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Clear out junk files and repair common Windows errorsFree Scan →Scan for outdated or missing drivers - takes under a minuteDriver Scan →Repair Windows errors before they cause bigger problemsFix Now →Run Ollama without GPU access by using the official ollama/ollama image, a persistent /root/.ollama volume, and port 11434. A CPU-only Compose file contains no GPU reservations, NVIDIA runtime options, ROCm image tag, or device mappings.
Antihost offers Linux VPS plans with full root access, unlimited traffic on a 1 Gbps port and a 14-day free trial.
Save the Compose file, start it with docker compose up -d, then launch a model with docker compose exec ollama ollama run llama3.2. The instructions below also show how to keep model files across container replacement, verify the API, restrict network exposure, and diagnose common failures.
As an Amazon Associate I earn from qualifying purchases.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
What you need before starting
You need Docker Engine with the Compose plugin, or a Docker Desktop installation that provides the docker compose command. The host must have enough memory, processor capacity, and disk space for the model you intend to run. Model downloads are stored locally, so the required disk space grows as you add models.
CPU compatibility
Ollama can select an appropriate CPU runner automatically. Most installations should leave that autodetection untouched. If you later force a specific runner, the CPU must support the instruction set required by that runner; the documented cpu_avx2 override, for example, requires AVX2 support and a matching runner in the image.
#1 Best Overall
- [Ultra-Lightweight Ear Buds Designed for Small Ears] Each earbud weighs only 3.7g and features a compact, ergonomic in-ear design made especially for small ears. Secure, low-profile, and comfortable for workouts, all-day wear or overnight sleeping.The ultra-slim profile sits flush in your ear with no hard edges poking or rubbing, so you can even sleep on your side without discomfort.
- [Immersive Stereo Sound with TOZO OrigX Technology] TOZO OrigX tuning delivers clear vocals, balanced mids, and natural stereo sound for music, podcasts, and videos.
- [Long Battery Life for Daily Use] Get up to 7 hours of playtime on a single charge, with up to 32 hours total using the charging case—ideal for workdays, commuting, and extended listening sessions.
- [Bluetooth 5.3 & Stable Connection] Bluetooth 5.3 provides fast pairing, stable wireless performance, and reduced dropouts as you move around home or office.
- [Deep Bass with Clear Vocals] High-performance drivers produce punchy bass while keeping vocals clean and detailed for everyday listening.
Why Compose is useful here
Compose keeps the image, port, volume, and restart policy in one declarative file. You can recreate or upgrade the container without manually reconstructing command-line flags, while the named volume keeps downloaded models separate from the disposable container filesystem.
The minimal CPU-only docker-compose.yml
Create a directory for the service and save this file as docker-compose.yml or compose.yml:
services:
ollama:
image: ollama/ollama:latest
container_name: ollama
ports:
- '11434:11434'
volumes:
- ollama:/root/.ollama
restart: unless-stopped
volumes:
ollama:This is intentionally small. The official image supplies the Ollama server, the port mapping makes the service available on port 11434, and the named volume maps persistent storage into Ollama’s standard data directory.
What each setting does
| Setting | Purpose | Operational consequence |
|---|---|---|
image: ollama/ollama:latest |
Uses the official Ollama container image. | The image includes CPU runtime support. The latest tag can change when a newer image is published. |
container_name: ollama |
Assigns a predictable container name. | Commands such as docker logs ollama address this container directly. |
ports |
Publishes host port 11434 to container port 11434. | Clients on the host can use http://localhost:11434. |
ollama:/root/.ollama |
Mounts the named volume at Ollama’s data directory. | Downloaded models survive container recreation. |
restart: unless-stopped |
Asks Docker to restart the service after routine failures and host restarts. | The service comes back automatically unless you explicitly stop it. |
Start the container and run a model
- Open a terminal in the directory containing the Compose file.
- Start the service in the background:
docker compose up -dCompose creates the named volume if it does not already exist, downloads the image when necessary, and starts the container.
- Check the service state:
docker compose psThe
ollamaservice should be listed as running. If it is still starting, inspect its logs before trying a model. - Launch a model from inside the container:
docker compose exec ollama ollama run llama3.2The first run downloads the model into
/root/.ollama. Subsequent runs use the copy in the volume. Replacellama3.2with another model name supported by your Ollama installation.
Call the HTTP API from the host
After the server is running, query the tags endpoint:
curl http://localhost:11434/api/tagsA successful response is JSON describing models already present in the volume. An empty model list is normal before the first download; a connection error means the host cannot reach the published port and should be investigated using the checks in the troubleshooting section.
Rank #2
- 【IP68 Waterproof Multifunctinal Smart Watch】Torjalph fitness activity tracker smart watch is IP68 waterproof technology ,you can wear it when swimming,handwashing,it's rain-proof which means that you perform multiple exercise modes outside without worrying about sweat or rain corroding your fitness smart watch,and this watch equipped with a 230 mAh large capacity battery magnetic charging ,and this sports activity tracker watch can be used 5-7 days If fully charged.
- 【Larger 1.83'' Full Screen Watch & Multi-dial Selection】1.83 inch full HD color touch screen smart watch for men / women provides excellent 240*284 resolution image quality ,This Bluetooth smart watch offers a better touch experience and remarkable immersion and ,more than 200 watch faces for you to choose from App"Da Fit ", you can change the dial style according to your mood to wear keeping the world on your wrist
- 【Accurate Activity Tracker Smart Watch 】sports watch activity tracker uses sweat and dust resistant breathable silicone strap, our Fitness Tracker Smart Watch for Android iOS Phones supports 12 sports modes,walking,running,bicycle,swimming,badminton,basketball,football,rope skipping. This bluetooth calls smart watch for men women is lightweight and stylish, bringing more convenient and relaxing sports experience while also providing more protection.
- 【Your 24 Hour Personal Daily Health Monitor】Fitness watch for women men full touch screen IP67 waterproof bluetooth smartwatch for Android iOS Phones with heart rate blood pressure monitor multifunction Andriod smart watch with steps counter sleep monitor sports fitness watch, monitoring blood oxygen saturation allows you to understand your physical condition in real time to track fitness progress ,to know your health status in time.
- 【More Functions Bluetooth Call Smart Watch】This fitness activity smart watch full screen for iPhone and Android phones includes a bunch of new features including:The bluetooth smart watch can hang up the phone, receive and display WhatsAPP /Text /Facebook /Twitter /Messenger reminder.Support sedentary reminder, alarm clock, stopwatch, timer,weather information,music control, remote camera control compatible with Android 5.0 and above or iOS 9.0 and above smartphones.
Why this configuration is CPU-only
CPU mode is achieved by not granting the container any accelerator access. Docker does not expose a host GPU to a container merely because one exists; an explicit runtime, reservation, or device mapping is normally required. The Compose file above contains none of those settings.
| Acceleration type | Settings normally required | What the CPU-only file does |
|---|---|---|
| CPU | Standard container start with the official image and no accelerator mapping. | Uses this path by omission of GPU-specific options. |
| NVIDIA GPU | A GPU-enabled Docker configuration, such as a --gpus=all style runtime request or an equivalent Compose reservation. |
Contains no NVIDIA runtime or GPU reservation. |
| AMD GPU | The rocm image variant plus access to /dev/kfd and /dev/dri. |
Uses neither the ROCm tag nor those device mappings. |
Do not add accelerator settings “just in case.” They change the execution path and can make a file fail on hosts that do not have the corresponding runtime or devices. A CPU-only deployment should also avoid copying GPU examples from a different operating-system guide.
Bind the API only where it is needed
The short port form in the example is convenient for a single-user setup. Docker commonly publishes that form on all host interfaces. If the service is intended only for applications on the same machine, bind the host side explicitly to loopback:
ports:
- '127.0.0.1:11434:11434'With this mapping, requests from the local host can use http://localhost:11434, while other machines cannot reach the published port through the host’s network addresses. If another local service already uses port 11434, choose a different host port while keeping the container side at 11434, for example 127.0.0.1:11435:11434; clients would then connect to port 11435 on the host.
Persist models with the named volume
Ollama stores downloaded models below /root/.ollama. The ollama named volume maps that directory outside the container’s writable layer. Removing and recreating the container therefore does not require every model to be downloaded again.
Recommended Free Tools
Rank #3
- [Low Noise High Fidelity] New bee guitar cord with 22AWG oxygen-free-copper center conductors provides enhanced signal transfer, the PVC insulation material and the high-density stranded copper double shield provide noise rejection for quiet operation
- [Ideal for Choice] New bee 1/4 straight and right angle metal connectors, ensures a secure connection in any instrument, pedal, or amp jack. L shape great fro on stage guitar stands
- [Wide Compatible] Suitable for mixer, power amplifier, microphone, effector, decoder, equalizer, electoronic organ, electric guitar, electric bass and other electronic equipments. Widely used in the installation of smart home, karaoke speaker, professional stage audio, etc
- [Easy Disassemble] Making incidental repairs quick and easy! Very nice, flexible cable
- [Dont Hesite to Order] 24 hours customer services and professional technology team are standing by
Operations that keep the volume
docker compose downremoves the Compose-managed container and network but, by default, leaves the named volume.docker compose up -dattaches the existing volume when the project and volume definition are unchanged.docker compose up -d --force-recreatereplaces the container while retaining the same volume.
Operations that delete model data
Deleting the volume, changing the volume name, or mounting a different host path means the new container no longer sees the old model files. Treat commands that include volume removal as destructive and verify that you have another copy before using them.
Verify the service and read its logs
Use these checks in order:
- Confirm Compose sees the service:
docker compose ps - Follow the service output while it starts or downloads a model:
docker compose logs -f ollama - Test the API endpoint from the host:
curl http://localhost:11434/api/tags
Container output is written to standard output and standard error, so Compose logs and Docker logs show the same primary diagnostic information. Because the file assigns the name ollama, the equivalent direct command is:
docker logs ollamaStop following a live log stream with the terminal interrupt key; this does not stop the container itself.
Troubleshoot common failures
The API is unreachable
- Run
docker compose ps. If the container is stopped or repeatedly restarting, readdocker compose logs ollamabefore changing the file. - Check that the host port in the Compose file is not already occupied. A changed host port must also be used in the URL you test.
- Confirm that the service was started from the directory containing the intended Compose file. Running Compose from another project can create or address a different container and volume.
- If you bound the port to
127.0.0.1, test from the same host. Other machines will not be able to use that loopback binding.
Models download again after recreation
Inspect the Compose file for an unchanged ollama:/root/.ollama mount. A typo in either side of the mapping, a renamed volume key, or a different project directory can point the replacement container at a new empty volume. List Docker volumes if necessary and compare the active mount shown by docker inspect ollama with the intended path.
The container appears to use an accelerator
Remove GPU reservations, runtime settings, device mappings, and accelerator-specific image tags. The CPU image should be the standard ollama/ollama image, not the rocm variant. Recreate the service so that old container settings are not retained:
Rank #4
- Digital Family Calendar: BSIMB digital calendar wall touch screen puts an end to your chaotic life. Designed for busy families, the digital calendar keeps you organised and in step with your family. Assign different colours and tasks to each family member to make family scheduling easier and more intuitive.
- Easy Setup and Auto-Sync: Plug in, connect to WLAN and auto-syn your electric calendar in minutes. Digital wall calendar compatible with Google calendar, iCloud calendar, Outlook calendar, Cozi calendar and Yahoo calendar. Add events directly on the device or through the ecalendar app. Electronic calendar stays current whether you're at work or traveling—no manual updates needed.
- All-In-One Home Assistant: Smart calendar builds habits and encourage independence with personalised chores. Display daily meal plan to reduce the daily ‘What's for dinner?’ stress. To-do lists keep track of everything you want. Turning chores into valuable achievements with stars and rewards, making life easier and more enjoyable.
- Manage Anytime Anywhere: BSIMB electronic calendar for family stay connected with the powerful ecalendar app. Access and manage your digital wall calendar anytime, anywhere with free features like stars & rewards, photo screensaver, chore charts, meals plans, to-do lists etc. Electronic planner paid features like Magic Import tailor advanced tools for users with special needs.
- Stay in Touch with Loved Ones: Share photos or videos from your phone to electronic wall calendar with app, instantly update your life moment, and keep in touch with your family, especially those who are far away; you can also surprise your smart calendar digital wall by sending him a wish card; it's also great for keeping tech-illiterate elders in touch with the youngsters.
docker compose down
docker compose up -d --force-recreateThen inspect the startup logs and run a model again. A fresh container is important because editing the YAML file does not modify an already-created container until Compose recreates it.
The service restarts continuously
Read the earliest log messages, not only the latest restart line. Common causes include an invalid image pull, an unsupported forced CPU runner, a port collision, or insufficient host resources. Temporarily remove nonessential environment overrides, confirm the standard image name, and start the service in the foreground with docker compose up so the first error remains visible.
A model command fails while the API is healthy
Separate server reachability from model availability. First query /api/tags to see what is installed. Then run the model command inside the same container with docker compose exec ollama ollama run model-name, replacing model-name with the exact model identifier. A model that is not present may need to be downloaded, and a model that exceeds available CPU memory may fail even though the server itself responds.
When to force the CPU runner
Autodetection is the preferred default. If troubleshooting shows that autodetection selected an unsuitable library, the documented override can be added to the service:
services:
ollama:
image: ollama/ollama:latest
container_name: ollama
ports:
- '127.0.0.1:11434:11434'
volumes:
- ollama:/root/.ollama
environment:
- OLLAMA_LLM_LIBRARY=cpu_avx2
restart: unless-stopped
volumes:
ollama:OLLAMA_LLM_LIBRARY=cpu_avx2 is an override for autodetection, not a universal compatibility switch. Use it only when the image contains that runner and the host CPU supports the required AVX2 instruction set. If the container fails after adding it, remove the variable, recreate the service, and return to autodetection.
Upgrade the image without losing models
The named volume and the container lifecycle are independent. To fetch a newer image and recreate the service:
- Pull the image declared by the file:
docker compose pull ollama - Recreate the service in the background:
docker compose up -d - Check the logs and API, then run a previously downloaded model to confirm that the existing volume is attached.
Using latest follows new image releases automatically when you pull. If predictable upgrades and rollbacks matter, replace it with a version or digest that you have chosen and keep that value in source control. Whatever tag you use, do not change the volume mount while upgrading.
Free tools Windows power users keep installed
One-click scans. No signup required.
CPU performance and resource planning
There is no single CPU performance figure that applies to every Ollama deployment. Response speed depends on the selected model, model quantization, available system memory, CPU architecture and instruction support, concurrent requests, and storage speed during loading. Official documentation does not publish a universal CPU benchmark suitable for predicting tokens per second on an arbitrary host.
Practical planning points
- Start with one model and observe memory use and response latency before adding concurrent workloads.
- Keep model storage on a reliable local disk; repeated downloads caused by a missing volume waste both time and bandwidth.
- Expect the first invocation of a model to take longer because the files must be downloaded and loaded.
- Choose a model that fits the host’s available memory. A healthy HTTP endpoint does not guarantee that every model will load successfully.
- For sustained workloads, monitor the host rather than assuming that a GPU-oriented example or benchmark predicts CPU behavior.
Compose versus a native Ollama installation
Both approaches run the same Ollama service, but they put operational responsibility in different places.
| Decision axis | Docker Compose | Native installation |
|---|---|---|
| Installation isolation | Ollama and its runtime are packaged in a container, separate from most host libraries. | Ollama integrates directly with the host operating system and its service manager. |
| Model-volume management | A named Docker volume explicitly owns /root/.ollama and follows the Compose lifecycle. |
Models live in the native installation’s configured data location. |
| Upgrade and rollback control | Image tags or digests can be changed independently of model data; recreation is declarative. | The host package or installer controls the update path. |
| Networking | Port publishing determines how the container API reaches the host and other clients. | The native service listens directly according to its host configuration. |
| Startup behavior | restart: unless-stopped lets Docker restart the container after failures and reboots. |
Startup and restart depend on the operating system’s service configuration. |
| CPU instruction compatibility | The container image supplies the runtime, but the host CPU still must support the selected runner. | The native runtime also depends on host CPU support and its installed libraries. |
| Operational complexity | Requires Docker, Compose, image management, and volume awareness. | Avoids container administration but places more responsibility on host packages and services. |
Choose Compose when repeatable container configuration, isolated dependencies, or easy service recreation matter most. A native installation can be simpler when Docker is not already part of your operating environment.
Platform and scope notes
The CPU-only file is deliberately platform-neutral at the Compose level: it does not request Linux GPU devices or an accelerator runtime. Docker Desktop behavior differs by host operating system, so do not paste Linux GPU mappings into a CPU deployment merely because the host documentation shows them.
Official guidance distinguishes CPU-only Docker from NVIDIA and AMD configurations. It also notes that GPU acceleration is unavailable for Docker Desktop on macOS. A CPU-only container avoids that GPU limitation because it does not depend on GPU passthrough.
Quick Recap
Operational checklist
- Use the official
ollama/ollamaimage. - Keep the
ollama:/root/.ollamavolume mount unchanged. - Publish container port 11434, preferably as
127.0.0.1:11434:11434for a local-only service. - Do not add
--gpus-style settings, GPU reservations,/dev/kfd,/dev/dri, or therocmtag. - Start with
docker compose up -dand verify withdocker compose ps. - Use
docker compose logs ollamaas the first diagnostic step. - Run models with
docker compose exec ollama ollama run model-name. - Use
OLLAMA_LLM_LIBRARY=cpu_avx2only when autodetection is unsuitable and the CPU and image support that runner. - Recreate the container after configuration changes so old runtime settings are removed.
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.
