Conseils et ressources - AIDA64 | Bonnes pratiques et aide-mémoire

Benchmarking Best Practices

Close other heavy applications before running benchmarks so results are consistent.
Run CPU benchmarks multiple times; the first run may be lower until the CPU reaches steady temperature.
For GPU benchmarks, ensure the GPU is not throttling (cool it down first or run a few warm-up passes).
When comparing systems, use the same AIDA64 version and similar settings (e.g. same stress test duration).
Save reports (HTML or CSV) so you can compare before/after overclocks or hardware changes.

Monitor temperatures during the System Stability Test. If CPU or GPU exceed safe limits (e.g. 95°C+ for many CPUs), stop and improve cooling.
Start with shorter runs (10–15 minutes) to verify cooling; extend to 30–60 minutes or more for overclock validation.
Do not run stress tests 24/7 unless you have enterprise cooling; it shortens component life.
If the system crashes or reports errors during a memory stress test, suspect RAM or memory controller; test with fewer sticks or lower speed.

Put the most important sensors at the top: CPU temperature, GPU temperature, and maybe one or two fan speeds.
Use gauges for quick visual feedback and graphs for trends over time.
Match the panel size to your display or external LCD so it fits without scaling artifacts.
Export your layout (backup the SensorPanel configuration) so you can restore it after an update or on another PC.
On external LCDs, avoid too much text; use large numbers and simple gauges for readability at a glance.

Place AIDA64 on a network share that all PCs can read. Use a logon script to run aida64.exe with /R (report path), /AUDIT, /CSV or /XML, /SAFE, and /SILENT.
Use /SAFE if some PCs hang or crash during audit; it disables low-level access that can cause issues on odd hardware.
Store reports in a folder structure that includes the PC name (e.g. \\server\reports\$HOSTNAME\report.csv) so you can identify each machine.
Schedule a task or run the script at logon. On Windows 8.1+, be aware of the default 5-minute logon script delay; disable it via Group Policy if you need immediate execution.
For SQL output, ensure every client can reach the database and that MDAC/drivers are up to date on problematic machines.

Use Quick Report for a fast HTML or text summary. Use Full Report for every detail when you need complete documentation.
Disable debug information in File → Preferences → Report if you want shorter, cleaner reports for sharing.
CSV is ideal for importing into Excel or other tools; XML is good for automated parsing; SQL is best when you want a central database.
Include the report date and AIDA64 version in the filename or report header so you can track when the data was collected.

DMI/SMBIOS: Standardised tables in the motherboard firmware that describe hardware (manufacturer, model, serial, etc.). AIDA64 reads these for the Computer and Motherboard pages. Accuracy depends on the vendor filling them correctly.
SensorPanel: AIDA64's feature to build custom on-screen or external LCD panels that display live sensor data (temperatures, voltages, fan speeds) using gauges, graphs, and text.
SMART: Self-Monitoring, Analysis and Reporting Technology for storage drives. AIDA64 can read SMART attributes to warn of potential drive failure (e.g. reallocated sectors, pending sectors).
System Stability Test: AIDA64's built-in stress test that loads CPU, memory, disk, and optionally GPU to check system stability under full load.
TJMax: Maximum junction temperature for Intel CPUs. Core temperatures are reported relative to TJMax. You can adjust TJMax in AIDA64 Preferences to align with other monitoring software.

Run aida64.exe from a script or scheduled task with options such as:

Full list: Help → Command-line Options in AIDA64, or the User Manual on the official site.