Why Page Files on Slow Drives Ruin Hardware Benchmarks

This article explains how allocating a system's page file (virtual memory) to a slow storage drive creates severe bottlenecks that invalidate computer hardware benchmarks. By examining the massive speed disparity between system RAM and slower storage mediums, how benchmarks trigger memory swapping, and the specific ways this impacts CPU and GPU scores, you will understand how to properly configure your virtual memory for accurate performance testing.

The Role of the Page File in Benchmarking

The page file, or virtual memory, is a reserved portion of a storage drive that the operating system uses as an extension of physical Random Access Memory (RAM). When a computer runs out of physical RAM, or when the OS decides to free up RAM for active tasks, it moves inactive data to the page file.

Benchmarking software is designed to push hardware to its absolute limits. During these extreme stress tests, memory consumption spikes rapidly. Even if a system has plenty of RAM, the operating system may proactively shift background data to the page file to ensure the benchmark has maximum access to physical memory. If that page file resides on a slow mechanical Hard Disk Drive (HDD) or an aging SATA Solid State Drive (SSD), the entire system's performance is compromised.

The Speed Disparity Bottleneck

To understand why a slow drive ruins benchmark results, look at the bandwidth and latency differences between memory types:

  • System RAM (DDR4/DDR5): Operates at speeds between 25,000 MB/s and 60,000+ MB/s, with latency measured in nanoseconds.
  • NVMe SSDs: Operate at speeds between 3,500 MB/s and 14,000 MB/s, with latency measured in microseconds.
  • Mechanical HDDs: Operate at speeds between 80 MB/s and 200 MB/s, with latency measured in milliseconds.

When a benchmark forces the system to read or write to a page file on a mechanical hard drive, the data transfer rate drops from tens of thousands of megabytes per second to just a tiny fraction of that.

How It Corrupts Benchmark Scores

When the page file is on a slow drive, the benchmark ends up testing the storage drive's limitations rather than the true capabilities of the CPU or GPU. This manifests in three specific ways:

1. CPU Starvation and I/O Wait

Modern CPUs can process billions of instructions per second, provided they have a steady stream of data. If a CPU needs data that has been paged out to a slow drive, the processor must pause and wait for the drive to locate and send that data. In computing, this is known as "I/O Wait." During a timed benchmark, every millisecond the CPU spends waiting for a slow drive is time it is not rendering or calculating, resulting in artificially low multi-core and single-core scores.

2. Ruined 1% and 0.1% Low Frame Rates

In gaming and GPU benchmarks, average frame rates only tell half the story; 1% and 0.1% lows measure system stability and frame pacing. When a benchmark loads heavy textures, it can fill up both the GPU's Video RAM (VRAM) and the system RAM, triggering page file usage. A slow drive cannot deliver assets fast enough, causing the screen to freeze momentarily. While the average frame rate might only drop slightly, these stutters completely destroy the 1% and 0.1% low metrics, falsely indicating that the GPU or CPU is unstable.

3. Background Task Interference

A benchmark does not run in a vacuum; the operating system is still running in the background. If the OS decides to swap background processes to a slow drive during the middle of a benchmark run, the storage controller becomes saturated. This creates a system-wide latency spike that interrupts the benchmark's execution thread, leading to inconsistent and unrepeatable scores across multiple runs.

How to Fix Page File Bottlenecks

To ensure your hardware benchmarks reflect the true performance of your CPU, GPU, and RAM, follow these configuration steps:

  • Disable Page Files on Slow Drives: Go into your operating system's advanced system settings and explicitly set the page file size to "None" for all mechanical hard drives and slow SATA SSDs.
  • Assign the Page File to the Fastest NVMe Drive: Set the page file to "System Managed" exclusively on your fastest PCIe NVMe SSD. While still slower than RAM, an NVMe drive minimizes the latency penalty if paging occurs.
  • Close Background Applications: Free up as much physical RAM as possible before running a benchmark to reduce the likelihood of the OS needing to use virtual memory at all.
  • Do Not Disable the Page File Completely: Even with 32GB or 64GB of RAM, some benchmarking suites and games are hardcoded to look for a page file and will crash if one does not exist. Always keep an active page file on your fastest drive.