Free · No signup · Runs in your browser
Bottleneck Calculator: Check Your CPU and GPU Bottleneck
Free bottleneck calculator for your CPU, GPU and RAM. See which part limits your PC at your resolution, an FPS range, and the upgrade that closes the gap.
- Instant result
- Nothing uploaded
- CPU + GPU + RAM + Storage
Average FPS by resolution
Recommended power supply
Upgrade suggestion
Estimates compare relative benchmark scores at your chosen settings. Real frame rates vary by game, drivers and system configuration. How this score is calculated, and when to trust it.
How to Use This Calculator (Step by Step)



- Choose your CPU: Use the exact model number, “Ryzen 5 7600,” not just “Ryzen 5.” Don’t know it? Press Windows + R, type dxdiag, hit Enter; your processor is listed on the System tab.
- Choose your GPU: Include the full name; variants matter (an RTX 4060 and RTX 4060 Ti score differently). Check the Display tab in the same dxdiag window if you’re not sure.
- Select resolution: Use what you actually play at, not your monitor’s maximum supported resolution.
- Set the extras if you know them: Target refresh rate, use case, RAM capacity, type, speed and channels, storage, upscaling, ray tracing, streaming and background apps. The defaults (144 Hz, general gaming, 16 GB of DDR4-3200 in dual channel, an NVMe Gen3 drive) fit a typical gaming PC.
- Click Calculate bottleneck: The result appears on the same page. There is nothing to download and no account, and the calculation runs in your browser.
- Read your results: You get a bottleneck percentage with a balance rating, the limiting part, a grade, a frame-time rating, an FPS range and an upgrade suggestion.
A result under 12% means you are set. A result over 22% points to a specific upgrade worth planning for, covered next.
What is a Bottleneck Calculator?
A free online tool that compares your CPU and GPU (and usually RAM, storage, and resolution) to show which part is limiting your system’s performance, expressed as a percentage.
What Percentage is Considered Good?
Under 12% earns the calculator’s Excellent or Very Good Balance rating and needs no action. From 12% to 22% is rated Good: a gap a benchmark can see but you rarely feel. Above 22% the rating drops to Fair, and the weaker part starts to hold the other back in demanding games.
Is a 10% Bottleneck Bad?
No. The calculator rates 10% as Very Good Balance. Some imbalance shows up in almost every build, since CPUs and GPUs are never released on matched schedules, and you will not feel a 10% gap while playing.
How Accurate is this Calculator?
It is reliable about direction and size, less so about exact frame rates. Each part carries one benchmark score, so the result dependably tells you which part is weaker and whether the gap is small or large. It does not measure your PC or a specific game, so treat the FPS range as a ballpark. Every step of the method is listed in How the Bottleneck Calculator Works below, and the longer answer is in are bottleneck calculators actually accurate.
How Much Bottleneck Is Safe?
The percentage from the calculator above is a gap, not a verdict. The calculator puts a label on every result, and the labels group into four practical bands.
Under 12% (Excellent or Very Good Balance): The CPU and GPU are close enough that the game will not feel limited by either one. Leave the build alone.
12–22% (Good Balance): One part is a little ahead. A benchmark can show it. You usually cannot feel it while playing, so this is not a reason to buy anything yet.
22–35% (Fair Balance): The weaker part is holding the other back in demanding games. Plan an upgrade for that part, sooner if you also see stutter, not just a lower average frame rate.
35% and above (Noticeable or Severe Bottleneck): The weaker part is costing real frames in most games. Replace it before you spend money anywhere else.
The percentage is not a property of the parts alone. The same pair can sit under 10% at 4K and over 20% at 1080p, because lower resolutions ask more of the CPU. Read the number at the resolution you play, then confirm it on the machine with a real usage check before you spend money.
Bottleneck Checker vs Bottleneck Test: Same Thing?
Same question, two tools: A bottleneck checker is the form at the top of this page. You pick a CPU and a GPU and get an estimate. A bottleneck test is a measurement on a PC you already own, usually CPU and GPU usage in Task Manager or MSI Afterburner while a game is running.
Checker: Use it before you buy, or when you want a number without installing anything. It compares benchmark scores. It cannot see a hot case, a single stick of RAM, or a driver bug.
Test: Use it on the PC in front of you. If the GPU sits near 99%, the graphics card is the limit. If the GPU stays well under that and the frame rate will not climb when you lower the settings, the CPU is the limit. A browser page that only animates a graphic and calls itself a bottleneck test does not know which parts are in the PC.
People search both phrases for the same job. Run the checker above, then use the steps in how to check a bottleneck without a calculator if the PC is already built.
Will My CPU Bottleneck My GPU?
It can, and resolution decides it: A much faster GPU paired with an older CPU is the usual case, most often at 1080p, where the processor has to prepare a lot of frames and the graphics card is not yet full. At 1440p and 4K the same CPU often stops being the limit, because the GPU is busy drawing more pixels.
You will not get a yes from the names alone. Select both parts in the calculator at the top of this page, at the resolution you play. If the result names the CPU and the gap is above 20%, the processor is the one to plan around. In a game, a CPU limit shows as low GPU usage with one or two busy CPU cores, and you can estimate the frame rate for the same pair. Whether the card physically fits, powers on and posts is a separate compatibility check.
Pairing guides for popular parts: If you already own one half of the build, these guides run that part against every common match in the same calculator. If you own the graphics card, there are CPU pairings for the RTX 3060, RTX 4060, RTX 4070, RTX 4080, RTX 4090, RTX 5060 Ti, RTX 5070, RTX 5070 Ti, RTX 5080 or RX 9070 XT. If you own the processor, there are GPU pairings for the Core i5-12400F, Core i5-13600K, Ryzen 5 5600, Ryzen 5 7600X, Ryzen 7 7800X3D or Ryzen 7 9800X3D. On an older Ryzen board, a drop-in AM4 upgrade may be all you need.
Comparing parts before you buy: The GPU tier list ranks 44 graphics cards by measured FPS. CPU compare and GPU compare put two parts side by side, and the RAM latency calculator turns a kit’s speed and CAS number into nanoseconds.
Why Your CPU and GPU Can Hold Each Other Back
Your CPU and GPU work as a two-step pipeline. The CPU works out what needs to happen next, physics, AI behavior, draw calls, then hands that off to the GPU, which turns it into a rendered frame. If your GPU is capable of 200 frames per second but your CPU can only prepare 90 sets of instructions per second, you’re capped at 90 FPS. The GPU spends the rest of its time idle, no matter how powerful it is on paper.
Flip the pairing and the opposite happens: a fast CPU paired with a weak GPU finishes its calculations early and waits while the graphics card struggles to catch up.
A bottleneck calculator puts a number on that gap. Instead of guessing whether your build is “good enough,” you get a percentage, a plain answer for which component is limiting you, and roughly by how much.
Most calculators worth using, including this one, go beyond CPU and GPU. Adding RAM, storage type, and target resolution sharpens the estimate, because a bottleneck at 1080p can look completely different at 4K.
Pro tip: Run the calculator twice, once at your current resolution and once at a resolution you’re considering moving to (say, 1080p to 1440p). Buying a monitor upgrade shifts your bottleneck just as much as buying new hardware does.
How the Bottleneck Calculator Works
The calculator does not run games or read your hardware. It compares two benchmark scores and weights the gap by how much of the work your settings put on each part. Every step is listed below, so you can check a result by hand.
Step 1: Each Part Has One Score
Every processor and graphics card in the list (619 CPUs and 594 GPUs) carries a gaming score on a 0–100 scale. The scores are placed from published review benchmarks, mainly Tom’s Hardware’s CPU and GPU hierarchies. Video editing, 3D rendering and 3D design use a separate rendering score, and machine learning and AI workloads use an AI score. The fastest CPU and the fastest GPU in the list both sit at 100, so the two scores are directly comparable.
Step 2: Resolution and Use Case Split the Load
Resolution decides how much of each frame depends on the CPU. The GPU carries the rest.
| Resolution | CPU share | GPU share |
|---|---|---|
| 720p | 68% | 32% |
| 1080p | 58% | 42% |
| 1440p | 42% | 58% |
| Ultrawide 1440p | 36% | 64% |
| 4K | 25% | 75% |
| 5K | 18% | 82% |
| 8K | 12% | 88% |
The use case then moves the CPU share. Competitive gaming adds 12 points, streaming adds 10, OBS streaming while gaming adds 14, video editing adds 8, general productivity adds 20 and programming adds 30. AAA gaming takes 6 points away, 3D design 5, 3D rendering 10, and machine learning and AI workloads 20. The share always stays between 5% and 90%.
Step 3: Settings That Move the Split
- Ray tracing: Moves 6 points to the GPU, because ray-traced lighting is graphics work.
- Upscaling (DLSS, FSR or XeSS): Moves 5 points to the CPU, because the GPU renders fewer pixels.
- Target refresh rate: 165 Hz moves 4 points to the CPU, and 240 Hz or higher moves 8, because the CPU has to prepare more frames each second.
- Streaming and background apps: Add 6 and 3 points to the CPU side.
Step 4: RAM Lowers the CPU Score
RAM counts against the processor, because a CPU waiting on memory prepares frames more slowly. Less than 8 GB costs 18 points, 8 to 15 GB costs 8, single-channel memory costs 12, and a kit running well below its type’s normal speed (3200 MHz for DDR4, 5600 MHz for DDR5) costs 6. 32 GB or more, quad channel or a fast kit each give back a point or two. The penalty never goes above 30.
Step 5: The Gap Becomes the Percentage
The calculator subtracts one score from the other. The lower score is the limiting part, and the gap is multiplied by that part’s share of the load. A Ryzen 5 3600 scores 61 and an RTX 5080 scores 97, a gap of 36 points. At 1080p the CPU carries 58% of the load, so the result is 36 × 0.58 = a 20.9% CPU bottleneck. At 4K the CPU share is 25%, and the same gap becomes 9.0%.
Step 6: The FPS Estimate
The frame-rate range scales from both scores, anchored so the strongest graphics card lands near 165 FPS at 1080p. It is then divided by the extra work of the chosen resolution: 1440p counts as 1.55 times the work of 1080p, and 4K as 2.6 times. RAM penalties trim it slightly. The low figure is 78% of the average and the high figure is 128%. Use it to compare builds, not as a measured result for one game.
Storage does not change the bottleneck or the frame rate. It feeds the frame-time rating instead, where a hard drive adds the most stutter risk.
Understanding Your Results
Here is what each part of the result card means and where it comes from.

Bottleneck Percentage and Balance Rating
| Bottleneck | Rating shown | What to do |
|---|---|---|
| Under 5% | Excellent Balance | Nothing |
| 5–12% | Very Good Balance | Nothing, this is normal |
| 12–22% | Good Balance | Visible in benchmarks, rarely felt |
| 22–35% | Fair Balance | Plan an upgrade for the weaker part |
| 35–50% | Noticeable Bottleneck | Upgrade the limiting part |
| 50% and above | Severe Bottleneck | Upgrade the limiting part first |
This number isn’t fixed to your hardware alone; it moves with resolution too. The same pairing can post a very different percentage at 1080p than at 4K, see How Resolution Changes Your Bottleneck for why.
Limiting Part
The badge names the CPU or the GPU. In games, a GPU limit is the normal and preferred state: it means the graphics card you paid for is fully used. A CPU limit caps the frame rate no matter how far you lower the graphics settings. The CPU load and GPU load figures under it are the split from Step 2 of the method above.
Grade
The grade runs from A+ to F. It averages the two scores and subtracts a share of the bottleneck, so it reflects how fast the build is, while the percentage reflects how well matched it is. A balanced budget pair and a balanced flagship pair can both show a small percentage, but only the flagship earns an A+.
Frame Time
Frame time is rated from Very Smooth to High Stutter Risk. It rises with the size of the bottleneck, a slow drive (a hard drive most of all) and less than 16 GB of RAM. A build can show a small bottleneck and still carry stutter risk from storage or memory.
FPS Range
The low, average and high figures show the range for your settings. The FPS calculator adds a chart of the average at 1080p, 1440p and 4K for the same parts. Two builds can share a bottleneck percentage and still sit in very different FPS brackets.
System Efficiency
System efficiency is 100 minus the bottleneck percentage. A 6% bottleneck shows an efficiency of 94. It is the same information as the percentage, shown the other way round.
Upgrade Suggestion
The suggestion names two parts for the limiting side: a budget step, the first part with a score at least 12% higher, and a bigger step, the first with a score at least 30% higher. Laptop parts are kept out of suggestions for desktop parts. Check the board before you buy, though: the chipset decides which CPU generations it accepts, and a suggested chip from a newer generation may need a different motherboard altogether.
What Is a PC Bottleneck?
A PC bottleneck happens when one component, usually the CPU or GPU, can’t keep up with the rest of the system, capping overall performance regardless of how strong everything else is. It isn’t a defect. Excellent hardware can still bottleneck another component simply from a tier gap, as explained in the pipeline example above.
Signs you’re dealing with one:
- One component sits near 100% usage while the other stays well below it
- FPS barely improves after upgrading one part
- Stuttering or frame drops during busy, demanding scenes
- Lowering resolution doesn’t meaningfully boost FPS (often a CPU bottleneck)
- Lowering graphics settings does boost FPS significantly (often a GPU bottleneck)
Open Task Manager while gaming. CPU near 100% with GPU usage staying low points to a CPU bottleneck. Reverse the pattern, reverse the conclusion.
The 9 Types of PC Bottlenecks (Most Guides Only Cover Two)
Nine distinct types can limit your system, and most bottleneck guides online only really cover the first two.
CPU Bottleneck
The GPU idles while it waits on the CPU to feed it work. Signs: CPU near 100%, GPU noticeably lower; FPS barely changes when you lower resolution. Fix: More cores/threads, close background apps, lower CPU-heavy settings (draw distance, physics). Watching CPU and GPU usage side by side in a game confirms it.
GPU Bottleneck
The CPU finishes its work and waits on the GPU to render. Signs: GPU near 100%, CPU comfortably lower; lowering resolution or settings noticeably boosts FPS. Fix: Upgrade the GPU, lower resolution/detail, enable DLSS/FSR.
RAM Bottleneck
Not enough memory (or fast enough memory) for the workload, forcing reliance on slower storage. Signs: Stuttering when loading new areas, high memory usage, slow app switching. Fix: 16GB minimum (32GB for heavy multitasking), dual-channel with XMP/EXPO enabled.
Storage Bottleneck
The drive can’t read or write fast enough. This mainly hurts load times, not FPS. Signs: Long load screens, texture pop-in, slow boot. Fix: HDD → SSD or NVMe.
VRAM Bottleneck
The GPU runs out of video memory for your resolution or texture settings. Signs: Texture pop-in, stuttering at high res, sudden frame-time spikes. Fix: Lower texture quality, or upgrade to a GPU with more VRAM (8GB+ for 1440p/4K), especially in VRAM-hungry games.
PCIe Bottleneck
The GPU isn’t getting the bandwidth it’s rated for, usually because it’s in a slot wired for fewer lanes, running at an older PCIe generation than it supports, or sharing lanes with an NVMe drive in the wrong M.2 slot. Signs: GPU-Z’s “Bus Interface” reading shows x8 or x4 instead of the card’s rated x16, or an older PCIe generation than the card supports; real-world performance lags behind benchmarks despite normal usage numbers.
Fix: Confirm the card is in the primary x16 slot, check PCIe generation settings in BIOS, and check your motherboard manual for which M.2 slots share lanes with the GPU slot. When you buy a new board, pick one whose main slot keeps all 16 lanes with the M.2 drives you plan to fit.
Thermal Bottleneck
CPU or GPU overheats and throttles its own clock speed to cool down. Signs: Performance drops after several minutes; FPS recovers once cooled. Fix: Better airflow, clean dust, reapply thermal paste, upgrade the cooler.
Power Supply Bottleneck
The PSU can’t deliver stable wattage under load. This causes crashes, not slow frame rates. Signs: Random restarts under load, GPU never reaching boost clocks, black screens. Fix: a higher-wattage PSU rated for your combined CPU and GPU draw.
Network Bottleneck
Your connection can’t move data fast or stably enough. This affects online play, not local FPS. Signs: high ping with normal FPS, rubber-banding, packet loss. Fix: Wired Ethernet, close background downloads, pick a closer server.
Common myth: “A bottleneck means one of my parts is broken.” Not true. A bottleneck describes a relationship between two components, not a defect in either one. A perfectly healthy CPU can bottleneck a perfectly healthy GPU simply from the tier gap between them.
CPU Bottleneck vs GPU Bottleneck: How to Tell Them Apart
Mixing these two up is the most common upgrade mistake people make: you spend money on the wrong part and see almost no improvement.
| CPU Bottleneck | GPU Bottleneck | |
| What’s waiting | GPU waits on CPU | CPU waits on GPU |
| More common at | Lower resolutions (1080p) | Higher resolutions (1440p, 4K) |
| Usage pattern | CPU ~100%, GPU lower | GPU ~100%, CPU lower |
| Resolution fix | Little effect | Lowering resolution helps a lot |
| Settings fix | Lower draw distance, physics, AI | Lower shadows, textures, ray tracing |
| Most affected games | Strategy, simulation, open-world with many NPCs | AAA titles with heavy visual effects |
The fastest way to tell them apart: watch how the number changes as you raise resolution. GPU workload grows fast with pixel count while CPU workload barely moves, so a CPU bottleneck at 1080p often shrinks or disappears at 4K (full explanation in theresolution section below).
If your CPU is the bottleneck, upgrading the GPU won’t fix anything; the new card just waits longer. Upgrade the CPU, close background apps, or lower CPU-dependent settings.
If your GPU is the bottleneck, a CPU upgrade is largely wasted money. Upgrade the GPU, lower resolution/settings, or turn on DLSS/FSR to cut the GPU’s rendering load without new hardware.
Reading CPU/GPU Usage Patterns
Checking usage percentages side by side (Task Manager or MSI Afterburner) is the fastest way to confirm a bottleneck without touching a calculator.
| Pattern | CPU | GPU | What’s Happening |
| CPU Bottleneck | ~100% | ~40% | CPU can’t feed the GPU fast enough |
| GPU Bottleneck | ~40% | ~100% | GPU can’t render fast enough |
| Balanced & Maxed | ~100% | ~100% | Both working hard together, normal for demanding games |
| Underutilized | Low | Low | Task isn’t demanding, no bottleneck present |
A 100/100 reading isn’t a problem, it’s what a well-matched system looks like under a genuinely heavy game. Watch temperatures here, though; sustained full load on both parts puts real thermal stress on your cooling.
A low/low reading just means the game isn’t demanding enough to stress your hardware. If you’re seeing this during a demanding AAA title, check that V-Sync or a frame cap isn’t artificially capping your FPS.

How Resolution Changes Your Bottleneck
This is the most misunderstood part of bottleneck analysis. People assume their CPU/GPU balance is fixed. It isn’t; it shifts with every resolution jump, because resolution controls how much work lands on the GPU versus the CPU.
1080p
The GPU renders quickly since there are fewer pixels to process, which exposes the CPU as the more likely limiter. High-end GPUs paired with older or budget CPUs often show their worst numbers here. If you play competitive, high-refresh titles at 1080p, your CPU matters more than most people expect.
1440p
For most modern builds, this is where CPU and GPU workloads come closest to balancing out, generally considered the “sweet spot” for mid-range and high-end builds.
4K
The GPU carries the overwhelming majority of the workload, four times the pixel count of 1080p, which shifts the bottleneck decisively toward the graphics card. CPU bottlenecks become far less common here, though not impossible with very weak processors. This is also where VRAM headroom starts to matter most, and the heaviest games of 2026 are where an 8GB card runs out first.
A CPU showing 25% at 1080p might drop to 8% once you push to 4K, because the GPU absorbs more of the workload as resolution climbs. Check your bottleneck at the resolution you actually play at, not your monitor’s maximum, and not a resolution you’re merely considering. Get this wrong and the calculator points you toward the wrong upgrade entirely, even though the math it ran was correct.
How RAM Affects Bottlenecks
Unlike CPU/GPU imbalance, a RAM bottleneck doesn’t show up as one component maxed out while the other waits. It shows up as stuttering, hitching, and inconsistent frame times, even when CPU and GPU usage both look reasonable.
Capacity: When RAM runs out, Windows starts using your storage drive as overflow memory (paging), which is dramatically slower and causes noticeable stutters.
- 8GB, a common stutter point in 2026’s newer titles, especially with background apps open
- 16GB, the realistic minimum for gaming right now
- 32GB, recommended for heavy multitasking, streaming while gaming, or content creation
Speed: Faster RAM narrows the gap between processor and memory. On AMD Ryzen builds, RAM speed has a measurable impact because the CPU’s internal Infinity Fabric bus scales with memory speed. On Intel, the effect is smaller but still present.
Single vs. Dual Channel
The factor most people overlook: Running two sticks in dual-channel mode roughly doubles memory bandwidth versus one stick of the same total capacity. 16GB in dual-channel will noticeably outperform 16GB in single-channel, especially in CPU-heavy games. If you’re running one stick, adding a matching second one is one of the cheapest upgrades available, often free performance you’re currently leaving on the table.
Common Mistake: Buying more RAM capacity to fix a bottleneck when the real issue is single-channel configuration or disabled XMP/EXPO. Check your channel setup before assuming you need more sticks.
How Storage Affects Performance
Storage speed doesn’t cause CPU or GPU bottlenecks and won’t change your FPS ceiling, but a slow drive limits load times, asset streaming, and how responsive your system feels overall.
| Storage Type | Relative Speed | Best For |
| HDD | Slowest | Bulk storage only, not OS/games |
| SATA SSD | ~5–6x faster than HDD | General use, most gaming |
| NVMe SSD | ~5–10x faster than SATA SSD | Large open-world games, fast streaming |
If you’re still running your OS or games off an HDD, this is the single easiest upgrade available, full stop.
DirectStorage lets supported games read data directly from an NVMe drive to the GPU, cutting CPU overhead in the loading pipeline. It only helps if the specific game you’re playing has implemented it; owning a fast NVMe drive alone doesn’t activate the feature automatically. Check a game’s system requirements or patch notes before expecting a difference.
Rule of thumb for diagnosing: low FPS points to CPU or GPU; long load screens and a sluggish system point to storage. A faster drive won’t touch your frame rate ceiling, and a faster GPU won’t shorten your load times.
CPU Core Count vs. Clock Speed
| Use Case | What Matters More | Why |
| Competitive/esports gaming | Clock speed | Lightly threaded games; single-core speed drives FPS |
| Modern AAA/open-world gaming | Core count | Workload spreads across many cores |
| Streaming while gaming | Core count | Encoding needs dedicated cores separate from the game |
| Video editing/rendering | Core count | Render tasks scale directly with available cores |
| General everyday use | Either (6+ cores is plenty) | Rarely CPU-limited at all |
A CPU with 6 fast cores at 5.0GHz will often beat a CPU with 12 slower cores at 3.5GHz in a pure competitive shooter, because that game barely touches more than a handful of cores. Flip to an open-world game with 100+ NPCs on screen, and the 12-core chip pulls ahead; the extra cores handle background calculations the 6-core chip has to queue up.
Practical rule: For gaming builds, look for strong single-core performance plus at least 6 cores/12 threads. That combination covers both competitive titles and modern AAA games. Chasing 16+ cores mainly benefits rendering, streaming, and productivity, not gaming FPS.
Frame Time, Frame Pacing, and Micro-Stutter: What Your Numbers Actually Mean
Average FPS hides a lot. Two systems can both average 90 FPS, and one feels smooth while the other feels stuttery, the difference is frame time consistency, not the average number.
| Metric | What It Shows | Best For Diagnosing |
|---|---|---|
| Average FPS | Overall performance level | General comparison between builds |
| 1% Low FPS | Consistency during demanding moments | CPU bottlenecks, RAM stutters |
| 0.1% Low FPS | Worst-case, isolated stutters | Background processes, throttling, driver issues |
A big gap between average FPS and 1% low FPS is a strong bottleneck signal, often a CPU struggling in busy scenes or a RAM bottleneck causing brief stutters.
Frame pacing is separate: uneven spacing between frames feels like stutter even when the FPS counter looks fine. Shader compilation stutter is another common cause, short hitches early in a session as the GPU compiles shaders for the first time, fixed by enabling “shader pre-caching” if the game offers it.
When comparing upgrades, weigh 1% lows as heavily as the average, a lower average with tighter 1% lows usually feels smoother than a flashier number with inconsistent delivery.
Gaming Technologies That Change Your Bottleneck
| Technology | Effect | Best For |
|---|---|---|
| DLSS | Reduces GPU workload | GPU bottleneck (RTX cards only) |
| FSR | Reduces GPU workload | GPU bottleneck (any GPU brand) |
| XeSS | Reduces GPU workload | GPU bottleneck (best on Intel Arc) |
| Ray Tracing | Adds GPU workload | Visual quality, not performance |
| Resizable BAR / SAM | Reduces CPU-GPU transfer inefficiency | Small, free performance gain |
DLSS/FSR/XeSS render at a lower internal resolution and reconstruct a sharper image, cutting the GPU’s actual rendering workload. If you’re facing a GPU bottleneck at 1440p or 4K, this is one of the most effective free fixes available. It won’t help a CPU bottleneck, since CPU workload stays roughly the same regardless of internal render resolution. DLSS Frame Generation is a different switch, and it needs hardware-accelerated GPU scheduling turned on.
Ray tracing can swing a bottleneck reading toward the GPU significantly, sometimes 20–30 percentage points on the same hardware. That’s why ray tracing and upscaling are usually paired in practice: upscaling offsets the extra load ray tracing adds.
Resizable BAR / Smart Access Memory let your CPU access the GPU’s full VRAM at once instead of the old 256MB-at-a-time limit. Gains vary by game, typically 2–8%. Not a bottleneck fix on the scale of upscaling, but free performance worth enabling in BIOS if your board and GPU both support it.
How to Check a Bottleneck Without a Calculator
| Tool | Best For | Key Strength |
| Task Manager | Quick, no-install check | Already built into Windows |
| MSI Afterburner | Real-time in-game overlay | Temperature + usage together |
| HWInfo | Deep diagnostics | Per-core, per-sensor detail |
| FrameView | Frame consistency analysis | 1% and 0.1% low tracking |
| CapFrameX | Frame time capture & review | Post-session analysis and graphs |
| GPU-Z | PCIe/bus diagnostics | Confirms slot generation and lane width |
Task Manager: Press Ctrl+Shift+Esc, click Performance, watch CPU vs. GPU usage while playing. Fast, but it updates slowly and doesn’t break down per-core usage, so it can miss a bottleneck caused by one or two overloaded cores while overall usage looks moderate.
MSI Afterburner (free, works with any GPU brand), paired with RivaTuner Statistics Server, shows real-time CPU, GPU, RAM, temperature, and FPS directly on screen. It’s the most commonly recommended tool because it updates instantly and shows temperature alongside usage, catching thermal throttling at the same time.
HWInfo (HWiNFO64) gives the most detailed sensor data available: per-core usage, individual voltage rails, granular temperatures. Use it when Task Manager or Afterburner show moderate usage but you’re still getting stutters; it can reveal a single core spiking to 100% while the rest sit idle.
FrameView (NVIDIA-made, works with AMD/Intel GPUs too) focuses on frame-time analysis: logging average FPS, 1% lows, and 0.1% lows. Use it when usage percentages alone aren’t giving you a clear answer.
CapFrameX builds on the same frame-time data Afterburner and FrameView capture, but it’s built for reviewing runs after the fact, useful for comparing a “before and after” of a settings change or driver update rather than watching numbers live.
Workflow: Start with Task Manager. Unclear result or suspected throttling → MSI Afterburner. Suspect a single overloaded core or unexplained stutters → HWInfo. Suspect PCIe lane issues → GPU-Z. Want to compare two configurations side by side → log both runs in CapFrameX or FrameView.

Bottlenecks Beyond Gaming: Streaming, Content Creation, and AI Workloads
Gaming gets most of the attention in bottleneck discussions, but the CPU/GPU relationship looks different once you add a second workload on top.
Streaming (OBS or similar)
Streaming while gaming effectively runs two jobs on your PC at once, the game itself and video encoding for the stream. Software (x264) encoding leans heavily on CPU cores the game isn’t using, which is why streamers benefit disproportionately from higher core counts even if their gaming-only bottleneck looks fine.
Hardware encoding (NVENC on NVIDIA, similar on AMD/Intel) offloads that work to a dedicated block on the GPU instead, freeing CPU cores for the game, the better default for most streamers unless you’re chasing the best possible encode quality at a given bitrate.
Video editing and rendering
Export and render times scale with core count far more directly than gaming FPS does, since rendering is a heavily parallel workload. A CPU that looks “balanced” for gaming can still be the clear bottleneck for a render-heavy workflow, and vice versa, a strong multi-core CPU paired with a modest GPU can render competently even while it’s GPU-bottlenecked in games.
Local AI workloads
Running local machine learning models or AI-assisted creative tools shifts the bottleneck conversation toward VRAM capacity and memory bandwidth more than raw core count or clock speed. A GPU that handles gaming comfortably can still run out of VRAM headroom for larger local models, a different failure mode than a gaming bottleneck, and one this calculator isn’t built to estimate.
VR gaming
VR headsets typically demand a higher, more consistent minimum frame rate than flat-screen gaming to avoid discomfort, which raises the bar for both CPU and GPU at once rather than shifting load toward one or the other the way resolution does on a monitor. Treat VR headroom requirements as a separate, stricter target than the flat-screen resolution guidance elsewhere in this guide.
Bottom line for multi-tasking use cases: a gaming-only bottleneck reading doesn’t tell you much about how your PC handles streaming, editing, or local AI workloads layered on top. Core count and VRAM headroom decide those outcomes far more than they decide gaming FPS.
Best CPU & GPU Pairings by Budget Tier
| Tier | Target Resolution | Typical Bottleneck Risk |
|---|---|---|
| Budget | 1080p | Low, components are naturally matched |
| Mid-Range | 1440p | Low, but RAM config matters more |
| High-End | 1440p/4K | Resolution choice significantly affects balance |
| Extreme | 4K | Shifts toward power/thermal considerations |
Budget
6-core mid-range CPU + entry-to-mid GPU with 8GB VRAM + 16GB dual-channel RAM. Targets smooth 1080p at high-medium settings. At this tier, CPU and GPU are typically evenly matched by design; budget GPUs simply don’t have enough power to expose a CPU bottleneck.
Mid-Range
6–8 core current-gen CPU with strong single-core clocks + mid-tier GPU with 12GB VRAM + 32GB dual-channel RAM with XMP/EXPO enabled. Targets 1440p high settings with headroom for 1080p high-refresh competitive play. RAM speed and channel config start mattering more here.
High-End
High-core-count current-gen CPU + high-end GPU with 16GB VRAM + 32GB high-speed dual-channel RAM. Targets 1440p ultra and playable-to-strong 4K. Resolution choice matters more than ever at this tier; the same components can lean CPU at 1080p and GPU at 4K.
Extreme
Flagship CPU + flagship GPU with 16GB+ VRAM + 32–64GB highest stable rated speed. Targets 4K ultra, high-refresh 1440p, or simultaneous streaming/rendering. At this tier, the conversation often shifts away from CPU/GPU balance entirely and toward PSU capacity and cooling; flagship parts draw serious power and generate real heat under sustained load. The motherboard matters too, because a chip like a Core i7 or i9 needs solid power delivery on the board to hold its boost clocks.
Common Mistake: Overspending on a flagship CPU for a budget-tier GPU build “to future-proof it.” In gaming, the GPU limits your experience long before CPU headroom becomes relevant. That money is better spent on the GPU or saved for the next upgrade cycle.
Worked Examples: Real Results From This Calculator
These are the calculator’s own results at its default settings: general gaming, 144 Hz, 16 GB of DDR4-3200 in dual channel and an NVMe Gen3 drive. The Core i9-14900K and Ryzen 7 7800X3D builds use 32 GB of DDR5-6000 instead. Enter the same parts above and you get the same numbers. The table leaves out the calculator's FPS range, which is only a ballpark; each card's measured frame rates at 1080p, 1440p and 4K come from TechPowerUp's testing.
| Pairing | 1080p | 1440p | 4K |
|---|---|---|---|
| Ryzen 5 5600 + RTX 4060 | 5.9% GPU | 8.1% GPU | 10.5% GPU |
| Ryzen 5 3600 + RTX 5080 | 20.9% CPU | 15.1% CPU | 9.0% CPU |
| Core i9-14900K + RTX 4060 | 16.4% GPU | 22.6% GPU | 29.3% GPU |
| Ryzen 7 7800X3D + RTX 5080 | 1.3% GPU | 1.7% GPU | 2.3% GPU |
Ryzen 5 5600 + RTX 4060: A Matched Budget Pair
Very Good Balance at all three resolutions, with a B+ grade. The graphics card is slightly the weaker part, so the gap grows a little as the resolution rises and the GPU carries more of the work. The suggested upgrade is an RTX 4060 Ti as the budget step or an RTX 4070 as the bigger one, and the 5600 handles cards well beyond the 4060. Going the other way, the RTX 4060 runs fine on older CPUs too.
Ryzen 5 3600 + RTX 5080: An Old CPU With a New Card
This is the classic upgrade trap. At 1080p the processor holds the card back by 20.9%, while a Ryzen 7 7800X3D leaves the same card at just 1.3%. At 4K the same pair drops to 9.0%, because the card does far more of the work. If you play at 1080p on a high-refresh monitor, the CPU is the part to replace; the calculator suggests a Core i5-13400F as the bigger step, and there are stronger RTX 5080 pairings if the budget allows.
Core i9-14900K + RTX 4060: Too Much CPU for the Card
Here the graphics card is the limit, and the gap widens to 29.3% at 4K. Look at the frame rates: they match the Ryzen 5 5600 build to the frame, because the RTX 4060 sets the pace either way. The extra money spent on the processor buys nothing in games; it belongs in the graphics card.
Ryzen 7 7800X3D + RTX 5080: A High-End Match
Excellent Balance and an A+ grade at every resolution, with a gap under 3%. Neither part waits on the other, and the frame rate is set by how hard you push the settings. A 7800X3D can drive even faster cards than this one.
Future-Proofing and Upgrade Strategy
A calculator tells you where you stand today. Planning your next upgrade means thinking about where you’ll stand in 18–24 months, a different question entirely.
GPUs age Faster than CPUs, in Practical Terms
New rendering techniques, VRAM requirements, and game engine demands tend to push GPUs toward obsolescence before a comparable-tier CPU becomes the limiting factor. A reasonable default: budget for a GPU upgrade roughly every other generation, and expect a well-chosen CPU to comfortably outlast two GPU upgrade cycles.
Buy the CPU for Today’s Games, not Tomorrow’s Rumors
Overbuying core count now “just in case” future games need it rarely pays off; game engines evolve, and by the time a game meaningfully uses 16 cores for gameplay logic, the CPU landscape has moved on anyway. Match your CPU tier to your GPU tier for the resolution you play at now.
VRAM is the Exception to “Buy for Today.”
Unlike core count, VRAM requirements have climbed steadily across recent years, and a GPU that’s short on VRAM ages out of usefulness faster than one that’s merely a tier behind on raw compute. If you’re choosing between two GPUs at a similar price with different VRAM capacities, the extra VRAM is usually the safer long-term bet, particularly at 1440p and above.
A Practical Upgrade Sequence for most Builders
Diagnose first (calculator + Task Manager), fix free issues (XMP/EXPO, drivers, background apps), then address the single largest bottleneck, not both components at once, unless your current bottleneck genuinely exceeds 30% on both. Spreading a limited budget across two half-upgrades usually delivers less than fully addressing one.
Calculator Limitations: What This Tool Can’t Tell You
Being upfront about this makes the number you get more useful, not less.
- It doesn’t know your specific game: Engine optimization varies enormously; a CPU-heavy strategy title and a GPU-heavy shooter will behave differently on identical hardware, and a general-purpose calculator can’t model every engine individually.
- It can’t see your cooling or case airflow: Two identical builds can perform differently if one is thermal-throttling and the other isn’t; the calculator assumes hardware running at its rated specification.
- It only knows the settings you give it: The Streaming and Background apps switches add load, but the calculator cannot see what actually runs on your PC, such as overlays, RGB software or an antivirus scan.
- Driver version and game patches shift real results: After the estimate is generated; benchmark databases update periodically, not in real time with every driver release.
- It trusts your RAM settings: If you pick dual channel but XMP or EXPO is off, or a PCIe slot runs below its rated generation, the result will look better than the PC performs. Only a check on the machine catches that.
- Laptop results are rougher: Laptop graphics cards and processors are in the list, but their scores have not been re-checked against current reviews the way desktop parts have, and power limits vary between laptop models.
Use the calculator to narrow down which component is the likely limiter and roughly by how much, then confirm with Task Manager, Afterburner, or GPU-Z on your actual system before spending money.
Common Upgrade Mistakes
| Mistake | Why It Costs You | The Fix |
| Upgrading without diagnosing | Spends money on the wrong part | Run a bottleneck check first |
| Flagship GPU + budget CPU | GPU sits underused | Match CPU tier to GPU tier |
| Flagship CPU + budget GPU | CPU sits underused | Match GPU tier to CPU tier |
| Chasing core count for gaming | Pays for cores gaming won’t use | Prioritize single-core speed + 6–8 cores |
| Ignoring resolution | Buys the wrong-sized upgrade | Confirm target resolution first |
| More RAM instead of dual-channel/XMP | Misses a free fix | Check channel config and BIOS settings first |
| Skipping the PSU check | Causes instability, not smooth gains | Verify PSU wattage before a GPU upgrade |
| Testing at the wrong resolution | Leads to an unnecessary upgrade | Test at your actual playing resolution |
| Ignoring VRAM for the resolution you play | GPU ages out early despite good compute | Prioritize VRAM headroom at 1440p/4K |
The most expensive mistake on this list, by a wide margin, is the first one: buying a new GPU because “the PC feels slow” without checking whether CPU, GPU, RAM, or storage is actually the culprit. That’s hundreds of dollars spent chasing a guess.
Frequently Asked Questions
Is a 20% Bottleneck Bad?
Not yet. The calculator rates 20% as Good Balance: a benchmark can show the gap, but you rarely feel it. Above 22% the rating drops to Fair, and that is the point to plan an upgrade for the weaker part, sooner if you also see stutter or frame drops.
Does RAM Cause a Bottleneck?
Rarely in the dramatic way CPU or GPU mismatches do. Insufficient capacity, slow speed, or single-channel configuration can quietly limit performance through stuttering and inconsistent frame times, even when CPU/GPU usage looks fine.
What’s a Good Bottleneck Percentage for Gaming?
The bands above hold for most builders, but your tolerance should shift with what you’re playing. Competitive shooters where 1% lows matter more than average FPS deserve a stricter target, closer to 10%, while a single-player, story-driven game at 20% will rarely be something you consciously notice.
Can a Bottleneck Damage my PC?
No. A bottleneck limits performance, not hardware longevity. The exception is a thermal bottleneck caused by inadequate cooling; sustained overheating can shorten component lifespan, a separate and more serious issue worth addressing directly.
Does Upgrading RAM fix a CPU Bottleneck?
No. RAM and CPU bottlenecks are different problems with different symptoms. If your CPU is the limiting factor, more RAM won’t close that gap; you need a CPU with more cores/threads or stronger single-core performance.
Is Bottlenecking Normal?
Yes. Some degree of bottleneck exists in nearly every PC build, since CPU and GPU releases never land on the exact same upgrade cycle. A small bottleneck under 10% is expected and not something to chase into zero.
Does Resolution Affect my Bottleneck?
Significantly. Lower resolutions shift the load toward your CPU; higher resolutions shift it toward your GPU. The same hardware pairing can show a very different bottleneck percentage at 1080p versus 4K, always check at the resolution you actually play at.
Does a Bottleneck Affect Streaming or Video Editing the same way it Affects Gaming?
Not exactly. Streaming and rendering lean much more heavily on CPU core count than gaming FPS typically does, so a system that’s well-balanced for gaming can still be CPU-limited for encoding or export tasks, and vice versa.
Will a Bottleneck Calculator tell me my Exact FPS in a Specific Game?
No. It gives a benchmark-informed range based on your components and resolution, not a per-game guarantee. Engine optimization, patches, and settings all shift real FPS beyond what a spec comparison can capture.
Should I Worry about a Bottleneck on a Brand-New, Top-Tier Build?
Generally no. A small bottleneck is expected even in flagship-tier pairings, since CPUs and GPUs rarely launch on perfectly matched cycles. Focus on whether the number stays under roughly 22%, not whether it is exactly 0%.
Is a CPU Bottleneck or a GPU Bottleneck Better?
A GPU bottleneck. In games you want the graphics card to be the limit, because it means the card is fully used and the frame rate is steady. A CPU bottleneck caps the frame rate however low you set the graphics, and it tends to show up as uneven frame times.
Does the Calculator Work for Laptops?
Roughly. Laptop graphics cards are listed with “Laptop” in the name, and you should pick that version, because a laptop RTX 4060 is slower than the desktop card. Laptop processors such as the Core i5-12500H are listed too. Power limits differ between laptop models, so treat a laptop result as a ballpark.
Does the Motherboard or Power Supply Change the Result?
No, neither one is an input. A motherboard does not change a processor’s gaming score as long as it supports the chip and runs the RAM at its rated speed. A power supply only matters when it is too weak, which shows up as crashes or shutdowns under load rather than as a percentage; see power supply bottleneck.
Should I Upgrade the CPU or the GPU First?
Upgrade the part the calculator names as limiting, at the resolution you actually play. For most gaming PCs at 1440p and 4K that is the graphics card. At 1080p with a high-refresh monitor, an older processor is often the limit instead.
Does Overclocking Fix a Bottleneck?
Only at the edges. An overclock speeds up one part by a small margin, which can nudge a borderline result into the next rating. It will not close a gap of 20% or more; a faster part will.
Why Use This Calculator
Most bottleneck tools stop at CPU and GPU. This one factors in RAM channel/capacity and storage type too, because a build can look perfectly balanced on CPU vs. GPU alone and still stutter from a single-channel RAM kit or an undersized drive. The upgrade suggestion it gives names an actual component tier rather than a generic “upgrade your GPU,” and the FPS estimate is shown as a range rather than a falsely precise single number, a more honest reflection of what benchmark-based prediction can actually promise.
Run your specs above, then use the sections on this page, CPU vs. GPU diagnosis, resolution effects, streaming and content-creation workloads, or the fix checklist, to figure out exactly what your result means for your next upgrade.