High memory usage in Windows 11 — Task Manager showing 70–90% RAM in use constantly, the machine feeling sluggish, applications taking long to open — is worth understanding before trying to fix it. Not all high memory usage is a problem. For a broader walkthrough, our Complete Guide to Fixing Windows, Browser, and Software Errors is a good next read.
Windows intentionally uses available RAM as a cache for recently accessed files and frequently used applications. “Used” memory in Task Manager includes this cache. Memory that’s “In use” (the solid colour in the graph) is actively needed. Memory that’s “Standby” (lighter colour) is cached data that can be freed instantly if applications need it. If your RAM shows 70% used but the machine is responsive, it’s working correctly — not wasting memory.
The actual problem is when memory pressure causes paging — reading from the hard drive when RAM runs out. Task Manager → Performance → Disk → consistently high disk activity (above 10 MB/s) without obvious file transfers, combined with high RAM: that’s paging, and that’s when to act.
Find what’s actually consuming memory
Task Manager → Details tab → sort by “Private Working Set” (right-click the column headers → Select columns → add Private Working Set). This shows real memory consumption per process, not shared memory. Processes consistently above 500 MB are worth investigating.
Common high-memory culprits: browsers with many tabs (Chrome and Edge allocate per-tab), memory-leaking applications that grow over time, antivirus with aggressive scanning, VMs running in the background, and cloud sync clients with large local indexes.
Browser tabs
The fastest win: reduce open tabs. Each Chrome or Edge tab is a process. 30 tabs easily consumes 3–4 GB. Enable Memory Saver: Chrome/Edge Settings → Performance → Memory Saver → On. This suspends inactive tabs and reduces active memory footprint significantly. On 8 GB machines, enabling Memory Saver can free 1–2 GB immediately.
Startup applications accumulating
Applications that launch at startup stay in memory for the entire session. Many install startup entries without prominently informing users. Over months and years, a machine’s startup list grows — and so does its baseline memory usage.
Task Manager → Startup apps → sort by Status (Enabled first) → identify applications you don’t need running at startup. Right-click → Disable for any that don’t need to start automatically. Cloud clients, communication apps, update helpers, and manufacturer utilities are common unnecessary startup items on machines that don’t actively use those services.
After disabling startup items: restart → check memory usage before opening anything. This shows the true baseline. A clean machine should start with 30–40% RAM used on 8 GB, lower on 16 GB.
Virtual memory and page file
When physical RAM is fully used, Windows writes memory contents to the page file on disk — virtual memory. If the page file is too small or on a slow drive, this causes severe performance degradation. By default, Windows manages the page file size automatically.
Check: System → Advanced system settings → Performance → Settings → Advanced → Virtual memory → change → confirm “Automatically manage paging file size for all drives” is checked. If it was manually set to a small value, this explains persistent memory pressure. Letting Windows manage it automatically ensures the page file grows as needed.
On systems with an SSD: the page file on SSD is much faster than on an HDD. If the system drive is an SSD and additional storage is an HDD: ensure the page file is on the SSD, not the HDD.
Memory leaks
Applications with memory leaks grow their memory usage over time without releasing what they no longer need. Characteristic pattern: memory usage at 50% after startup, 65% after 2 hours, 80% after 4 hours, without opening more applications. A specific process grows steadily in Task Manager → Details.
Identify the leaking process by watching Task Manager over a few hours. After identifying it: update the application (memory leaks are often fixed in updates), restart it periodically, or if the application is dispensable, close it when not actively needed. Restarting the machine clears all memory leaks temporarily, but they return — the application itself needs to be fixed or replaced.
Windows services consuming memory
Win + R → services.msc → some services run at startup and consume memory for features you may never use. Services worth reviewing for disabling on personal machines: SysMain (Superfetch — pre-loads applications; on SSDs the benefit is minimal), Windows Search (if you use a third-party search tool instead), Connected User Experiences and Telemetry, and Print Spooler (if no printer is used). Disabling unneeded services frees both RAM and some CPU.
Warning: only disable services you understand. Disabling the wrong service can cause instability. Research each service before disabling it — the SysMain and Print Spooler are the safest to disable on SSD machines without printers.
RAM itself might be the issue
On 4 GB machines: high memory usage is expected. Windows 11 with basic applications fills 4 GB readily. The fix there is hardware — 8 GB minimum is the practical threshold for Windows 11 with any meaningful work. Upgrading from 4 to 8 GB typically costs £25–50 in laptop RAM and transforms the experience more than any software optimisation.
On 8 GB machines feeling pressured: the fixes above (startup apps, browser tabs, Memory Saver) provide meaningful relief. On 16 GB: genuine high memory pressure usually indicates a specific leaking application rather than capacity issues.
Our guide on Windows 11 slow startup covers startup application management in more depth, including services that affect boot performance alongside memory — the same startup items that affect boot also affect baseline memory usage throughout the session. The high disk usage guide covers the paging and indexing activity that accompanies genuine memory pressure. Microsoft’s Windows 11 memory management documentation covers the Resource Monitor (resmon.exe) which provides more detailed memory breakdown than Task Manager, including the hard fault rate that confirms whether paging is actually occurring.
Using Resource Monitor for deeper analysis
Resource Monitor (search in Start → resmon.exe) provides more detail than Task Manager. The Memory tab shows “Hard Faults/sec” — each hard fault is the machine reading from the page file rather than finding data in RAM. Above 10 hard faults/sec during normal use indicates active paging, which explains slow performance. This number alone confirms whether memory pressure is actually causing the performance issue versus just the RAM gauge being high.
The Memory tab also shows “Commit Charge” — the total virtual memory committed across all processes. If Commit Charge significantly exceeds physical RAM: more memory is committed than physically exists, and the system relies heavily on the page file. Reducing Commit Charge through closing applications is the immediate fix; adding RAM is the permanent one.
Antivirus memory footprint
Security suites vary dramatically in their RAM consumption. Some products routinely use 400–600 MB; others run at under 100 MB. On machines with 8 GB, a bloated antivirus consuming 600 MB represents 7.5% of total RAM as overhead before any user applications open.
Malwarebytes and Microsoft Defender are among the lighter options. If the antivirus is consuming significant memory (check the specific processes in Task Manager → Details — look for the product’s service names), switching to a lighter product is a legitimate RAM conservation strategy, particularly on 8 GB machines. Microsoft Defender (built into Windows 11) provides solid protection with a lighter footprint than most commercial alternatives.
Virtual machines and container applications
If you run Hyper-V, VirtualBox, VMware, or Docker: each VM or container requires dedicated RAM that Windows cannot reclaim. A VM configured with 4 GB RAM consumes that 4 GB whether the VM is busy or idle, leaving 4 GB for everything else on an 8 GB machine. Reduce VM memory allocation when not actively using the VM’s full capacity, or shut down VMs entirely when not needed. Dynamic memory allocation in Hyper-V (adjusting VM RAM based on demand) helps but doesn’t eliminate the overhead entirely.
WSL2 and developer environment memory
WSL2 (Windows Subsystem for Linux) runs a full Linux kernel in a lightweight VM. By default, it can claim up to 50% of physical RAM (capped at 8 GB). If WSL2 is installed and used: it may be consuming significant RAM even when not actively being used for tasks.
Check WSL2 memory: Task Manager → look for “Vmmem” process — this is WSL2’s memory usage. Limiting WSL2 memory: create a .wslconfig file in the user’s home directory (C:Users[username].wslconfig) with:
[wsl2]
memory=4GBRestart WSL2 (wsl --shutdown in Command Prompt). This limits WSL2 to 4 GB, leaving more for Windows. Adjust the value based on actual WSL2 needs.
RAM hardware issues
Memory module problems can cause Windows to report incorrect memory usage — the OS reserves memory it detects as problematic, reducing available RAM. If only some of the installed RAM appears available (e.g., 16 GB installed but only 12 GB available), Windows may have remapped problematic memory regions.
Check: Task Manager → Performance → Memory → “Installed” vs “Available.” A significant gap between installed and available (more than 100–200 MB for hardware-reserved) warrants investigation. Also check BIOS → should show the correct total. MemTest86 identifies failing memory modules. On dual-channel setups (two RAM sticks), removing one stick and testing with each individually identifies whether a specific module is failing.
Third-party memory “optimization” software
Tools that claim to “optimize” or “free” RAM — RAMMap-style tools that forcibly clear the standby list — often make performance worse rather than better. Standby memory is cache that Windows will use productively. Clearing it means Windows has to reload data from disk the next time it’s needed, increasing load times and disk activity. True freed memory (completely unused, not cached) is what actually improves performance when low.
These tools show a satisfying “RAM freed” number but are trading cached data (fast) for available memory (potentially unused) — a net performance loss. Windows manages its own memory cache intelligently. Third-party RAM management tools generally don’t improve this management and often interfere with it. Avoid them.
Memory-heavy application settings
Individual applications can be configured to use less memory. Specific examples:
- Visual Studio Code: Extensions can each consume 50–200 MB. Disable unused extensions.
- Slack desktop app: Settings → Advanced → “Store workspace data” → disable offline message storage to reduce Slack’s local cache memory use
- Spotify: Settings → Quality → download cache size → reduce from default
- Adobe Creative Cloud: reduces background process memory via Creative Cloud app → Preferences → General → disable “Launch Creative Cloud at login”
Application-specific memory reduction, aggregated across all running applications, often reclaims 500 MB – 1 GB on machines with many applications installed and configured for heavy use.
64-bit vs 32-bit application memory limits
32-bit applications are limited to 4 GB of RAM regardless of how much the system has. For memory-intensive work — video editing, large databases, complex spreadsheets — a 32-bit application hitting its limit causes “out of memory” errors and slow performance even when the machine has ample RAM. Check whether the application causing memory issues is 32-bit: Task Manager → Details tab → right-click any process → Select columns → Platform shows “32-bit” or “64-bit.” If the heavy application is 32-bit, the 64-bit version handles large workloads correctly. Most modern applications are available in 64-bit versions.
The practical impact: a 32-bit Photoshop or video editor will run out of its own memory allocation before Windows runs out of system RAM. Windows shows RAM as “available” but the application shows “out of memory” errors. Switching to the 64-bit application version resolves this immediately and typically improves performance on those workloads as well.
Determining the right amount of RAM to add
If the fixes above improve the situation but memory pressure persists, the practical question is: how much RAM do you actually need? Resource Monitor → average “In Use” memory during your typical workload, then add 30% for headroom. If typical In Use is 10 GB: 16 GB is the minimum and 32 GB provides comfortable headroom. Buying exactly enough RAM without headroom means hitting the limit as soon as any additional task runs — the headroom matters.
RAM upgrade compatibility: laptop RAM type (DDR4 vs DDR5, SO-DIMM vs SO-DIMM slots) must match the board specification. Crucial’s System Scanner (crucial.com) identifies compatible upgrades by scanning the installed system. Check whether the laptop has user-accessible RAM slots before purchasing — many thin laptops have RAM soldered to the motherboard and cannot be upgraded. The laptop’s service manual (available from the manufacturer) specifies which RAM slots, if any, are user-accessible.
To summarise the approach: high memory numbers in Task Manager alone aren’t a problem — hard fault rate in Resource Monitor and actual paging disk activity are what matter. If those are low and the machine is responsive: the memory is being used productively, and no action is needed. If paging is occurring: reduce the processes consuming memory (startup apps, browser tabs, idle VMs, WSL2) or add RAM. Memory “optimization” tools are counterproductive. The Task Manager → Details → Private Working Set column is the most useful diagnostic tool for identifying which specific processes are the biggest consumers and whether any are growing abnormally over time. Related: Chrome High Memory Usage.
