Over the years, manufacturers have implemented various techniques to increase computer performance, like increasing the cores in a CPU and allowing multiple threads to run simultaneously on a single core. However, this means that statistics like gigahertz or core count are no longer a good way to compare the performance of two different laptops. If you want to know whether a laptop can process photo edits, run physics simulations, or compile code quickly enough to suit your needs, you can look to a Geekbench benchmark. These measurements are a good way to obtain comparable results among laptop models, helping you get a better idea of the kind of performance you can expect when running day-to-day tasks.
This article explains the conditions we perform our Geekbench tests in, and what the results mean in practical use.
Geekbench benchmarks are an easy way to determine the general performance of a laptop at a glance. The benchmarks measure how well the CPU performs a wide variety of workloads, mainly in encryption, general-purpose computing, and computationally intensive tasks like 3D renders. Geekbench also performs these tasks using one or all of a CPU's threads to determine the laptop's single- and multi-thread performance independently.
Single-thread performance affects scenarios where CPU instructions have to be performed in a very specific order to obtain the desired result, such as physics simulations that calculate the trajectory of one ball after it's thrown. Some programs like Adobe Photoshop benefit most from good single-thread performance. This is in contrast to multi-thread performance, which mostly affects applications that benefit from having other instructions being run simultaneously. A good example would be simulating the motion of a thousand stars around a black hole. Most modern applications are well-optimized for multiple threads, but if your laptop has good multi-thread performance, you'll also get a smoother experience when multitasking heavily or playing complex open-world video games.
Newer versions of Geekbench, including Geekbench 5, also measure the compute performance. GPUs are designed to perform graphical workloads like rendering video games, but this benchmark measures how well they can perform computational tasks, like dividing large matrices. GPUs have become increasingly prevalent in computation-heavy scenarios like animation rendering, so compute APIs like CUDA have been developed to increase the GPU's efficiency in these tasks.
Generally speaking, the higher the Geekbench score, the faster the laptop feels overall. It'll launch applications, load webpages, and complete heavy tasks like renders and code compiles faster. Keep in mind that a fast CPU and GPU doesn't necessarily mean you'll have a smooth, responsive laptop, as there may be other bottlenecks elsewhere in the system like a slow hard drive or RAM.
We test the laptop's performance by simply running the full suite of Geekbench 5 tests until completion. Depending on the operating system and manufacturer, some tests may not be available; scroll down to each individual test to see the details.
We use Geekbench 5 to measure the performance of a laptop alongside our Cinebench R23, Blender, Basemark GPU, and game benchmarks. We perform these tests one after another in a small, temperature-controlled room set to 22°C (71.6°F), with a tolerance of ±0.5°C. We keep the laptop plugged in using its included adapter and ensure that the battery is at full charge before beginning our tests.
Note: Our Geekbench benchmark determines the "cold performance" of the laptop. This means that the test isn't designed to take into account possible performance degradation due to thermal constraints. For more information, see our Performance Over Time test article.
We use the same versions of the available Geekbench 5 app for each operating system:
Windows: Version 5.3.1
macOS: Version 5.3.1.
Chrome OS: Android APK, version 5.2.5.
The single-thread benchmark score is a weighted result of the CPU's performance while performing cryptographic, integer, and floating point workloads, using a single thread on one core. Each workload type is described in further detail below. We run the test three times, with two-minute idle intervals between each run, then note the average as our result.
The numerical score doesn't mean anything in itself but is useful in comparisons. The baseline score of 1,000 corresponds to the single-thread performance of an Intel Core i3-8100, an entry-level quad-core desktop CPU released in late 2017; because the score is designed to be linear, double the score means doubled performance, half the score means halved performance, and so on. The i3-8100 is more than enough for medium productivity tasks and multitasking, so a laptop that scores lower than 1,000 may still be more than enough for your needs.
The higher the CPU's single-thread score, the faster each of the CPU's threads runs tasks dedicated to it. This means, generally speaking, if other threads are busy working on background tasks, the CPU can still run main tasks quickly. If a CPU's multi-thread score is excellent, yet its single-thread score is mediocre, workloads will take a while to finish if the system's other threads are under load.
Like the single-thread CPU benchmark, the multi-thread benchmark score is a weighted result of the CPU's performance while performing cryptographic, integer, and floating point workloads. However, this test utilizes all available threads on all cores to test how well they perform and schedule tasks among themselves. We run the test three times, with two-minute idle intervals between each run, then note the average as our result.
As above, the numerical score doesn't mean anything in itself but is useful in comparisons. The baseline score of 1,000 corresponds to the single-thread performance of an Intel Core i3-8100, an entry-level quad-core desktop CPU released in late 2017. Again, because the score-to-performance relationship is linear, a CPU with a multi-core score of 4,000 can generally run a task four times faster than a single thread on the i3-8100 if all system resources are dedicated to that task. At the same time, it can't complete that same task as quickly if its other threads are busy. Additionally, each program utilizes a CPU's cores and threads differently, so even if you're only running a single foreground task, you might experience worse-than-expected performance, especially on older programs.
A system generally has good multi-thread performance if it has many threads and efficient task scheduling. On the flip side, this doesn't necessarily mean that it also has good single-thread performance. Likewise, better single-thread performance doesn't necessarily equate to better multi-thread performance if the CPU doesn't have many cores or threads. It's good to keep in mind that having a comparatively high multi-thread score doesn't necessarily indicate that the CPU as a whole can run tasks in a fraction of the time as a single one of its threads.
The GPU compute benchmark measures how well a laptop's graphics card performs compute tasks like image processing, face detection, and physics simulations. Even though these tasks are vastly different than graphical workloads, they're still a good indication of how well the GPU runs graphical tasks like 3D rendering and video games. However, keep in mind that different compute APIs and graphics driver versions interface in different ways with the GPU, meaning the same GPU might perform very differently depending on which options you choose for certain tasks.
We choose different compute APIs that best reflect the experience we expect most users will have on their laptop's corresponding hardware:
Windows: We use the CUDA API if it uses an NVIDIA dedicated graphics card. Otherwise, we use the OpenCL API, which we use for Intel or AMD integrated graphics, or a dedicated AMD graphics card.
macOS: We use the Metal API. This graphics API is used in many games on iOS, as well as modern macOS games coded for Apple silicon.
Chrome OS: Unfortunately, the Android APK we use on Chrome OS doesn't support any GPU Compute tests. As a result, we can't give any direct comparisons regardless of whether the CPU is ARM- or x86-based. However, as most Chromebooks only have integrated graphics, we expect this value to be in line with Windows devices using similar CPUs that don't have a dedicated graphics card.
Note: The Vulkan API is most commonly used as a graphical backend in video games. We don't use it in our Geekbench tests because this test isn't designed strictly to measure game performance. For more information, see our articles for our Basemark GPU and game benchmarks.
The final numerical score that Geekbench presents for single-thread, multi-thread, and GPU compute workloads are only a weighted value of the laptop's performance in different types of operations. There are three main types of workloads that are tested, and each factor differently into the final scoring: cryptography (5%), integer (65%), and floating point (30%). A CPU can perform better in some workloads compared to others, depending on its architecture and how it handles (schedules) different instructions.
The cryptographic tests measure how well the CPU performs instructions related to encryption. It's particularly important to AES encryption, which secures communication channels like the HTTPS protocol used by every major website since around 2016. These typically involve manipulating very large numbers and matrices.
The integer workloads measure how quickly the CPU performs calculations with integer numbers; that is, whole numbers that don't involve any decimal points. These calculations are most commonly found in general computing, like when decompressing files, compressing images, rendering PDF documents, and compiling code.
Finally, the floating-point workloads measure how quickly the CPU performs calculations with floating point numbers; that is, numbers that are fractions of a whole number. These types of calculations are necessary when more precision in the final output is necessary, like in soft-body and fluid physics simulations, advanced image transformations like HDR image generation and Gaussian blur, computational operations like ray tracing, and even advanced operations like speech recognition and machine learning. Generally speaking, these computations are better executed on dedicated gaming or workstation graphics cards.
On the other hand, the GPU Compute workloads measure the compute performance; in other words, how well the graphics card performs at non-graphical tasks. Some of these tests used by Geekbench include edge-finding algorithms, automatic contrast adjustment of an image, face detection, and fluid/particle simulations.
There isn't one single laptop that performs incredibly well for every workload. As a consumer with a limited budget, getting the most out of your laptop is a compromise between finding the laptop model that best suits your needs and its cost.
For instance, if you intend to perform only light productivity tasks and don't need to multitask very much, you probably only need a laptop with a dual-core, 2-thread CPU. On the flip-side, a CPU with many cores, which individually run tasks more slowly, will very likely not provide any extra benefits to running a few light productivity workloads at a time. If you need to run more demanding workloads like games or video editing, or you multitask more often, you'll have a much smoother experience with a quad-core CPU, whether or not it can run multiple threads per core. If you intend to run very computationally expensive workloads like CPU rendering or physics simulations, you probably want something with many cores and threads, like the AMD Ryzen 9 5900HX or Intel Core i9-10980HK, both of which have 8 cores and 16 threads.
Some CPUs can run multiple threads on a single physical core, which improves multi-thread performance. Intel's implementation is called "Hyper-Threading Technology," or HTT, while AMD uses the term "simultaneous multithreading," or SMT. The principle of operation is similar in both cases, but Intel's implementation is proprietary, so its exact mechanism of action isn't publicly known. A processor with multithreading technology performs better than a processor with the same amount of cores without the capability; however, it performs worse than a processor with the same number of physical cores as the CPU with multiple threads per core. This is the reason why the dual-core, 4-thread Intel Core i3-10110U performs worse in online benchmarks compared to the quad-core, 4-thread AMD Ryzen 3 4300U.
If you want to have a laptop with performance that suits your needs, a Geekbench benchmark is a good reference. It scores a laptop's CPU performance when running several tasks, using a single thread or multiple threads. It also scores a laptop's GPU performance in computational, as opposed to graphical, workloads. The final benchmark results are a good reference point that can help you compare different laptops so you can find the best one that suits your needs.