 7, 10th Generation Ice lake processor. Image: Microsoft.

An example Comet Lake device is the Lenovo ThinkCentre M90t with i3-10100, an i3 dual core processor. Note this processor does not have a name suffix because it does not include special features.

An example Ice Lake device is the Microsoft Surface Pro 7 with i7-1065G7 processor. Note, when using graphic suffix, if the last digit before the “G” is “0”, it’s a Y series, and if it’s a “5” is a U series, and “8” is a H series based on their power needs. As such, the quad core i7, 10th Generation Ice Lake Processor is a mobile efficient (U) processor with Iris Plus graphics.

11th Generation: Tiger Lake

Tiger Lake 11th Generation, 10nm processors, are marketed under a new branding targeting light and thin laptops called Intel Evo.

Tiger Lake also introduces a new generation of integrated graphics called Iris Xe, with improved capability compared to previous top of range Iris Plus.

11th Generation Rocket Lake-S due in 2021 will be the first architecture change for desktop processors since Skylake was introduced in 2015.

AMD processors

4th Generation Ryzen APU with on Zen 2 CPU architecture (3rd Generation CPU). Image: AMD

AMD has a smaller market share to Intel for mobile and desktop computers (15-20% v 80-85%) however in recent years their market share is growing assisted by improved performance processors achieved with 7 nm manufacturing process compared to Intel 10 nm processors.

The below table illustrates AMD processor development covering just over a decade.

Zen processor generations

Zen family of processors is marketed under the Ryzen brand name, with “generations” identified by a series number per the following table. Note that APUs (AMDs term for a processor with an embedded graphics capability) receives a “generation uplift” compared to a CPU of the same architecture.

AMD markets it’s processors with integrated graphics as Accelerated Processing Units (APUs). While there are some technical differences between the graphic integration approach by AMD and Intel, an AMD APU and an Intel CPU with integrated graphics refer to the same capability of a combined graphics and computer processor.

Ryzen APUs have integrated graphics branded using RX Vega with a trailing digit, e.g. RX Vega 8. For a given generation of processors, the higher the number the more capable the graphics with 11 the maximum at October 2020.

Outsourced or insourced?

AMD uses 3rd parties to manufacture their processors citing technical specialism and economies of scale as key benefits. In contrast, Intel almost exclusively manufactures its own processors citing this as a strategic advantage. With that said, Intel recently announced it may use third party foundries as part of its strategy to release their next (overdue) micro-architecture processors.

AMD processor naming convention

AMD markets their Zen processors under the Ryzen brand name using the below naming convention.

 Buyer’s guide checklist

  • For any business computer where graphics processing is not the primary objective Intel dominates the market and generally dominates the industry performance reviews for price comparable machines with an AMD processor. As such, consider AMD for graphics intensive needs, otherwise go with Intel unless you have an ideologic reason to avoid.

  • Processor “Generations” are marketing terms and comparison between generations requires consideration of other processor specific capabilities such as Cores, Threads, Clock Speed and Instructions Per Second (IPS) (see Page 2). As such, there is no substitute for processor benchmarks from sites such as https://www.cpubenchmark.net/ for independent performance measurements.

  • Processor capability is directly related to device capabilities such as cooling and power. When considering graphics processors, capabilities are also tied to display capabilities. Fortunately, device manufactures aim to match and optimise the capability of a processor for the intended device use. Device reviews provide the best proxy to determine the success of this aim.

  • Integrated graphics are sufficient when graphics processing is not the primary objective and discrete graphics apply if graphics processing is the primary objective.

  • Evaluate the processor for operating system support. Taking Windows as an example, new Intel processors are not supported on older OS versions. For example, Intel 10th Generation processors are not supported on Windows 10 version 1803 even though Microsoft supports this version for enterprise customers until May 2021.

  • The need to continually improve processor performance is making Intel consider significant changes in approach, first illustrated in the move from the Tick Tock to POA development model, and more recently illustrated to the release of hybrid processors using Core processors with low powered Atom processors (see Pages 4 – 5). We recommend being aware of processors released with a new development model and consider assessing the risk of being a first adopter. It may make sense to initially divert devices with new processor approaches to less critical use cases.

  • In general, buying the latest generation of processor will provide more capabilities, however as illustrated by Intel having two different processor ranges marketed as “10th Generation” it is important to understand the changes introduced by a new processor generation to evaluate if the benefits match your needs (see Page 8).

  • The market for processors continues to evolve rapidly, indeed Intel’s market leadership and recent product development success has waned in recent years. As such, keeping across developments when considering device selection is prudent. As such, we recommend subscribing to updates from devicecraft.co.uk!