What was tested
The rig has a Ryzen 7 9800X3D and a GeForce RTX 5070. It used one G.Skill DDR5-6000 kit throughout, and only the installed capacity changed between runs. The author checked average FPS and memory use across four games at 1080p. The report names three of them: Wardogs, Cyberpunk 2077 and Marvel Rivals.
No frame-rate tables, memory logs, test durations or graphics settings were published. The fourth game isn't named. Read the findings as practical impressions, not measured data.
16GB versus 32GB: it depends on what else is running
With few background apps, FPS at 16GB and 32GB was "virtually the same" in the three named games. Task Manager memory use reached about 90% during intense moments at 16GB, but the games stayed stable. The games were also installed on a fast NVMe drive, which the author suggests may have masked problems that a slower drive would show.
The picture changed once the author added their normal background load: Chrome with about six tabs, Discord, Spotify and a few other apps. At 16GB the author saw frame dips during busy scenes and brief stutters when entering new areas. At 32GB those symptoms went away. In a 100-player game with about a dozen apps open, at least 10GB stayed free.
Those figures point to one lesson. Raw FPS may not move between capacities, but smoothness under a realistic desktop workload can.
64GB: no gaming gain
At 64GB, the author reports identical performance to 32GB in the tested games, because 32GB never came close to filling. The author kept 64GB for a different reason. The PC doubles as a development machine that runs several virtual machines, Docker containers and a local LLM. That is a capacity case driven by non-gaming work.
The article mentions anecdotes that Microsoft Flight Simulator 2024 can exceed 32GB with heavy add-ons, but it doesn't substantiate them. Microsoft's FAQ lists 16GB as minimum, 32GB as recommended and 64GB as ideal. Those tiers also step up the CPU and GPU, and they are requirement categories, not measured usage. The Xbox store page lists 32GB in its PC requirements.
Why "90% used" doesn't prove paging
The author attributes the 16GB stutter to Windows shuffling data between RAM and the SSD page file. That's plausible, but the article didn't measure it, and a high RAM percentage alone doesn't prove paging caused any particular hitch.
Microsoft documents a related concept worth checking. The system commit limit is the sum of physical memory and all page files combined. Commit charge is the total memory Windows has promised to all committed virtual memory. If commit charge reaches the limit, the system and processes might not get committed memory. Microsoft's documentation says this can cause freezing, crashes and other malfunctions.
Microsoft's page is about commit behavior, not frame times. It doesn't promise anything about gaming performance.
How to check your own PC
Reproduce your normal gaming session, then look at the numbers before buying anything:
- Open Task Manager and go to Performance > Memory. Watch the Committed readout, which shows commit charge against the commit limit. The first number is current committed memory and the second is the commit limit.
- Optionally log the
\Memory\Committed Bytesand\Memory\Commit Limitcounters. Microsoft says both Task Manager's Performance tab and these counters measure them. - Play once with your usual apps open, then again with the browser and chat apps closed. If stutter disappears, background load is the issue, and more RAM is a plausible fix.
- Note the peak values, not just the idle ones. If your peak commit stays well under your installed RAM, extra capacity probably won't help.
Two caveats apply. A heavily used page file isn't automatically a problem. Microsoft says a page file showing 100 percent usage isn't a performance issue as long as the commit limit isn't reached. That claim comes from a secondary summary of Microsoft's guidance. Also, a Committed reading such as "18/32GB" does not mean 18GB sits in the page file.
What to buy
The author recommends 32GB for new gaming PCs. They treat 16GB as a reasonable stopgap while DDR5 prices are inflated, and they suggest upgrading when prices ease or when hitches appear. The article gives no price data, so that is budget advice, not a verified market assessment.
The "upgrade later" plan also assumes a free memory slot and a compatible kit. The article doesn't specify either, so check your motherboard before relying on it.
| Capacity | Reported result | Best fit |
|---|---|---|
| 16GB | Playable, but little headroom with background apps | Tight budgets, mostly gaming and light browsing |
| 32GB | Smooth with a typical background load | Most new gaming builds |
| 64GB | No added gaming performance | VMs, containers, local LLMs, heavy creative work |
Limits of the evidence
- One system, one resolution (1080p) and four games, only three of them named.
- No published FPS or memory data.
- A fast NVMe drive may have hidden penalties that slower storage would show.
- Higher resolutions, heavily modded games and very large open-world titles weren't covered.
These findings fit the general rule that extra RAM only helps once you run short of it. They don't show that 16GB is always enough or that 32GB can never be outgrown. Measure your own workload first.
References
- I tested 16GB, 32GB, and 64GB of RAM in the same gaming PC, and found out when more memory actually matters XDA · 2026-11-21T12:01:22+00:00
- Microsoft Flight Simulator 2024 | XBOX xbox.com
- Introduction to the page file - Windows Client | Microsoft Learn learn.microsoft.com