Concepts / Using Jupyter Notebooks with AWS

Using Jupyter Notebooks with AWS

A recent NVIDIA GPU is recommended for faster execution, not required for running the examples.

  • Programming

Choose the Computation Location

Before running the deep-learning code examples in a Jupyter Notebook, make one practical decision: where will the computation take place? The two broad routes are a local machine with a recent NVIDIA GPU or a cloud environment that provides access to GPU-based computing. This decision affects the execution experience, but it does not determine whether the examples are allowed to run at all.

beginyesnoNotebook examplesChoose a computationlocationRecent NVIDIA GPU?Available locallyLocal machineRecommended local routeCloud environmentAWS GPU instance or GoogleCloud instance
Given whether a suitable local NVIDIA GPU is available, which execution path should you choose?

Trace the First Decision

Choosing Between Local and Cloud Execution

You want to run the notebook examples and have a local workstation with a recent NVIDIA GPU.

Inspect the local workstation: The workstation matches the recommended local setup because it has a recent NVIDIA GPU.

Choose the local route: Run the notebook on the local machine so the computation takes place there.

Change the route if the workstation is unavailable: If the suitable local workstation is not available, use a cloud environment that provides GPU-based computing instead.

A recent local NVIDIA GPU leads to the recommended local route; otherwise, a cloud GPU environment is a practical alternative.

The decision is not a choice between running and not running the examples. It is a choice between execution environments. A recent NVIDIA GPU gives the local route the recommended performance advantage. If that local setup is unavailable, the source identifies cloud environments as alternatives, including Google Cloud instances and Amazon Web Services GPU instances.

Recommended Does Not Mean Required

A recent NVIDIA GPU is recommended for faster execution, but it is not required for running the examples.

This distinction is central. The recommendation describes the better-performing experience, not a strict entry condition. A TITAN X is given as an example of the recommended kind of GPU. Without the recommended local GPU, the examples can still be run; the practical question becomes which available execution environment is most suitable.

supportsrunscan still runcan still runRecent NVIDIA GPURecommended local setupFaster executionRecommended experienceOther setupNot the recommended localGPUExamples can runGPU is not mandatoryNo GPUGPU not available locally
What changes when a recent NVIDIA GPU is available, when another setup is used, or when no GPU is available?

Follow the Computation Path

The notebook is the place from which you work with the examples, but the important execution question is where the computation takes place. In the local route, the local machine performs the computation. In the cloud route, a cloud environment supplies access to GPU-based computing. For this topic, an AWS GPU instance is one named cloud option.

usesperformsusesprovidesJupyter NotebookLocal routeLocal machineRecent NVIDIA GPUrecommendedLocal computationExamples execute locallyJupyter NotebookCloud routeAWS GPU instanceCloud-based computingCloud computationExamples execute in thecloud environment
Where does the notebook computation run in a local environment compared with an AWS-based environment?

The cloud route replaces the need for access to a suitable local workstation. It does not change the underlying recommendation: GPU-based computing is useful because it affects how quickly the examples run. The difference is where that computing resource is supplied.

Understand the Speed Trade-Off

The practical benefit of a recent NVIDIA GPU is execution speed. The source recommends one for faster execution, so the presence of a recent NVIDIA GPU changes the expected pace of the examples rather than changing the basic route from notebook to computation.

recommended fordoes not preventRecent NVIDIA GPURecommendedFaster executionRecommended experienceWithout recentNVIDIA GPUStill not automaticallydisallowedExamples runGPU is not required
How does a recent NVIDIA GPU affect execution time compared with running without the recommended GPU?

Make the execution-location decision before beginning the deep-learning examples. If a recent NVIDIA GPU is available locally, use that recommended local setup. If it is not available, select a cloud environment with GPU-based computing, such as an AWS GPU instance or a Google Cloud instance.

Mistakes Beginners Make

  • Treating a recent NVIDIA GPU as mandatory.

    The GPU is recommended for faster execution, not required for running the examples.

    Fix: Use the available route, including a cloud environment when a suitable local workstation is unavailable.

  • Assuming that the only alternative to a recommended local workstation is to stop.

    The source identifies cloud environments as alternatives that can provide access to GPU-based computing.

    Fix: Consider an AWS GPU instance or a Google Cloud instance.

  • Choosing a cloud environment without identifying why it is being used.

    The first practical decision is the computation location: local machine or cloud environment.

    Fix: Decide whether the examples will compute locally or in a cloud environment, then select the matching route.

Practice the Route Choice

EASY

For each situation, choose the most practical route and explain whether the GPU is a recommendation or a requirement: a local workstation has a recent NVIDIA GPU; a suitable local workstation is unavailable; no recent NVIDIA GPU is available locally but you still want to run the examples.

Hints
  • A recent NVIDIA GPU supports the recommended local setup.
  • The absence of a suitable local workstation points toward a cloud environment.
  • The GPU recommendation affects execution speed, not whether the examples may be run.

What do you think happens?

You do not have a suitable local workstation. Which route does the source identify as the practical alternative?

  • Use a cloud environment with GPU-based computing
  • Treat the examples as impossible to run
  • Require a recent NVIDIA GPU before continuing
Reveal answer

Answer: Use a cloud environment with GPU-based computing, such as an AWS GPU instance or a Google Cloud instance.

The recent NVIDIA GPU is recommended for faster execution, but it is not a strict requirement. Cloud environments can replace the need for access to a suitable local workstation.

Execution Route Summary

  1. Decide where the notebook computation will take place before running the deep-learning examples.
  2. A recent NVIDIA GPU, such as the source's TITAN X example, is recommended because it supports faster execution.
  3. The recommended GPU is not mandatory; the examples can still be run without that local setup.
  4. When a suitable local workstation is unavailable, use a cloud environment with GPU-based computing.
  5. AWS GPU instances and Google Cloud instances are the named cloud alternatives in the source.

Key Takeaways

  • Choose the computation location first: local machine or cloud environment.
  • A recent NVIDIA GPU is recommended for faster execution, not required for running the examples.
  • A local workstation with a recent NVIDIA GPU is the recommended local route.
  • AWS GPU instances and Google Cloud instances are alternatives when a suitable local workstation is unavailable.