How to Benchmark Your PC

Benchmarking a PC measures the performance of the CPU, GPU, and storage by running standardized tests and comparing the scores to baselines. A benchmark produces a repeatable number that quantifies a component’s speed, which confirms stability, compares hardware, and validates an upgrade. This article walks through benchmarking in phases ordered from purpose to comparison: understand why benchmarking matters, benchmark the CPU with Cinebench, benchmark the GPU with 3DMark, benchmark storage with CrystalDiskMark, run an in-game benchmark, monitor temperatures and clocks during the runs with HWiNFO, and compare the scores to online baselines.

The article also explains how thermal throttling lowers scores when a component overheats. Each phase states its goal and gives the exact steps. The result is a set of measured scores for the processor, graphics card, and drive, recorded alongside the temperatures during each run, ready to compare against published baselines for the same hardware to confirm the system performs as expected.

What You Need to Benchmark Your PC

Benchmarking a PC requires the benchmark tools for each component, a monitoring tool for temperatures, and a stable system state during the runs. The items required to benchmark a PC are listed below, in the order each is needed:

• Cinebench measures CPU performance through a multi-core and single-core rendering test that produces a comparable score.
• 3DMark measures GPU performance through graphics tests such as Time Spy and Steel Nomad with published baselines.
• CrystalDiskMark measures storage performance, reporting sequential and random read and write speeds for a drive.
• HWiNFO monitors temperatures, clock speeds, and power during the benchmark runs to detect thermal throttling.
• A stable system with background apps closed produces consistent scores, since running applications skew the results.

Each component has a dedicated benchmark: Cinebench for the CPU, 3DMark for the GPU, and CrystalDiskMark for storage, with HWiNFO tracking temperatures across all runs. Temperatures reaching the thermal limit lower scores, which the guide to monitoring CPU and GPU temps covers in full. A benchmark validates an upgrade or confirms a build performs as published baselines for the same hardware indicate.

Why Does Benchmarking Matter?

Benchmarking matters because it confirms stability, compares hardware, and validates that an upgrade or build performs as expected. A benchmark turns performance into a repeatable number that reveals whether a component meets its rated speed. Benchmarking serves three purposes listed below:

• Stability testing runs a sustained load to confirm the system does not crash or throttle under stress.
• Hardware comparison measures a component against published scores for the same and competing hardware.
• Upgrade validation compares scores before and after a hardware change to confirm the upgrade delivers its expected gain.

A benchmark score below the published baseline for the same hardware signals a problem, such as thermal throttling, a misconfigured power setting, or a background process consuming resources. Comparing before-and-after scores validates an upgrade, confirming a new graphics card or processor delivers the expected gain. A low score relative to baseline points to the same causes the guide to fixing a slow computer addresses for general performance.

Benchmark the CPU With Cinebench

Benchmarking the CPU with Cinebench measures processor performance through a rendering test that produces a comparable score. Cinebench renders a 3D scene across all CPU cores and reports a multi-core and single-core score that compares directly against other processors. Follow these steps:

1. Download Cinebench R23 or Cinebench 2024 from Maxon, the tool’s developer.
2. Close background applications, since running programs consume CPU cycles and lower the score.
3. Run the multi-core test, which loads every core and thread to measure full processor performance.
4. Run the single-core test, which measures per-core performance relevant to many games and applications.

Cinebench reports a multi-core score for full processor performance and a single-core score for per-core speed, both comparable against other processors. The multi-core score reflects rendering and productivity workloads, while the single-core score reflects gaming and lightly threaded tasks. A score below the published baseline for the same processor points to thermal throttling or a power limit, which the temperature monitoring step confirms.

Benchmark the GPU With 3DMark

Benchmarking the GPU with 3DMark measures graphics performance through standardized tests with published baselines. 3DMark runs a graphics workload such as Time Spy or Steel Nomad and reports a score that compares against other graphics cards. Follow these steps:

1. Download 3DMark from UL Solutions, available on Steam and the developer’s site.
2. Run Time Spy for a DirectX 12 test at 1440p, the standard benchmark for modern gaming graphics cards.
3. Run Steel Nomad for a heavier modern GPU test, which replaced the older Fire Strike workload for current cards.
4. Record the graphics score and overall score, then compare against the 3DMark database for the same card.

3DMark Time Spy tests DirectX 12 performance at 1440p, while Steel Nomad provides a heavier workload for current graphics cards, both with large comparison databases. The graphics score isolates GPU performance from the processor, producing a number that compares directly against the same card. A GPU score below baseline points to thermal throttling or a driver issue, which the guide to monitoring CPU and GPU temps helps diagnose.

Benchmark Storage With CrystalDiskMark

Benchmarking storage with CrystalDiskMark measures the read and write speeds of a drive in megabytes per second. CrystalDiskMark runs sequential and random transfer tests and reports the speeds, which confirm a drive performs at its rated specification. Follow these steps:

1. Download CrystalDiskMark from its developer, then select the target drive.
2. Run the default test, which measures sequential and random read and write speeds.
3. Read the sequential speeds, which indicate large-file transfer performance such as copying video.
4. Read the random 4K speeds, which indicate small-file performance relevant to system responsiveness.

CrystalDiskMark reports sequential speeds for large-file transfers and random 4K speeds for small files, with a SATA SSD reaching around 550 MB/s sequential and an NVMe PCIe 4.0 drive reaching 7,000 MB/s. The random 4K speed affects system responsiveness more than the sequential figure for everyday use. A drive below its rated speed signals a connection or configuration issue, which the comparison of HDD and SSD storage relates to drive types.

Run an In-Game Benchmark

Running an in-game benchmark measures real gaming performance in frames per second within an actual game. Many games include a built-in benchmark that runs a fixed scene and reports average, minimum, and maximum frame rates. Follow these steps:

1. Open the graphics or display settings in a game with a built-in benchmark, such as Cyberpunk 2077 or Shadow of the Tomb Raider.
2. Set the resolution and graphics preset to the settings the benchmark should measure.
3. Run the built-in benchmark, which plays a fixed scene and records the frame rate.
4. Record the average, minimum, and 1 percent low frame rates, which describe smoothness beyond the average alone.

An in-game benchmark measures real gaming performance at chosen settings, reporting the average, minimum, and 1 percent low frame rates that describe smoothness. The 1 percent low frame rate reveals stutters the average hides, since a high average with low minimums still feels uneven. A frame rate below expectations for the hardware connects to the guide to increasing FPS in games, which addresses settings and bottlenecks.

Monitor Temperatures and Clocks During Runs

Monitoring temperatures and clocks during benchmark runs reveals whether thermal throttling lowers the scores. HWiNFO logs the temperature, clock speed, and power of the CPU and GPU during a benchmark, exposing throttling that caps performance. Follow these steps:

1. Download HWiNFO and open the sensors view, which lists CPU and GPU temperature, clock speed, and power.
2. Start logging in HWiNFO before the benchmark, which records the values across the run.
3. Run the benchmark while watching the temperatures, noting whether the CPU or GPU reaches its thermal limit.
4. Check whether the clock speed drops as temperature rises, which indicates thermal throttling capping performance.

A clock speed that drops as the temperature reaches the thermal limit indicates thermal throttling, where the component slows itself to avoid overheating. A CPU reaching around 95 degrees Celsius or a GPU near its limit throttles, lowering the benchmark score below baseline. Monitoring during the run identifies throttling as the cause of a low score, which the guide to monitoring CPU and GPU temps addresses through cooling.

Compare Scores to Online Baselines

Comparing scores to online baselines confirms whether the system performs as expected for its hardware. Each benchmark publishes a database of scores for known hardware, against which a measured score reveals normal or below-normal performance. Follow these steps:

1. Find the baseline score for the exact hardware in the 3DMark database or Cinebench comparison charts.
2. Compare the measured score to the baseline, noting whether it falls within the expected range.
3. Investigate a score more than 10 percent below baseline, which points to throttling, drivers, or background load.
4. Record the scores for future comparison, establishing a reference for validating later upgrades.

A score within a few percent of the published baseline confirms the hardware performs normally, while a score more than 10 percent below points to thermal throttling, outdated drivers, or background processes. The 3DMark database and Cinebench charts hold baselines for specific hardware to compare against. Recording the scores establishes a reference that validates a future upgrade against the guide to fixing a slow computer for any later regression.

Common Mistakes to Avoid

Benchmarking produces misleading scores when the system state is inconsistent or temperatures go unmonitored. The mistakes that distort benchmark results are listed below:

• Leaving background applications running consumes resources, so closing them produces a consistent score.
• Ignoring temperatures during the run hides thermal throttling, so HWiNFO monitors the thermal limit.
• Comparing scores across different test versions misleads, so the same benchmark version is used for comparison.
• Running a single pass can catch an anomaly, so several runs confirm a consistent score.
• Benchmarking on a laptop on battery caps performance, so the laptop stays plugged in for full clocks.

A benchmark score below baseline most often results from thermal throttling or background processes rather than faulty hardware, which monitoring temperatures and closing applications confirm. Comparing scores requires the same benchmark version, since Time Spy and Steel Nomad produce different numbers. A laptop on battery throttles to save power, so a plugged-in run measures full performance, a factor the guide to monitoring CPU and GPU temps relates to sustained clocks.

Key Takeaways

• Benchmark the CPU with Cinebench, recording the multi-core and single-core scores for comparison.
• Benchmark the GPU with 3DMark, running Time Spy or Steel Nomad against the comparison database.
• Benchmark storage with CrystalDiskMark, reading the sequential and random 4K speeds.
• Run an in-game benchmark, recording the average and 1 percent low frame rates for real gaming performance.
• Monitor temperatures with HWiNFO, since thermal throttling lowers scores when a component overheats.
• Compare scores to online baselines, investigating any result more than 10 percent below the expected range.

Last Thoughts on Benchmarking Your PC

Benchmarking a PC follows a fixed order: understand why benchmarking matters, test the CPU with Cinebench, the GPU with 3DMark Time Spy or Steel Nomad, and storage with CrystalDiskMark, run an in-game benchmark, monitor temperatures with HWiNFO, and compare the scores to online baselines. Thermal throttling lowers scores when a component overheats, so monitoring temperatures during each run explains a result below baseline.

Readers can continue with the guide to monitoring CPU and GPU temps, the guide to increasing FPS in games, or the PC tutorials hub. The guide to fixing a slow computer addresses a score that falls below the expected baseline.