Why Reviewers Skip the First Game Benchmark Run
Hardware reviewers routinely discard the first benchmark run after launching a game to ensure accurate and repeatable test results. During the initial playthrough, a computer performs heavy background tasks like shader compilation, asset caching, and thermal stabilization, which cause artificial stuttering and inconsistent frame rates. By treating the first pass as a warm-up and only recording subsequent runs, testers isolate the true, sustained performance of the CPU and GPU from temporary initialization hiccups.
Shader Compilation Stutters
When a modern game loads a scene for the first time, the system must translate the game’s graphics shaders into code the specific graphics card can understand. This process, known as shader compilation, demands heavy CPU usage.
If a reviewer records the very first run, the benchmark will capture severe frame drops and stutters caused by the CPU scrambling to compile these shaders on the fly. Once compiled, shaders are saved to a local cache. Subsequent benchmark runs pull directly from this cache, eliminating the stutters and allowing the reviewer to measure the actual rendering power of the hardware rather than its ability to compile code.
RAM and VRAM Caching
The first time a benchmark sequence runs, the system must pull gigabytes of textures, audio files, and geometry data from the storage drive (SSD or HDD) and load them into the system memory (RAM) and graphics memory (VRAM). Even with fast NVMe SSDs, this initial data transfer creates a bottleneck that can lower average frame rates and ruin 1% low metrics.
After the first run finishes, most of these necessary assets remain cached in the much faster RAM and VRAM. Discarding the first run removes storage bottlenecks from the equation, ensuring that CPU and GPU benchmarks are actually testing the processors, not the drive speed.
Thermal Stabilization and Boost Clocks
Modern CPUs and GPUs use dynamic boosting algorithms—such as AMD Precision Boost or Nvidia GPU Boost—that adjust clock speeds based on current temperatures. When a game first boots, the hardware is relatively cold. This allows the components to boost to higher-than-normal clock speeds for the first few minutes, resulting in artificially high frame rates.
Conversely, as the first run progresses, the hardware heats up and clock speeds drop to a sustainable thermal equilibrium. Reviewers skip the first run so the cooling system and silicon temperatures can stabilize. This ensures the recorded benchmark reflects the performance a gamer will actually experience during a normal, extended gaming session, rather than a brief burst of "cold-boot" speed.
Background System Initialization
Launching a game triggers several operating system and software events that only happen once per session. These include:
- Anti-Cheat Software: Programs like Easy Anti-Cheat or Vanguard perform heavy initial scans upon game boot.
- Overlays and APIs: Steam, Discord, and graphics driver overlays hook into the game's rendering API, causing brief frame-time spikes.
- OS Resource Allocation: Windows Game Mode and thread directors take a moment to deprioritize background tasks and allocate maximum resources to the game.
Letting the benchmark run once allows all of these software layers to settle. The following runs provide a clean slate where the hardware's performance can be measured without interference from the operating system's startup routines.