Windows 11 24H2 Throttling Explained: Why Windows 10 Feels Faster

A programmer’s deep dive into Windows 11’s behavior has given a clear, if uncomfortable, explanation for why many users say Windows 10 still feels faster: aggressive background throttling tied to window focus and power-management policies on some hardware, combined with UX regressions in File Explorer and other shell components. The discovery—published as a community write-up and summarized in recent coverage—helps explain repeatable “snappy vs. sluggish” impressions, shows a path Microsoft can follow to close the gap, and highlights practical steps users and administrators should take now as Windows 10 approaches end of support. (techradar.com)

Background / Overview​

Windows 10’s official security and feature updates end on October 14, 2025. Microsoft is offering Extended Security Updates (ESU) as a one‑year bridge for some customers, but that is a stopgap—most users and organizations are being nudged toward Windows 11. The timing makes performance perceptions especially important: if users feel the new OS is slower, the migration story weakens. (microsoft.com)
The debate over perceived speed has two overlapping strands. One is apples‑to‑oranges benchmarking and marketing—where newer hardware running Windows 11 naturally outperforms older Windows 10 PCs. The other is apples‑to‑apples side‑by‑side comparisons on the same hardware, where many independent testers and community members have observed small regressions in everyday responsiveness, File Explorer behavior, and occasional throttling under certain conditions. New findings about minimized/focused-window behavior squarely address the latter.

---

What the programmer found: Minimizing a window can throttle CPU throughput​

The claim in plain terms​

The core observation is straightforward: on some setups running Windows 11 24H2, when an app window loses focus or is minimized, CPU clocks and application throughput can drop sharply, sometimes to roughly half the previous performance in CPU‑bound tests. The effect was reported across synthetic workloads (Prime95) and real utilities (7‑Zip), suggesting it isn’t limited to one program or benchmark. (techradar.com)

Why this matters​

Modern operating systems often bias resources toward the foreground to improve perceived responsiveness. But the discovery shows those policies can be too aggressive on some hardware — throttling background work that users expect to continue at full speed. For anyone running background builds, long compressions, virtualization, or scientific workloads, a minimized window shouldn’t translate into a cut in throughput measured in the tens of percentage points.

What the evidence looks like​

Public posts and screenshots show real‑time reductions in CPU utilization and lower clock frequencies the moment a window is minimized. In at least one reported reproducible case, 7‑Zip and Prime95 performance dropped by roughly half immediately after minimization, and partially recovered when the window was restored. Those reproducible traces are the starting point for debugging whether the root cause is a power policy, a scheduler decision, microcode/firmware interaction, or a Windows bug. (techradar.com)

Caveat — correlation versus root cause​

While the behavior is demonstrable on affected machines, the exact root cause is still being sorted by the community and vendors. Early signals point to an interaction between Windows’ process/foreground prioritization, power management defaults, and certain Intel platform behaviors, but full causality (for every affected model) requires vendor telemetry and driver/firmware validation. Treat individual reproductions as high‑quality symptom evidence, not a final verdict on a single-code defect until vendors confirm a specific fix path. (techradar.com)

---

Technical anatomy: Why an OS might throttle minimized apps (and when it goes wrong)​

Foreground prioritization and power policies​

Windows 11 includes enhanced heuristics that prioritize the foreground app to improve interactivity. Those heuristics are reasonable for desktop responsiveness but become problematic when:

• The OS interprets “not focused” as “low priority” even for long‑running compute tasks.
• Power management (ACPI/OS policy) allows deeper frequency reductions for background processes.
• Hardware/firmware (CPU microcode or platform power tables) reacts to those OS signals by changing performance states more aggressively than expected.

If the policy margins are too wide, background workloads suffer noticeably.

Hybrid CPU scheduler interactions​

Modern Intel hybrid CPUs (P‑cores + E‑cores) depend on the OS scheduler to place latency‑sensitive threads on P‑cores and background threads on E‑cores. A subtle scheduler change or a misapplied heuristic can push important threads to less capable cores, or wrongly reduce the P‑core frequency when the system believes foreground responsiveness is preserved. That can create the double‑hit of lowered clocks and fewer cycles available for a background job.

Power plans, driver and firmware factors​

Some workarounds in community reports required forcing a maximum power plan or adjusting background processing settings. That’s consistent with a throttling mechanism exposed through the usual ACPI/OS power interfaces. But vendor firmware (BIOS/UEFI microcode) and chipset drivers also shape behavior; manufacturer updates sometimes restore expected performance where OS tweaks do not.

---

Real‑world impact: Why users feel Windows 11 is slower​

Accumulation of small regressions​

Three classes of everyday problems appear to have contributed to the “Windows 10 feels snappier” impression:

• File Explorer latency — modernized UI components (WinUI/XAML rewrites) introduced small but visible delays on common tasks like opening a folder, loading previews, and context menus.
• Edge‑case throttling — minimized/focus‑loss throttling creates jarring slowdowns for background tasks that used to run uninterrupted.
• Micro‑stutters in games and apps — hybrid scheduler and driver issues can cause microstutters or worse minimum‑frame behavior for specific games or workloads on some CPUs.

Each problem alone is tolerable; together they change the subjective feel of the OS.

Benchmarks vs. the feeling of speed​

Synthetic benchmarks often show parity or small differences between Windows 10 and Windows 11 on identical hardware, because they measure raw throughput under controlled conditions. The user experience is often decided by interaction latency — how quickly Explorer responds, how a minimized backup finishes, and whether an app resumes instantly. That’s where the reported regressions matter most.

---

Microsoft’s response so far: fixes and the 25H2 roadmap​

KB5055627 — targeted improvements in 24H2​

Microsoft’s April 25, 2025 preview update KB5055627 (OS Build 26100.3915) included explicit fixes aimed at shell responsiveness, notably improving the performance of extracting zipped files and addressing several File Explorer issues. Microsoft documented the changes in the KB notes, and independent reporters observed measurable Explorer startup improvements on many systems after the patch. That indicates Microsoft is iterating on the shell and that at least some regressions have been identified and mitigated. (support.microsoft.com) (windowslatest.com)

Insider telemetry and new diagnostics​

Microsoft has broadened diagnostic collection for “system sluggishness” via Insider builds to gather better traces from affected users. This decision signals that the company recognizes these issues as multi‑factor and is seeking the telemetry to separate scheduler, driver, firmware, and application causes. Internal collection and Feedback Hub traces are a practical next step to root‑cause more complex interactions. (theverge.com)

25H2: an enablement package, not a performance reboot​

Windows 11 version 25H2 is being released as an enablement package layered on top of 24H2 (an approach Microsoft has used for prior annual updates). That means 25H2 will not fundamentally rework the kernel or servicing branch; it enables features already present in 24H2. Early independent benchmarks show little to no raw CPU performance difference between 25H2 and 24H2, which tempers expectations for wholesale speed improvements from that update alone. The Release Preview availability confirms Microsoft’s intention to focus on stability and enablement rather than dramatic performance gains. (blogs.windows.com) (tomshardware.com)

---

Cross‑checking the claims: what’s verified and what needs more proof​

• Verified: Windows 10 end of support date (October 14, 2025) and the availability of ESU options are documented by Microsoft. (microsoft.com)
• Verified: KB5055627 includes Explorer‑related fixes (zip extraction performance) in the official release notes; independent outlets reported improved Explorer startup on patched machines. (support.microsoft.com) (windowslatest.com)
• Verified: Reports of performance drop when minimizing apps on 24H2 have been published and reproduced by community members; major outlets have covered the reports. The symptom is real on affected configurations. (techradar.com)
• Partially verified / needs vendor confirmation: the exact single root cause (for example "a Windows scheduler bug" vs "interaction with a specific Intel microcode") remains under investigation; community reproductions implicate a policy interaction but the final classification requires driver/firmware diagnostics and vendor confirmation. Treat individual hardware‑specific claims as high‑quality but potentially narrow in scope until a coordinated vendor fix or Microsoft statement ties them to a single change. (techradar.com)
• Verified: 25H2 is an enablement package and early benchmarks show no material CPU gains vs 24H2. That makes 25H2 a limited vehicle for immediate perceptible speed improvements. (blogs.windows.com)

---

Strengths in Microsoft’s approach — and remaining risks​

Notable strengths​

• Targeted fixes are being shipped. KB5055627 shows Microsoft is addressing shell regressions and file‑operation performance directly. (support.microsoft.com)
• Telemetry and Insider diagnostics will help Microsoft separate OS, driver, and firmware causes faster than public debugging alone. That’s the right engineering response for multi‑factor performance bugs. (theverge.com)
• Enablement packaging for 25H2 reduces risk of regressions for mass rollout by limiting surface area for new changes. That helps administrators plan updates with lower disruption risk. (blogs.windows.com)

Remaining risks​

• Edge‑case regressions can linger. Policies tuned for user experience can overreach on certain hardware families; unless OEM/driver fixes are coordinated, affected users may see repeated cycles of partial fixes and new regressions. (techradar.com)
• Perception vs. measured gains. Even if average benchmarks are neutral, visible day‑to‑day latency can drive stubborn resistance to migration—perception matters nearly as much as throughput for adoption.
• Support timeline pressure. With Windows 10’s end of support approaching, enterprises must decide whether to tolerate short‑term regressions, buy ESU time, or accelerate driver/firmware testing before a broad rollout. (microsoft.com)

---

Practical checklist — what users and IT teams can do today​

For individual users (desktop/laptop)​

• Update Windows and drivers first. Install the latest cumulative updates and vendor GPU/chipset drivers; KB5055627 and other cumulative releases have already addressed some explorer and unzip performance problems. (support.microsoft.com)
• If you see minimized‑window throttling, test with a “maximum performance” power plan as a temporary mitigation and compare results. If that helps, report logs to Feedback Hub and your OEM. (techradar.com)
• Disable nonessential visual effects temporarily (Animations / Transparency) to improve perceived responsiveness.
• For long background jobs: avoid minimizing the active window during runs, or move the workload into a true background service/task scheduled explicitly to run at full priority.
• Consider enrolling in the Windows Insider Release Preview (on a test machine) if you want early access to 25H2 for validation, but don’t deploy preview builds broadly in production. (blogs.windows.com)

For IT teams and system administrators​

• Inventory hardware and prioritize machines with hybrid Intel architectures for driver/firmware validation.
• Test Windows 11 24H2 + 25H2 enablement package on representative fleets, especially where CPU‑bound background work is important.
• Use controlled rollouts and telemetry collection to identify reproducible regressions before broad deployment.
• If needed, budget for Windows 10 ESU as a controlled migration buffer while drivers and OS fixes converge. (microsoft.com)

---

What to watch next​

• Vendor statements and firmware updates from Intel and OEMs clarifying whether specific microcode or BIOS patches address the minimized‑window throttling.
• Microsoft cumulative updates that explicitly reference scheduler, power policy, or background throughput fixes in release notes.
• Enterprise feedback and broader benchmark sets on 25H2 once it reaches general availability; early independent testing shows little raw CPU uplift vs 24H2, so real improvements must come from targeted fixes at the shell or policy level. (tomshardware.com)

---

Conclusion — the migration decision will be won by polish, not features​

The programmer’s discovery reframes the Windows 10 vs Windows 11 speed conversation in a useful way: many of the performance complaints are not mystical cultural preferences but traceable interactions between OS prioritization, power policy, and platform firmware — and those are fixable. Microsoft has already shipped patches that improve File Explorer responsiveness and is collecting richer diagnostic telemetry; 25H2 is positioned to be a low‑risk enablement release rather than a performance panacea. (support.microsoft.com)
For users and IT managers, the pragmatic path is clear: test and validate on representative hardware, apply the latest cumulative updates (including KB5055627 fixes where applicable), collaborate with OEMs on firmware/driver updates, and use ESU only as a planned buffer—not a long‑term excuse to postpone a careful migration. If Microsoft, OEMs, and driver vendors coordinate fixes for the minimized‑window and Explorer regressions, Windows 11’s modern UX and security advances will be easier to recommend. If not, Windows 11 risks lingering in Windows 10’s shadow for users who prize day‑to‑day snappiness over polished visual features. (microsoft.com)

---

(If you’re tracking this issue, focus on these milestones: Microsoft’s monthly cumulative releases and KB notes, OEM BIOS/chipset driver updates, and the general availability telemetry post‑25H2 rollout. Those are where the practical fixes will appear.)

---

Source: Glass Almanac A programmer found out why Windows 10 outperforms Windows 11