Cold Boot vs Warm Boot Explained

This page sits within our benchmarks section and addresses a specific measurement question: when someone quotes a boot time, did they measure from a cold start or a warm restart? The search intent is straightforward - you want to understand why two seemingly identical tests produce different numbers and which one reflects real-world performance more accurately.

I have been measuring boot times on refurbished hardware for over a decade, and the cold-versus-warm distinction is the single most common source of misleading numbers in PC performance discussions. A machine that cold boots in 45 seconds might warm boot in 18 seconds, and both numbers are technically correct - but they describe very different things. Below we cover the mechanics of each boot type, where the time goes in each phase, how caching creates the gap, benchmark data from real machines, and guidelines for when each measurement is useful. For the full range of our performance measurements, visit the Benchmarks hub.

What Happens During a Cold Boot

A cold boot begins from a fully powered-off state. RAM is empty, caches are cleared, and every component must initialise from scratch. The sequence goes through several distinct phases, each consuming measurable time.

POST (Power-On Self-Test): The firmware tests RAM, detects storage devices, initialises the display, and checks connected peripherals. On older machines with legacy BIOS, this can take 5-15 seconds. UEFI firmware is generally faster at 3-8 seconds, though some manufacturers add their own splash screen delays.

Bootloader: GRUB, Windows Boot Manager, or another bootloader loads from the boot partition. This phase is usually 1-3 seconds on an SSD and 3-8 seconds on a mechanical drive.

Kernel and driver initialisation: The operating system kernel loads into RAM, initialises drivers for storage, display, network, input devices, and USB controllers. On Linux, this phase is visible via systemd-analyze. On Windows, it is the period between the loading animation and the login screen.

User session startup: Login services, desktop environment, and startup applications load. This is where accumulated bloat has the most impact - every background service adds seconds on mechanical drives.

What Happens During a Warm Boot

A warm boot is a restart without a full power cycle. The machine goes through the same sequence as a cold boot in theory, but several caching mechanisms shorten it in practice.

Drive controller caches: SATA and NVMe controllers retain internal metadata about the drive layout. After a warm restart, the controller skips some of the enumeration and negotiation steps that occur during a cold power-on.

UEFI variables: Modern firmware caches boot configuration data in NVRAM. A warm restart reads these cached values rather than re-enumerating all devices.

Filesystem caches: If the OS writes cached metadata to disk before restarting, the subsequent boot can read it back faster than building it from scratch. Windows Fast Startup takes this further by saving the entire kernel session to disk.

The combined effect is that warm boot numbers are consistently 30-60% faster than cold boot numbers on the same hardware. This gap is wider on mechanical drives than on SSDs because the caching advantage is proportionally larger when base read speeds are slow.

Side-by-Side Timing Comparison

The following measurements come from three machines tested with both cold and warm boot protocols. Each figure is the median of five consecutive runs. Cold boot runs include a 10-second power drain between each cycle. Warm boot runs use a standard OS restart command.

Why the Distinction Matters

If you are evaluating whether a hardware upgrade or OS change is worthwhile, the boot measurement you use changes the conclusion. Warm boot numbers make every configuration look better than it actually performs from a full shutdown - which is the scenario most relevant to real-world use.

Consider the HDD configuration above: warm boot says 38 seconds, which sounds acceptable. Cold boot says 87 seconds, which clearly indicates a problem. If you based your upgrade decision on the warm boot number, you might skip the SSD upgrade that would cut the cold boot to 22 seconds.

When cold boot is the right measurement: After a power outage, an overnight shutdown, a battery drain, or whenever the machine has been off long enough for caches to clear. This is the realistic worst case - and on older machines, it is often the daily case.

When warm boot is useful: If you are measuring the effect of a software change within a single session - for example, testing whether disabling a startup program reduces restart time. In this context, the warm boot isolates the software variable without the noise of POST and firmware initialisation differences.

A Note on Windows Fast Startup

Windows 8 and later include a feature called Fast Startup that blurs the line between cold and warm boot entirely. When you "shut down" with Fast Startup enabled, Windows saves the kernel session to a hibernation file on disk. The next power-on loads this file rather than performing a full kernel boot.

This produces boot times that look like warm boot numbers even after a full power cycle. It is misleading for benchmarking purposes and can also cause issues with dual-boot configurations and driver updates. For accurate measurements, disable Fast Startup: open Power Options, click "Choose what the power buttons do," then uncheck "Turn on fast startup."