Affiliate Disclosure: By buying the products we recommend, you help keep the lights on at MakeUseOf. Read more.
Whenever a new GPU or CPU enters the market, the online world floods with benchmarks. Benchmarks allow users to take temperature readings under load, gauge PC performance through FPS, and test PC components for stability. Furthermore, benchmarks can also indicate how well your rig performs compared to other systems across the globe.
For our purposes, we will duplicate the process undergone by tech experts when testing the performance of new PC parts.
Benchmarking compares the performance output of your PC components with other, similar components. Benchmarking also allows users to keep track of how well their parts function, and whether or not a particular part is malfunctioning.
For example, if you have a high functioning CPU, but a low functioning GPU, you may experience issues with ultra-high video game settings. If both your CPU and GPU are functioning well, but either one overheats, your PC may shut down without warning.
Furthermore, benchmarking components lets you gauge how well games will run on your PC.
Note that benchmarking is different from stress-testing. PC stress tests push the components to their limit. This is useful for overclocking, wherein pushing parts past their limit may crash the PC under a heavy load. Benchmarking, on the other hand, tests performance across differing levels of intensity.
Caution: To gather a proper benchmark reading, close all other running programs during benchmark tests. All will require 100% of your CPU in order to gauge performance.
You can find a number of different benchmarking tools online and not one is better than the other. Yet, benchmarking tests do become more complex, depending on what you wish to test.
A general benchmark will gauge three simple variables: clock speeds, temperatures, and voltages. HWMonitor provides this exact information. Although using monitoring software does not officially qualify as benchmarking, HWMonitor will allow users to keep track of different readings throughout the benchmarking process. As your GPU and CPU work harder, your temperature readings will rise.
This simple monitoring method displays two key variables: clock speeds and temperature readings. If your temperature readings are high (80-90 °C) at idle — meaning your PC is not working very hard — you should consider taking measures to cool it down. If your components are cool at idle conditions, but increase dramatically under load, your GPU fan may not be working or may be working improperly.
Gauging FPS is the best method to test PC performance. It’s also reliable, because FPS relies both on CPU and GPU performance. A higher FPS typically indicates a faster overall PC.
Online tech reviewers use minimum, maximum, and average FPS scores to test a component’s performance. There is an inverse ratio between FPS and in-game activity. Minimum FPS readings mean more in-game activity, which stretches PC performance, while maximum FPS readings mean little is happening in-game.
There are two types of FPS benchmarks: simulated and real-time. When a program runs pre-rendered visuals on your PC to test FPS, it’s simulated. These simulated tests gauge performance during both high and low activity.
Real-time FPS ratings are taken during regular gaming sessions. These ratings test how well your PC performs with daily use. For example, users may switch from a low-intensity game like League of Legends to a graphics intensive game like Grand Theft Auto V.
FPS readings may differ widely between these two sessions. FPS readings will also differ based on the in-game video settings. Real-time FPS ratings give more realistic results than simulations, so tech reviewers often present both test results.
Not all benchmarking tools give scores. Most test the performance of your components. Certain software, such as 3DMark or RealBench, will provide a score. This score is particular to the software. You can use these scores to compare them with other PCs.
These scores will not gauge what games or programs will run on your PC. They do, however, show how your PC matches up with other PCs around the globe.
GPU benchmarking is a great method of checking the ability of a newly acquired or overclocked GPU. As mentioned above, you can benchmark your GPU using two different methods: simulated and real-time benchmarking.
GPUs affect gaming performance more than they do general PC usage and multi-tasking. The best method for real-time benchmarks is to run a graphics intensive game and track your FPS. If your FPS ranges from 10-20, consider lowering graphics settings for better gaming results.
3DMark also provides a free demo version of their Time Spy benchmark (available on Steam), which works to benchmark PC components.
Unigine’s Heaven Benchmark and 3DMark’s Time Spy test different versions of DirectX. The free version of Heaven benchmarks using DirectX 11, while Time Spy uses DirectX12. DirectX12 is only available for Windows 10.
Pick your video settings in the splash screen and run the software to get started. The higher setting, the more stress the simulation puts on your PC. I used Heaven Benchmark on High settings to benchmark the effectiveness of a slight GPU overclock. These results were taken using 1440 x 980 resolution. Higher resolutions will provide better graphics, but will also stress your GPU more so than lower resolutions.
I increased my core clock from 980 MHz to 1070 MHz, and my memory clock from 1070 MHz to 1550 MHz, and saw an FPS and Heaven Benchmark score increase after overclocking my GPU. You can also save your results when the test finishes. Heaven Benchmark will save in your C:\Users\[PC Name] folder by default.
This is the simplest way to benchmark your GPU’s performance.
Real-time benchmarks gauge the effectiveness of your GPU during normal game play. This is a great way to test the different FPS ratings you would receive using different video quality settings. They’re also great measuring tools for GPU performance across different game types. AAA-rated games, for example, are more intensive on the GPU than online multiplayer games like League of Legends and World of Warcraft.
To perform a real-time benchmark test, download the popular FPS program Fraps. When Fraps is running, your FPS reading will automatically appear on the corner of your screen. Open Fraps and open the FPS tab to orient your indicator.
Ensure MinMaxAvg is checked under Benchmark Settings. Leave everything else as it. Now, you can begin real-time benchmarks. Minimize Fraps and open a game of your choice. You will notice that Fraps provides a large, yellow FPS indicator on the corner of your screen.
While your game is open, press F11 on your keyboard to begin benchmarking. Fraps will begin tallying your minimum, maximum, and average FPS rating. These specifications are saved to your C:\Fraps\Benchmarks folder as a .csv (spreadsheet) file. The image below is an example of Fraps readings. The left benchmark was taken running Middle Earth: Shadows of Mordor in Medium settings, while the right shows similar gameplay in Ultra settings.
For further results, check FPS in your Benchmark Settings to receive a spreadsheet of every FPS rating received while gaming. Games like Middle Earth: Shadows of Mordor also provide in-game benchmarks to test your video settings. Although these tests are simulated, they test in-game graphics and activity rather well.
The above results were taken from the in-game benchmark at Ultra settings, which closely match our Fraps readings.
CPU benchmarks differ slightly from GPU benchmarks. While GPU benchmarks focus on the ability to run games smoothly, CPU benchmarks often test your PCs ability to multitask.
Make no mistake; a better CPU will allow games to run more smoothly on your PC. You can even use the FPS tests to benchmark your CPU speeds. Yet, testing the CPU itself depends on how many programs you have open and run efficiently.
CineBench‘s CPU benchmark only tests render times, but provides a convenient comparison chart directly in the software.
Render and playback speeds rely heavily on CPU speeds. Experts often use render tests to gauge the performance of different CPUs.
The render test is a rather effective way to test processor speed, but isn’t indicative of how well a CPU would work under regular conditions. This also doesn’t directly compare to real-world tests like multitasking.
Software like GeekBench allows users to test their CPU and compare their scores with Geekbench’s entire benchmark library. It also claims to run real-world tests, ensuring the accuracy of the test.
I prefer RealBench, which uses a variety of real-world tests to benchmark CPUs. The tests include image editing, video encoding, programming, and multitasking. RealBench also displays every process being conducted directly on your desktop.
Once you have your results, compare your RealBench results with the RealBench library and leaderboards as well. Not only is RealBench an effective benchmarking tool, it’s an excellent, real-world stress test as well.
How Much Do You Bench?
If you’ve never benchmarked your PC, now is the perfect time to try. Benchmarking won’t speed up your computer or increase performance, but being able to benchmark your component parts can be an effective troubleshooting tool. If not, you’ve finally caught a glimpse as to what benchmarks are, and how PC enthusiasts use them to test part performance.
Do you benchmark your PC? What for and what software do you use? Let us know in the comments below!