
Four gigabytes of RAM was generous in 2014. In 2026, it is a constraint that shapes every aspect of how a laptop behaves. Millions of machines in the refurbished and hand-me-down market ship with 4 GB — sometimes soldered, sometimes upgradeable — and the question is always the same: can it actually do the things I need?
This page answers that question with specific numbers. We tested three 4 GB laptops across Windows 10, Windows 11, and Linux Mint Xfce to find the practical limits: how many browser tabs before the experience degrades, whether 1080p video playback works under memory pressure, and how each operating system's swap behaviour differs. The goal is not to declare 4 GB "good" or "bad" — it is to show exactly where the boundary falls on each OS so you can make an informed decision. For the broader resource comparison with 8 GB machines, see our Windows 10 ESU vs Windows 11 vs Linux benchmark. For all performance data, visit the Benchmarks hub.
Test Protocol and Hardware
Each OS was clean-installed on all three machines with a SATA SSD (Crucial MX500 250 GB) to eliminate the HDD bottleneck — we wanted to isolate the RAM constraint, not the storage one. All installations were fully updated as of April 2026. Browser tests used Chrome 124 on Windows and Firefox 126 on Linux (each platform's default/dominant browser). Each tab loaded a reference set of pages: a news article (BBC), a web app (Google Docs), a media-rich page (Reddit front page), a video page (YouTube), and additional plain text pages.
Tabs were opened one at a time with 15-second intervals to allow each to fully render. Memory and swap were recorded after each tab. The "swap thrashing threshold" was defined as the point where the system became noticeably unresponsive — typing in the URL bar lagged by more than 2 seconds, or switching tabs took more than 5 seconds.
| Machine | CPU | RAM | Storage | RAM Upgradeable? |
|---|---|---|---|---|
| 2013 Dell Latitude E5440 | Core i5-4310U (2C/4T, 2.0 GHz) | 4 GB DDR3L (1 slot free) | SATA SSD 250 GB | Yes — 1 SODIMM slot available |
| 2015 Lenovo ThinkPad T450s | Core i5-5300U (2C/4T, 2.3 GHz) | 4 GB DDR3L (soldered + 1 slot) | SATA SSD 250 GB | Partially — can add to reach 12 GB total |
| 2016 HP EliteBook 820 G3 | Core i5-6200U (2C/4T, 2.3 GHz) | 4 GB DDR4 (1 slot free) | SATA SSD 250 GB | Yes — 1 SODIMM slot available |
All three are common refurbishment models that frequently ship with 4 GB configurations. Windows swap was set to system-managed (default). Linux swap was a 2 GB swap partition, matching the common automatic installer default.
Browser Tab Limits
The table below shows how many tabs of our reference page set each machine could hold open before reaching the swap thrashing threshold. "Comfortable" means under 500 MB swap and responsive. "Degraded" means 500 MB - 1 GB swap with noticeable lag. "Thrashing" means over 1 GB active swap with the system effectively unusable for real work.
| OS / Browser | Comfortable | Degraded | Thrashing |
|---|---|---|---|
| Windows 10 / Chrome | 5 tabs | 6-7 tabs | 8+ tabs |
| Windows 11 / Chrome | 3 tabs | 4-5 tabs | 6+ tabs |
| Linux Mint / Firefox | 10 tabs | 11-14 tabs | 15+ tabs |
Memory breakdown at 5 tabs — Windows 10
OS idle: 2.1 GB
Chrome (5 tabs): 1.4 GB
Total committed: 3.5 GB
Swap active: 320 MB — usable but tight
Memory breakdown at 5 tabs — Windows 11
OS idle: 2.8 GB
Chrome (5 tabs): 1.4 GB
Total committed: 4.2 GB
Swap active: 890 MB — noticeable lag when switching tabs
Memory breakdown at 10 tabs — Linux Mint
OS idle: 680 MB
Firefox (10 tabs): 2.6 GB
Total committed: 3.3 GB
Swap active: 180 MB — fully comfortable
Video Playback Under Memory Pressure
We tested 1080p YouTube playback in two scenarios: with no other tabs open (best case) and with 4 background tabs open (realistic light use). The metric was dropped frames over a 5-minute 1080p 30fps video clip.
| OS / Scenario | Dropped Frames (5 min) | Swap Active | Subjective Quality |
|---|---|---|---|
| Windows 10 / Video only | 2 | 140 MB | Smooth |
| Windows 10 / Video + 4 tabs | 38 | 780 MB | Occasional stutters |
| Windows 11 / Video only | 8 | 410 MB | Smooth with brief hiccup |
| Windows 11 / Video + 4 tabs | 124 | 1.3 GB | Unwatchable stuttering |
| Linux Mint / Video only | 0 | 0 MB | Smooth |
| Linux Mint / Video + 4 tabs | 4 | 210 MB | Smooth |
The results confirm what the tab testing suggested. Linux Mint handled video playback alongside four tabs with negligible swap and near-zero dropped frames. Windows 10 managed single-video playback well but degraded with background tabs. Windows 11 was already in swap with video alone and became unwatchable when tabs were added.
The dropped frame count on Windows 11 with tabs — 124 in five minutes — means roughly one visible stutter every 2.4 seconds. That is not a marginal issue. It renders the machine unsuitable for video consumption while doing anything else, which is a common use case for a secondary or family laptop.
Swap Behaviour: How Each OS Manages Scarcity
The three operating systems handle memory pressure differently, and those differences matter significantly at 4 GB.
Windows 10 begins paging to swap relatively early — we saw swap activity start at approximately 75% physical RAM usage (around 3 GB committed). The swap is gradual and the system remains responsive during light swap. Task Manager showed the pagefile growing smoothly from 0 to 800 MB as tabs were added.
Windows 11 uses more aggressive memory compression (introduced in Windows 10 but expanded in 11), which trades CPU cycles for reduced swap volume. In theory this helps. In practice, on these dual-core machines, the compression CPU overhead competed with the browser's rendering thread, creating a double penalty — the system was both swapping and consuming extra CPU to compress pages.
Linux Mint with default swappiness (60) began swapping later — at roughly 85-90% physical RAM usage. The smaller OS footprint meant this threshold corresponded to far more application data in memory. When swap did activate, the SSD-backed swap partition handled it cleanly with minimal impact until the 1 GB active swap mark.
Swap onset comparison (same workload: browser + 5 tabs)
Windows 10: Swap begins at 5 tabs (~3.5 GB committed)
Windows 11: Swap begins at 3 tabs (~3.8 GB committed)
Linux Mint: Swap begins at 12 tabs (~3.5 GB committed)
The physical threshold is similar — around 3.5 GB committed. The difference is how much of that 3.5 GB is OS versus application.
What the Numbers Do and Do Not Prove
These results prove that 4 GB is a meaningful constraint in 2026 on Windows but manageable on lightweight Linux. They prove that OS idle footprint directly determines how much memory is available for user tasks, and that this difference is the primary factor in tab counts and multitasking capability.
They do not prove that Chrome is worse than Firefox in absolute terms — Chrome on Linux would use similar per-tab memory to Chrome on Windows. The comparison reflects the typical default browser on each platform. They do not prove that 4 GB is "enough" for Linux in all scenarios — running development tools, VMs, or image editing on 4 GB will strain any OS. And they do not account for the RAM upgrade option: if your machine has a free SODIMM slot, spending £10-20 on a DDR3L module to reach 8 GB eliminates the entire constraint more effectively than any software optimisation.
Recommendation by Situation
| Situation | Action | Expected Outcome |
|---|---|---|
| RAM is upgradeable | Add RAM to 8 GB (£10-20 for DDR3L) | Eliminates the 4 GB constraint entirely; all three OSes run comfortably |
| RAM is soldered, need Windows | Stay on Windows 10 + aggressive startup cleanup | 5-7 comfortable tabs, workable for light use; Windows 11 is too tight |
| RAM is soldered, browser/office use | Switch to Linux Mint Xfce | 10+ comfortable tabs, smooth video, genuine daily-driver usability |
| Still on HDD | SSD swap first, then evaluate RAM | SSD makes swap tolerable; HDD + 4 GB is the worst combination |
The startup fix guide is particularly relevant for 4 GB Windows machines — every background process you eliminate frees RAM for your actual work. For the full OS comparison on machines with more headroom, see the Windows 10 ESU vs Windows 11 vs Linux benchmark. And for the broader question of whether a constrained machine is worth investing in at all, our assessment of which laptops are worth saving can help. If you are considering Linux but unsure which distribution to try, the lightweight Linux guide covers the practical options for resource-constrained hardware.