PC slowness usually comes from a single bottleneck (CPU, GPU, RAM, or SSD) that hits 90-100% usage or causes high latency during the task you care about. Start with read-only checks: monitor per-core CPU load, GPU utilization/VRAM, memory pressure, and disk active time/queue. Then apply the smallest, safest fix before considering upgrades.
Quick Diagnostic Checklist for Slow PCs
- Reproduce the slowdown and note the exact action (boot, app launch, game load, compile, video export).
- Check CPU per-core utilization and clocks while slow; look for one core pegged near 100%.
- Check GPU utilization and VRAM use; watch for VRAM saturation and power/thermal limits.
- Check RAM "in use", "available", and swap/pagefile activity; confirm if memory pressure is present.
- Check disk "active time", latency, and queue length; confirm if storage is the blocker.
- Check thermals: sustained throttling (CPU/GPU clocks dropping under load) is a common hidden cause.
Identifying a CPU Bottleneck: Symptoms and Tests
What you typically notice
- UI stutter or delays during multitasking even when GPU load looks low.
- Game FPS drops in crowded scenes; GPU usage stays low or fluctuates.
- High fan noise with lower-than-expected performance (possible thermal throttling).
- Slow compilation, encoding, compression, or heavy browser tabs.
Safe tests (read-only) to confirm
- Windows: Task Manager → Performance → CPU. Switch graph to Logical processors. A CPU bottleneck often shows one or more cores near 100% while the GPU is not fully utilized.
- Linux:
htop(per-core view). If a single thread maxes one core, overall CPU may look "not 100%" but you are still CPU-limited. - Clocks/thermals: check if CPU clock drops under load (throttling). On Windows, tools like HWiNFO (sensors view) are read-only; on Linux,
sensors(lm-sensors) andlscpu. - Background load: sort processes by CPU usage; verify antivirus scans, update services, or runaway tabs are not stealing cycles.
Low-risk fixes before "อัปเกรดคอม เพิ่มความเร็ว"
- Disable unnecessary startup apps (Windows: Task Manager → Startup apps).
- Update chipset drivers and BIOS only if you have a clear reason (stability/performance notes); schedule downtime.
- Improve cooling (clean dust, restore airflow) to prevent throttling; this often beats "upgrade" expectations for free.
Spotting GPU Limits: When Graphics Are the Culprit
- GPU utilization stays near 95-100% during the slowdown (game/3D/render).
- VRAM usage approaches the card limit; stutter appears when textures load or scenes change.
- Lowering resolution or graphics settings improves FPS immediately.
- FPS is stable but low; CPU per-core is not pegged.
- GPU clocks drop under load (thermal/power limit), even though utilization is high.
- Frame time spikes align with asset streaming (often VRAM or storage-related).
- External displays/high refresh or multi-monitor setups increase GPU load at idle.
- Driver issues: sudden stutter after driver update; verify version and known issues.
- Windows overlay/recording (Game Bar, Discord, OBS) increases GPU load; test with overlays off.
- In games, switching from "Ultra" textures to "High" reduces stutter more than changing shadows (VRAM clue).
Quick workflow to "ตรวจคอขวด CPU GPU"
- Run the same scene twice: once at native settings, once at lower resolution/settings.
- If lowering settings boosts FPS significantly and GPU stays high, you are GPU-limited.
- If lowering settings barely helps and one CPU core is pinned, you are CPU-limited.
Detecting RAM and Memory Bottlenecks
Common causes: too little RAM for workload, heavy background apps, memory leaks, slow swapping to disk, single-channel configuration, or iGPU using shared memory.
Safe checks (read-only)
- Windows: Task Manager → Performance → Memory. Watch In use, Available, and Committed. In Resource Monitor → Memory, watch Hard Faults/sec (frequent spikes can mean paging).
- Linux:
free -h,vmstat 1(watchsi/sofor swap in/out),top/htop(RES, VIRT),swapon --show.
| Symptom | Possible causes | How to verify (read-only) | How to fix (safe-first) |
|---|---|---|---|
| Apps take long to switch; frequent stutter when multitasking | RAM pressure; paging/swapping; too many background apps | Windows: low Available memory + frequent Hard Faults/sec; Linux: swap in/out in vmstat |
Close heavy apps/tabs; reduce startup apps; keep pagefile/swap enabled; then consider adding RAM |
| Browser becomes slow with many tabs | Tab memory usage, extensions, memory leak | Task Manager shows browser processes growing; Linux: per-process RES growth over time | Disable extensions; restart browser periodically; use tab-suspend features; test a clean profile |
| Integrated GPU performance drops and stutters | Shared RAM bandwidth/size limits; single-channel RAM | Check memory channel mode in BIOS/UEFI or tools; observe iGPU using shared memory | Enable dual-channel (matched sticks); ensure XMP/EXPO only if stable; allocate iGPU memory prudently |
| System feels fine until one app starts, then everything slows | Memory-hungry app; leak; VM/container overcommit | Windows: per-process memory in Details tab; Linux: ps aux --sort=-%mem | head |
Cap the app/VM memory; update the app; fix configs; add RAM if workload is legitimate |
| "เพิ่ม RAM คอมเร็วขึ้นไหม" is unclear for your case | Not actually memory-limited; bottleneck is CPU/GPU/SSD | During slowness, RAM Available stays healthy and swap is near idle | Do not buy RAM first; re-check CPU per-core, GPU utilization/VRAM, and storage latency |
Verifying SSD and Storage Performance
Storage bottlenecks often feel like "random pauses": slow boot/login, long app launches, and freezes when loading files. Use safe, read-only checks first; only then change settings.
- Confirm it's disk-related: during the slowdown, check Windows Task Manager → Performance → Disk for high Active time and elevated response time; on Linux, use
iostat -x 1(look for high utilization and latency indicators). - Identify the process doing I/O: Windows Resource Monitor → Disk (sort by Total B/sec); Linux:
iotop(run with appropriate permissions) orpidstat -d 1. - Check free space: keep comfortable headroom; very low free space can worsen write performance and updates. (Windows: Settings → System → Storage; Linux:
df -h.) - Verify drive health (SMART): Windows: read SMART via vendor tool; Linux:
smartctl -a /dev/nvme0n1(or your device). If SMART warns, prioritize backup and replacement. - Confirm link/speed basics: ensure NVMe runs on expected PCIe slot and SATA SSD is on a proper SATA port (not through a slow bridge). This is inspection-level, no data change.
- Measure sequential and random performance: use a trusted benchmark with a short run. Avoid running heavy write tests on production data; schedule downtime. If performance is far below typical for that class of drive, look for thermal throttling or firmware issues.
- Trim/Discard status: Windows: Optimize Drives (should show SSD and scheduled optimization); Linux:
systemctl status fstrim.timer. Enable scheduled trim if it is disabled and you understand the environment. - Only if you have backups and downtime: update SSD firmware and storage controller drivers; test stability afterward.
- Upgrade decision: if the system is HDD-based or a small/old SSD is saturated, "เปลี่ยน SSD คอมเร็วขึ้น" is often the highest-impact change for boot/app launch times, but verify with the checks above before buying.
Essential Tools, Benchmarks and What Metrics Mean
- Windows built-ins: Task Manager, Resource Monitor, Event Viewer (disk warnings), Performance Monitor (advanced logging).
- Linux built-ins:
htop,free,vmstat,iostat,journalctl,smartctl(smartmontools). - Key metrics to interpret quickly:
- CPU: one core pegged + low GPU usage → CPU bottleneck; dropping clocks under load → throttling.
- GPU: 95-100% utilization + setting changes affect FPS → GPU bottleneck; VRAM near limit → stutter risk.
- RAM: low available memory + active swap/pagefile → memory bottleneck; fix workload first, then add RAM.
- Storage: high active time/queue + high response time/latency during pauses → disk bottleneck.
Escalate to a technician or vendor support when
- SMART reports warnings, increasing error counters, or the OS logs repeated disk I/O errors.
- System hard-freezes, reboots under load, or shows WHEA/MCE hardware errors.
- Thermals are abnormal and basic cleaning/airflow checks don't stop throttling.
- Performance is inconsistent across reboots and you suspect failing PSU/RAM/SSD.
- You must not risk production downtime and need controlled diagnostics (spare drive imaging, memtest scheduling, A/B hardware swap).
How to Prioritise Fixes: Impact, Cost and Sequence
- Start with measurement: capture CPU per-core, GPU/VRAM, RAM/swap, and disk latency while slow. This answers "คอมช้า แก้ยังไง" with evidence, not guesses.
- Remove obvious background load: startup apps, browser extensions, sync clients, and overlays; re-test the same scenario.
- Fix thermals before upgrades: dust cleanup, fan curves, reseat coolers only if you're trained and have downtime; otherwise keep it inspection-only.
- Right-size RAM only when memory pressure is proven: if swapping is active during your workload, adding RAM is meaningful; if not, "เพิ่ม RAM คอมเร็วขึ้นไหม" is usually "not for this issue."
- Storage upgrades for responsiveness: HDD → SSD is the biggest felt improvement for boot/app launch; saturated/aging SSD may also justify replacement.
- CPU vs GPU upgrades: use the reproducible test (lower resolution/settings). If FPS rises a lot, consider GPU; if not, consider CPU/platform.
- Plan "อัปเกรดคอม เพิ่มความเร็ว" as a sequence: smallest change that removes the measured bottleneck first (often SSD/RAM), then re-measure before buying the next part.
- Keep systems stable: avoid aggressive overclocks on production machines; stability issues often masquerade as "slowness."
Concise Answers to Typical Troubleshooting Concerns
How do I quickly tell whether it's CPU or GPU?
Re-test the same scene at lower resolution/settings. Big FPS gain points to GPU; little change with one CPU core pinned points to CPU.
Is high CPU usage always a problem?
No. It's a bottleneck when CPU is high during the slowdown and GPU isn't fully utilized, or when clocks drop because of throttling.
Will adding RAM always make my PC faster?
Only if you are memory-limited (low available memory and active swap/pagefile). If swapping is near idle, more RAM often won't change performance.
Does switching to an SSD improve gaming FPS?
Usually it improves load times and reduces streaming stutter, not average FPS. FPS is mainly CPU/GPU-bound unless the game constantly streams assets from disk.
What's the safest first step when I suspect storage issues?
Check disk active time/latency and identify the process doing I/O, then check SMART health. Don't run heavy write benchmarks on production data.
What should I do if I see SMART warnings?
Back up immediately, then plan a drive replacement. Performance troubleshooting comes after data safety.
Why does my PC feel slow after waking from sleep?
Drivers (especially storage/GPU), background updates, or power-state bugs can reduce clocks or spike disk I/O. Reboot and re-test; if it repeats, update drivers/firmware with downtime and backups.
