A computer is a system because its components work together to transform inputs into results. This cooperation brings together hardware, the physical parts, and software, the instructions and data organised to perform tasks. The CPU executes instructions, and RAM holds the working information that programs need. Storage preserves files when the power is off. The motherboard enables components to communicate through interfaces, while the power supply provides suitable energy. Peripherals, such as a keyboard and display, make it possible to enter information or receive results.

These functions become clearer when we follow a task. Editing a poster involves the program, CPU, RAM and display; preserving the changes also involves the save operation and its storage destination. Seeing the poster therefore establishes that there is visible output, but not that the file has been saved. If the application is remote, the pathway also includes network dependencies. A single task needs several functions, and observing one does not automatically verify the others.

A functional inventory connects this pathway with a particular computer: it identifies the model, function and intended task. A specification therefore gains meaning through what it enables and how it relates to the whole system. System information and the authorised computer’s technical documentation allow details to be checked without collecting serial numbers or taking parts apart to discover information that is already available to consult.

RAM and storage can express capacity in bytes, although they serve different functions. Ordinary RAM is volatile working memory: its contents are lost when the power is off. Storage preserves files, so having free space does not mean having enough memory for the current task. A bit is a binary unit, and a byte contains eight bits. Decimal multiples use powers of one thousand: 1 kB equals 1000 bytes. Binary multiples use powers of 1024: 1 KiB equals 1024 bytes. The label tells us which quantity the number represents.

Besides the unit, what is being measured matters. Capacity expresses how much can fit, and occupancy expresses how much space is used relative to the total. Activity describes work over an interval. An occupancy percentage and an activity percentage may therefore match without describing the same situation. Transfer speed relates information to time, as in MB/s. An isolated RAM percentage is insufficient too: it needs the task and time of observation.

Interpreting indicators means connecting them with working conditions. Comparing several resources during an equivalent task helps identify possible limitations without immediately attributing them to a fault. The record preserves the indicator’s name, unit, denominator where applicable, task and time. This separates observation from provisional explanation while retaining tools’ original labels when comparing capacities.

Compatibility brings together the conditions that allow components and software to work together. Having the same function or an apparently matching connector is insufficient. For memory, check generation, form factor, supported capacity and configuration. For storage, interface, dimensions and system support matter. Physical, electrical and logical compatibility must be checked against the specific model’s documentation, rather than a photograph of a similar computer.

A compatible improvement must also address the observed need. A bottleneck is a resource that limits a task under particular conditions; it can change when the workload changes. Adding capacity to a resource that does not limit the task may achieve little. Connect requirement, specification and evidence before comparing options. The classroom cost assumption supports reasoning, but is not a market price. A recommendation includes outstanding questions and conditions that would invalidate it, allowing someone else to review its usefulness before an intervention.

A symptom is what you observe: a task takes time, a file is missing or there is no image. A hypothesis proposes a provisional cause that can be tested. The test should produce different observations depending on whether it is correct. Before running it, write what you expect and which result would make you revise the explanation. This avoids selecting only favourable information afterwards.

Start with reversible, low-impact checks suited to the symptom. Keep the task comparable and change one condition at a time when possible. Record environment, action, result and next decision. Improvement after a change supports a relationship, but does not automatically identify a defective part: alternative explanations may exist. If the symptom cannot be reproduced, preserve the test conditions. The aim is to reduce uncertainty and verify the final task, rather than guess a component immediately or make a chain of changes that hides which one mattered.

A diagnostic tree represents decisions that depend on observations. Each question should separate pathways and lead to a check, rather than an automatic purchase. “It does not work” needs precision: no visible power, power but no image, or an image with a failing task. These situations do not support the same cause. External connections and recent context may provide information before an internal intervention.

The laboratory retains three cases: slowness with many tabs, failure to save to a full destination and no image after moving the computer. For each, connect symptom, possible causes and an authorised test; then verify the original operation. Label fictional records separately from real observations. Internal inspection is limited to the prepared computer, disconnected where appropriate and under supervision. Do not open the power supply or measure mains electricity. Documenting this scope allows the reasoning to be repeated without causing faults.

The life cycle includes obtaining materials, manufacturing, transport, use, maintenance and end-of-life management. Energy during use is only part of the impact. Power expresses the rate of consumption and is measured in W; energy accumulated over time can be expressed in Wh or kWh. For constant assumed power, energy equals power multiplied by time. A maximum power label is not a measurement of actual consumption during a task.

Compare keeping, repairing, upgrading, repurposing and replacing according to need, compatibility, support, accessibility and destination. Extending use requires continued suitability and maintainability. The classroom amounts of 35 and 280 monetary units are fictional repair and replacement comparison data, not commercial quotations. Also state which consumption values are assumptions. Without manufacturing and other life-stage data, usage savings cannot establish a total footprint. When a computer reaches end of life, identify the authorised electronic-waste management channel.

A technical report lets readers reconstruct the relationship between need, evidence and recommendation. Separate observations, inferences and proposals: recording a number is neither interpreting its cause nor deciding a purchase. The inventory, tree and tests should share names and conditions so another person can follow the pathway. A provisional conclusion is valid when it explains what is known, what is missing and which evidence would require a change.

The laboratory integrates the three cases and a revised hypothesis; it creates no third submission. A01 contains design, testable requirements and an already executed small diagnosis. A02 preserves the editable product, before/after tests, correction and a verified improvement. Another pair uses part of the guide and records where help is needed. This review checks communication and procedure as well as appearance. Each member explains a decision and predicts a result for a variation of the case. Screenshots support the record but replace neither running the test nor preserving reviewable files.