Throttling: Thermal Management and the New MacBook Pro

10 August, 2018

As consumer demands evolve, and devices grow ever thinner and more powerful, the need for thermal simulation is increasing. In this new environment, the margins and limits that engineers must work to are becoming narrower, with even a 5°C temperature ‘safety’ margin being too expensive to incorporate within most electronic equipment.

While also getting thinner, modern laptops and smartphones are also packing in greater processing power, more functionality and a longer lasting battery. To run such devices steady state at full power would cause them to overheat, making them either unsafe, uncomfortable for users to hold, or risking significant damage to the internal components. As a result, the management of power consumption – and by extension heat generation – is critical to ensuring the very best user experience while protecting the electronics.

This balance is often achieved by managing the speed and frequency at which the main processor is running – a practice known as ‘throttling’. Such throttling has recently become a hot topic in the electronics community, following the suggestion that the new 2018 Macbook Pro laptop can appear slower than its 2017 predecessor as the new i9 processor slows down to avoid overheating.

While it’s to be expected that a laptop as thin as the MacBook Pro should incorporate some degree of throttling, experts in the Mac community have rightly taken issue with the sheer amount of throttling that appears to be happening in Apple’s new high-end machine.

Given that the MacBook Pro is advertised as a high-end device (available for over £5000), one of its core audiences has been video editors and graphic designers – people who genuinely require a device with the highest possible specs. Given this fact, the idea that the 2018 MacBook Pro may be slower than its older, cheaper alternative has raised a lot of concerns.

Following several tests by independent experts, engineers and Mac community members, the general consensus has been that – although the chassis and the thermal design power for both the 2017 and 2018 MacBook processors are the same - Apple is in fact over-throttling the i9 processor on the new Macbook Pro.

In designing the MacBook Pro, the designers will have run multiple simulations to examine the performance trade-offs with power consumption and heat dissipation before production. The simulation of throttling scenarios may sound relatively straightforward however there are a number of things to consider. The heat dissipation will usually be calculated from a power vs temperature curve for the processor. But that curve is usually defined at a single (typically maximum) operating frequency. But if the system is to be throttled, the frequency of the processor will be lowered and the heat dissipation will be defined by another curve. Indeed, the software needs an input curve for each and every frequency the processor could adopt. This is a set of circumstances that would be difficult to model using legacy simulation tools.

In examining the importance of this issue, we at 6SigmaET have recently developed the use of multiple power vs temperature curves for the upcoming Release 13 of our simulation platform. Using this feature, engineers testing the throttling of their designs will be able to develop even faster and more accurate simulations of different throttling scenarios.

In the last week, Apple has been working hard to address the MacBook’s own throttling issues, recently putting out a new firmware update as part of macOS High Sierra designed to solve the device’s aggressive over-throttling. The success of this update, however, is yet to be decided.

Blog written by: Matt Evans, Product Engineer

Other Recent Posts

6SigmaET R15: Thermal Simulation for Effective Electronics

Thermal simulation using Computational Fluid Dynamics (CFD) provides users with an accurate visual…

Read More

Modeling Liquid Cooled Electronics with 6SigmaET Release 15

We recently launched the 15th release of our thermal simulation software 6SigmaET. 6SigmaET uses co…

Read More