I stopped disabling this Windows feature and gained back the performance I thought I’d lost

I’ve spent enough time around Windows to know my way around the settings that most people never touch. Memory compression is one of those features I have always understood well enough to leave alone, but last year, curiosity got the better of me. The question was simple: if memory compression trades CPU cycles for RAM efficiency, would a system with plenty of RAM actually run better without it?

There was only one way to find out. I opened PowerShell as an administrator, ran Get-MMAgent to confirm the feature was on, then followed it with Disable-MMAgent -mc and restarted my PC. What came after, though, was not quite the performance jump I had been hoping for. In fact, the system felt subtly worse under load, and that pushed me to dig deeper, and eventually led me straight back to the command that started it all: Enable-MMAgent -mc.

Disabling memory compression is not the performance tweak it sounds like

In defense of the background squeeze

The thinking behind my little experiment was that memory compression shrinks data before it’s stored in RAM, which means you can fit more into the same physical space and lean less on the page file. The catch is that your CPU has to do extra work, compressing data on the way in and decompressing it when you need it again. So on a system with plenty of RAM, it starts to sound a bit unnecessary. If there’s already enough room, why bother compressing anything at all?

That line of thinking misses what actually happens when memory starts to fill up. Windows doesn’t immediately dump data to disk. It first compresses inactive memory and keeps it in RAM. Only when that compressed space runs out does it start pushing data to disk. And the page file is the real slowdown here, not compression. Because CPUs are now incredibly fast, decompressing data already in RAM happens quickly. Pulling that same data from a page file, even on a fast SSD, takes microseconds. That’s roughly a 1,000x difference in response time — even the fastest SSD still can’t compete with RAM sitting right next to the CPU.

To put numbers to it: imagine you have 14GB of active data in RAM on a 16GB system. With compression enabled, Windows might squeeze 3GB of inactive, highly compressible data (browser sessions, background app state, cached files) down to around 1.5GB, keeping your total footprint at roughly 15.5GB — comfortably within physical RAM. Without compression, that same 3GB sits uncompressed, pushing your total closer to 17GB and forcing the overflow to the page file. While some enthusiasts argue for turning virtual memory off entirely, disabling the page file actually makes Windows less stable, so that overflow is inevitable. The end result was subtle but noticeable. Multitasking felt heavier, and switching back to previously opened apps sometimes involved a brief pause that was not there before.

You can watch this play out in real time by opening Task Manager with Ctrl + Shift + Esc, navigating to the Performance tab, and selecting Memory. (If you want a more detailed breakdown, using Resource Monitor instead of Task Manager can help you understand exactly what is eating your RAM, but Task Manager works well enough for a quick glance). The “In use (Compressed)” figure shows how much of your RAM is currently holding compressed data. On a busy system, that number represents memory compression actively keeping your workload off the disk. When I disabled the feature via PowerShell and restarted, that value dropped to zero. Consequently, the overall memory footprint of my open applications climbed noticeably as the data “expanded” to its full size, and the page file was forced to pick up the slack to keep the system stable.

Knowing when to flip the switch and when to leave it alone

Treat the disease, not the symptom

After two weeks without compression and no meaningful gains to show for it, I re-enabled the feature. I opened PowerShell as an administrator, ran Enable-MMAgent -mc, restarted, then confirmed the change with Get-MMAgent once I was back on the desktop. The difference was not explosive, but it was immediate in the same subtle way the slowdown had been. Disk activity settled down; previously opened apps resumed more quickly, and browser tabs stopped refreshing as often. The system felt lighter again during heavy multitasking, which made it clear that the performance I thought I had gained by disabling compression had never really existed in the first place.

The case for leaving compression disabled only makes practical sense on systems with at least 32GB of RAM, where “memory pressure” is almost never reached. Even then, the benefit depends on the processor’s capabilities and the nature of the workloads being run. On an older or weaker CPU already running close to its ceiling, removing the compression overhead might create a little more room. However, on my system with 16GB and a capable modern processor, I did not observe any improvement in application launch time, browser tab switching, or file extraction speed with compression disabled. If anything, the system lost its ability to absorb memory pressure gracefully.

The only notable exception to this rule is within virtual machines (VMs). In virtualized environments, it is often better to disable compression within the “guest” OS to allow the “host” hypervisor to manage memory deduplication and allocation more efficiently. But for a standard desktop or gaming rig, if compressed memory is causing slowdowns, the solution is to install more physical RAM rather than disabling the feature and letting the page file absorb the difference. Even ensuring your Windows page file is moved to the right drive is a another alternative to mitigate system stutter. Disabling compression merely treats the symptom while worsening the underlying performance condition.

I gained back performance by reversing the tweak

The experiment was worth running. It confirmed something that is easy to forget when you are deep in optimization mode: Windows’ default memory behavior is not accidental. It holds up under real conditions precisely because it was designed around them, even when the theory suggests a smarter path forward.

Curiosity is usually worth following, especially when it leads somewhere instructive. My compression experiment did not deliver faster performance, but it gave me a tested understanding of exactly what the feature is doing and why removing it made my system feel slightly worse instead of better. The performance I thought I had gained by disabling memory compression only really returned when I stopped fighting the feature and let Windows handle memory the way it was designed to.