A 100% CPU reading is a symptom, not a diagnosis. First establish whether one process is consuming a single core, several threads are saturating the machine, or the server is mainly waiting on storage. Then map the workload to its service, correct the cause, and verify the result under the same load. Do not reboot or kill a process until you know what owns it and what user-visible work it performs.
Antihost offers Linux VPS plans with full root access, unlimited traffic on a 1 Gbps port and a 14-day free trial.
The workflow below uses Ubuntu’s built-in tools to identify the process or thread, distinguish CPU work from I/O wait, apply the least disruptive fix, and measure whether the fix improved latency, throughput, errors, and load.
As an Amazon Associate I earn from qualifying purchases.
The Tool Desk
Outbyte PC Repair FREEClear out junk files and repair common Windows errorsFree Scan →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →What “100% CPU” means on Ubuntu
In top, a task’s %CPU is its share of elapsed CPU time. On a multicore (SMP) server, the number is not necessarily capped at 100: a multithreaded process can report more than 100% when several cores are busy. Compare the task’s value with the host’s logical CPU count from nproc or lscpu. A process at 100% on a 16-CPU host is using roughly one core; a process at 1,600% could be occupying all 16 if the measurement and display support that range.
The host-wide CPU summary in top also matters. User time indicates application work, system time indicates kernel or system-call work, and wait time indicates time spent waiting for I/O. A machine can feel slow even when the user-plus-system percentages are modest if storage or another device is keeping many tasks blocked.
#1 Best Overall
- BLUE LIGHT BLOCKING GLASSES: Our blue light glasses for women boast UV400 protection, blocking harmful blue rays from computer screens. With these bluelight glasses women, you can enjoy gaming, reading, or watching TV with no worries of eye strain.
- EYE RELIEF: Our blue light glasses are engineered to reduce blurry vision or strain caused by prolonged screen time. These womens blue light glasses aid in better sleep by preventing blue light from disrupting your natural sleep cycle, helping you fall asleep faster and enjoy a more restful night's sleep, making them the perfect christmas present.
- COMFORTABLE: Crafted from durable material, our trendy blue light glasses for woman or blue light glasses men ensure long-lasting comfort. The high-transmittance lenses accessories, gamer gifts for men, help keep your eyes relaxed. Perfect stocking stuffers, gifts for gamers men, christmas gifts for men, halloween costumes.
- LIGHTWEIGHT: These clear blue light glasses feature a lightweight frame, meaning they’ll stay comfy for long-term wear. Perfect as blue blockers glasses for women or mens blue light glasses, these non-prescription fake glasses for men can be worn throughout the day. The blue light blocking glasses women feature polycarbonate HD lenses, which are anti-reflective and restore true color.
- WORRY-FREE PURCHASE: If you experience any issues, our support team is ready to assist you quickly. These are the perfect costume glasses for cosplay dressing up.
CPU saturation versus a busy single core
Record the number of online CPUs before judging a percentage. A single-threaded game server, database worker, or script may legitimately hold one core at 100% while other cores remain idle. That is different from a workload that has spread across every core, where run-queue pressure and response latency generally rise. Thread-level inspection is necessary when the process total looks high but the individual workers behave differently.
Why load average is not the same as CPU usage
Ubuntu’s /proc/loadavg definition counts jobs in the run queue (state R) and jobs waiting for disk I/O (state D), averaged over one, five, and 15 minutes. Consequently, a high load average with lower CPU utilization can indicate storage pressure rather than a CPU-bound program. A high load with a high %wa value is a reason to investigate disks, filesystems, network-backed storage, or other blocking operations before restricting the apparent top process.
Run uptime to see the three load windows, then compare them with the CPU state breakdown in top. The one-minute value reacts quickly to a new event; the five- and 15-minute values show whether the condition is persistent. Compare the load to the number of CPUs rather than treating a particular numeric threshold as universal.
Step 1: Establish a system-wide picture
-
Check uptime and the three load averages:
uptime -
Open the live process and CPU view:
top -
Count available processors:
nprocUse
lscpuwhen you also need topology and online-CPU details.
In top, sort tasks by CPU so the largest consumers are visible first. Note whether one core or many are busy, and record the user, system, idle, wait, and other state percentages. Capture the observation before changing anything; it becomes the baseline for validation.
What to record in the baseline
- One-, five-, and 15-minute load averages.
- Logical CPU count and whether all CPUs are online.
- The process or processes at the top of the CPU list.
- User versus system CPU share and the host-wide I/O-wait share.
- Whether latency, throughput, error rate, or player-facing responsiveness is degraded.
Step 2: Identify the process and, when necessary, the thread
Use a point-in-time process snapshot
For a sortable snapshot that includes ownership, parent process, state, CPU, memory, elapsed time, and command, run:
Free tools Windows power users keep installed
One-click scans. No signup required.
Rank #2
- WINDEX GAMING WIPES: effortlessly clean all your gaming gear and gaming electronics in one easy step with our new Windex Gaming Wipes
- AMMONIA FREE: our ammonia-free formula is safe to use on all of your favorite gaming equipment like consoles, headsets, controllers, screens, keyboards, and more
- PRE-MOISTENED WIPES: each wipe instantly removes dust, dirt, and stubborn smudges and fingerprints leaving a streak-free shine
- ANTI-STATIC: great for cleaning and wiping down your gaming gear and electronics leaving behind no visible residue
- COMPACT RESEALABLE PACKAGE: store your pack of Windex Gaming Wipes anywhere in your home for whenever a quick and easy wipedown is needed
ps -eo pid,ppid,user,stat,pcpu,pmem,etime,cmd --sort=-pcpuThe PID identifies the process; PPID helps reveal whether it was launched by a supervisor or job; USER identifies the account; STAT exposes states such as running or uninterruptible sleep; and ETIME helps distinguish a long-lived daemon from a newly started batch task. The command line can reveal a map-generation job, backup, compiler, game-server instance, or unexpected executable.
Inspect threads in top
If a process is multithreaded, enable thread view in top and sort by CPU again. A process total can hide one hot worker, a set of evenly busy workers, or a thread stuck in repeated system calls. The distinction guides the fix: an application-level loop may need configuration or code correction, while a kernel-heavy pattern may require storage, networking, or driver investigation.
Sample per-thread behavior with pidstat
For repeated one-second samples of thread CPU and context-switch behavior, use:
pidstat -u -w -t 1The -t option exposes threads, -u reports CPU statistics, -w reports context switches, and 1 sets the sampling interval. Look for a thread that remains busy across samples, unusually high voluntary or involuntary context switching, or a process whose CPU falls while wait behavior rises. A single sample can catch a transient burst; several identical samples establish whether the condition persists.
Recommended Free Tools
Use cpustat for repeated host context
cpustat can report active tasks and add load, frequency, and online-CPU information. Use it when you need a compact, repeated view that combines process activity with host conditions. Correlate its output with the same time window from top and pidstat instead of relying on one tool’s ordering.
Step 3: Decide whether the work is CPU-bound or waiting on I/O
Do not assume the process at the top of a list is the root cause until the CPU-state evidence agrees. A task can appear prominent because many other tasks are blocked, while the actual bottleneck is a disk, filesystem, or network path.
Rank #3
- Ergonomic & Adjustable Design: Fight "text neck" and improve your posture. Designed for comfort, it offers an adjustable height (7.1in-8.5in) (1.4in adjustable range) and angle (5°-85°) Please note: cell phone stand is only 9 inches tall—the length of a standard pencil
- Without blocking subtitles Perfect for office desk Tiktok Face Time Zoom video—this is the ultimate kitchen desk essential for phones! Perfect viewing experience while reducing neck back strain, A must-have for gamers streamers desk, and anyone who values both style and support!
- Sleek & Stable Design fits Otterbox Case: Say goodbye to bent cables, tangles, and tip-overs! LISEN’s stylish adjustable desk phone holder features a built-in charging hole and smart cable management, keeping your phone secure and your workspace clutter-free cool gadgets for men
- ANTI-SLIP DESIGN free hands: The under 10.00 items easy gadgets back and the bottom of this kitchen essentials for women are fully covered by anti-skid silicone, which LISEN stand up desk decor can provide maximum protection device from any scratches and slides
- Your Perfect Viewing Companion:If your product arrives damaged or breaks within 365 days, reach out to LISEN for a quick and hassle-free solution. Struggling with assembly, like tightening the screw? We’ve got your back! Your experience matters to us. (Product including: phone Holder stand*1, screw*1, hexagon screwdriver*1, specification*1)
Signs of CPU-bound work
- High user CPU (
%usr) persists across samples. - One or more runnable threads remain active, with little I/O wait.
- Throughput or latency worsens as the workload increases and improves when that workload is reduced.
- The same command, query, retry loop, or scheduled job appears at the top repeatedly.
Signs of kernel or system-call pressure
A high system CPU share suggests kernel or syscall activity. Investigate what the process is asking the kernel to do: intensive filesystem operations, networking, synchronization, or device access can all raise system time. The percentage alone does not identify the subsystem, so confirm with disk latency, filesystem, network, or storage telemetry.
Signs of I/O wait
- Host-wide
%waitis elevated while user CPU is lower. - Many processes sit in
D(uninterruptible sleep). - Load average is high even though the CPUs are not fully busy.
- Disk latency, queue depth, filesystem errors, or remote-storage symptoms coincide with the spike.
Use iostat and the relevant disk, filesystem, network, or storage telemetry to confirm the bottleneck. If the evidence points to I/O, capping the process’s CPU may not solve the incident and can make recovery slower.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Step 4: Map the PID to its service and cgroup
Knowing a PID is not the same as knowing who should control it. A process may belong to a systemd service, a transient scope, a container-related unit, a user session, or a manually launched job.
Ask systemd for the owner
For a service-owned PID, run:
systemctl status <PID>You can also query a known unit directly:
systemctl status <service-name>The status output exposes the unit, main PID, recent log lines, and cgroup context. That context tells you which service controls the process and where a restart or resource change should be applied. Use the service’s logs before stopping it; a restart can hide the original failure and interrupt users.
Inspect cgroup membership directly
When systemd’s relationship is unclear, read:
cat /proc/<PID>/cgroupModern Ubuntu systems use a unified hierarchy managed by systemd. The cgroup membership view can reveal the resource domain that contains the process, including CPU controls and the other tasks affected by a change. Apply limits to that domain only after confirming it does not include unrelated workloads.
Step 5: Choose the least disruptive remedy
Interventions differ in scope, reversibility, service impact, evidence required, and the quality of validation they permit. Use the narrowest action that addresses the demonstrated cause.
Rank #4
- Easy Installation: Our gaming lights have been designed to enhance your entertainment experience. Simply position the led light bar next to your TV or game console to create a game room environment with ease. They come with USB power and a 60 inch cord for direct mounting and mobility
- Audio Syncing & Extensive Mode: The built-in high-sensitivity music sensor allows you to sync the TV led lights to your favorite music. Lights react and move to the beat of music, film and game audio, creating an incredible immersive sound and light experience – and all without the need for an additional hub. You can choose from 64+ patterns, including strobe, illusion, jump, energy, gradient, running, chasing, fluttering mode etc. Each mode can be applied to different scenes
- 64+ Chasing IC+RGB & W Effect: The independent control design of IC+RGB and white all-in-one multifunctional rgbw led bar lights monitor backlight allows you to easily switch the led lights for desk between IC+RGB and white to enjoy a variety of lighting effects, you can choose white lighting mode with high brightness for better illumination when working, etc. Choose IC+RGB lighting mode when you need to create an entertaining atmosphere for the light show, 16 million colours of Color changing rgb light bar let you dance to the beat and create a dynamic party atmosphere
- Dynamic and Dimmable Lights: 1%-full brightness free brightness adjustment gives you the flexibility to adjust dimmable led light bar for room decorations or a variety of occasions
- USB Powered: USB light bars for room can be powered by powernank, usb charger, TV, Monitor, laptop PC mainframe, making behind TV led lighting bar easy to install in any location.TV led lights are powered by the TV usb interface / PC mainframe, can be turned on/off synchronously with TV / PC
| Intervention | Scope | Reversibility | When it fits | Main trade-off |
|---|---|---|---|---|
| Correct the workload | Application, query, job, or configuration | Depends on the change | A repeatable loop, oversized batch, retry storm, or scheduled task is the cause | Requires understanding the application and may need deployment work |
| Restart the managed service | One systemd service | Temporary until the cause recurs | A known daemon failure mode is causing the spike and restart impact is acceptable | Brief interruption and possible loss of in-memory state |
| Stop the managed service | One systemd service and its cgroup | Temporary | Containment is more important than availability and ownership is confirmed | Service becomes unavailable |
| Set a cgroup CPU limit or weight | Service or cgroup | Runtime or persistent, depending on how configured | You need containment or prioritization while investigating | Can reduce throughput or increase latency for that service |
| Set CPU affinity or isolation | Selected process, service, or CPUs | Usually reversible | Measurements show a placement requirement for a specific workload | May strand capacity or worsen performance if not validated |
Correct the workload first
Look for the direct trigger: a runaway retry loop, an inefficient query, an unexpectedly large batch, a scheduled job colliding with peak traffic, or a game-server task generating work faster than it can complete. Reduce the workload, correct the query or loop, adjust the batch size, or reschedule the job. This is safer than masking the symptom because it removes the source rather than merely rationing its CPU.
Restart or stop a service only after confirming ownership
If the PID belongs to a managed daemon, use its systemd unit rather than killing an arbitrary process. Before restarting, inspect status and logs, identify the failure mode, and decide whether the interruption is acceptable. A restart is appropriate for a known recoverable fault; stopping is an emergency containment measure when preserving the host matters more than availability.
Do not delete binaries, erase data, or repeatedly kill a process you cannot explain. Unexpected miners, crashes, or unauthorized workloads require an investigation of package, authentication, and network logs. Preserve evidence and determine how the process started before removing it.
Contain a service with systemd or cgroups
When immediate containment is needed, systemd can apply cgroup properties to a service with systemctl set-property. CPU limits cap consumption; CPU weights change relative priority when CPU is contested. Apply the smallest change that protects critical work, document the value and scope, and understand whether the setting is runtime-only or will persist according to the command and unit configuration you choose.
Containment is not a cure. A cap can lengthen queues, increase request latency, and leave a faulty job running longer. Monitor the affected service and the host while the limit is active, then remove or revise it after the underlying workload is corrected.
Use affinity only for a measured requirement
Affinity and isolation can reserve CPUs or place a workload on selected CPUs, but proving that a process moved does not prove that performance improved. Placement can reduce cache locality, strand idle capacity, or push other work onto fewer cores. Treat affinity as an experiment: measure application throughput and latency under identical load before and after, and keep the change only when the service objective improves.
Best Value
- Compatible with Wide Screens - To ensure compatibility with the dual monitor mount, your each monitor must meet three conditions at the same time: First, computer screens size range: 13 to 32 inches. Second, screen weight range: 4.4 to 19.8 lbs. Third, the back of the monitor screen must have VESA mounting holes with a pitch of 75x75mm or 100x100mm.
- Regarding the compatibility with desks - Your desk must meet three conditions at the same time: First, desk material: Only wooden desks are recommended, plastic or glass desks cannot be used. Second, desk thickness range: 0.59" - 3.54". Third, the bottom of the desk should not have any cross beams or panels, as this will interfere with installation. We recommend carefully checking that your desk and monitors meets all above conditions before purchasing.
- Dual C-Clamp Hold - Worried your dual monitors might wobble or slip? Our upgraded base uses a larger platform plus a dual C-clamp structure to lock the dual monitor arm firmly to your desk. Each arm safely keeps your screens steady while you type, click and game—no shaking, no sliding, just a clean and secure setup you can trust every day. It also provides Grommet Mounting installation choice, both options ensure stable and secure fixation for your 0.59" - 3.54" desk.
- Full-Motion Adjustment For Comfortable View - Pull the screen closer when you’re deep in a spreadsheet, push it back to watch videos, or rotate to portrait for coding — moving everything smoothly with just one hand. The monitor stand offers +85°/-50° tilt, ±90° swivel and 360° rotation. Raise your monitor up to 15.75″ to support a healthy sitting posture. Whether you’re working from home, gaming through the night, or switching between video calls and documents, getting the screens to your natural line of sight helps relieve neck, shoulder and back strain so you can stay focused longer with less fatigue.
- Keep Your Desk Organized: By lifting both screens off the desktop, this dual monitor stand opens up valuable space for your keyboard, notebook, docking station or a simple, clutter-free work area. Built-in cable management guides wires along the arms, keeping cords out of sight and out of the way. Enjoy a tidy, modern workstation that looks as good as it feels to use.
Step 6: Validate the fix instead of trusting a lower percentage
Repeat the same observation window and, where possible, the same workload used for the baseline. A lower CPU number is not success if requests are slower, queues are longer, or errors have increased.
Compare these measurements
- CPU utilization by host and by process or thread.
- One-, five-, and 15-minute load averages.
- User, system, and wait CPU shares.
- Application latency, throughput, and error rate.
- Context-switch behavior from
pidstat. - Disk or network latency when the original evidence indicated waiting.
Use identical workload intensity and observation duration for before-and-after comparisons. If the spike was caused by a one-time job, compare the same job phase rather than an idle period. Keep a change only when the user-visible service objective improves; an affinity mask or a capped percentage is not proof by itself.
Quick wins for a faster PC:
Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Common failure modes and recovery paths
The top process is not actually the bottleneck
If CPU is moderate, wait is high, and many tasks are in D, investigate storage and filesystem behavior. Remove CPU limits you added for a workload that is not CPU-bound, and address the I/O path instead.
A PID disappears before you can inspect it
The task may be a short-lived child or scheduled job. Capture repeated ps or pidstat samples, note the parent PID, and inspect the parent service or scheduler. A one-time snapshot can miss the process that launches the real workload.
A service returns immediately after being killed
A supervisor is probably restarting it. Stop or restart the owning systemd unit after confirming the cgroup and service name; otherwise the supervisor will recreate the process and the CPU spike will continue.
A CPU cap hides the symptom but the service becomes slow
Remove or raise the cap, then correct the workload that caused the spike. Compare latency and throughput, not just CPU utilization. If the service is competing with critical work, use a measured priority or scheduling policy and validate it under representative load.
Crashes, No Sound, or Screen Glitches?
Random freezes, missing sound and display glitches usually trace back to one bad driver. Find and replace yours safely.Free scan · under a minutePC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11Affinity changes produce no improvement
Restore the previous placement and investigate the original bottleneck. A placement change can be technically successful while leaving throughput and latency unchanged or worse.
Operating practice after the incident
Keep the baseline commands and the successful remediation in the service’s runbook. Record which PID belonged to which unit, what workload triggered the spike, what evidence separated CPU work from I/O wait, and which measurements justified the final change. Ongoing Linux server CPU monitoring with process-level CPU alerts can shorten the next investigation by preserving per-process history instead of waiting for a user to report slowness.
Alerts should include host CPU, load averages, wait time, and the identity of the consuming process or service. A CPU-only alert can miss I/O-bound incidents; a load-only alert can page on blocked storage. Pair thresholds with a short history and the service’s latency or error objective so operators can distinguish a brief batch burst from a sustained failure.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
