How are my scores determined?

Performance values and scores are determined by the resources of the graphics card (GPU) and the main processor (CPU) provided to the Redshift renderer during a benchmark test.

Since the performance of a processor (CPU or GPU) is influenced by different factors, it is possible that two systems with the same hardware achieve different performance values. If the values vary upwards and downwards again, this is normal.

If you notice a system is markedly slower than the same hardware of others, any of these components may have affected the result:

• Can your computer be cooled sufficiently? 
  Both the ambient temperature and the temperatures inside the case can influence the performance values. If the processor gets too hot, you will get different values than a processor with good cooling.
• Can your power supply provide enough power?
  If a power supply is too weak or has fluctuations in the output power, it can affect the GPU and thus, in addition to poorer values, also the stability of the graphics card.
• Is the CPU / GPU overclocked?*
  While most processors can tolerate overclocking, if the cooling of the system and the performance of the power supply aren't adequate, both points above come into effect.
• Did you run the test for less than 10 minutes?
  Tests should run for 10 minutes on account processors heat up over time, which in turn affects their speed. Performance values may be higher for shorter tests, but if the test runs for 10 minutes, an average value is calculated which then represents a more accurate performance measurement.
• Is your graphics card driver up to date? 
  Since manufacturers ensure that drivers receive performance improvements and bug fixes as quickly as possible, GPU drivers can quickly become outdated. For Windows users this means downloading the driver directly from the developer's website; for Mac users this means downloading the latest macOS version.

Remember the values determined by Cinebench should only be regarded as a guideline. 

Technical Notes

• Some Intel CPUs offer P-Cores (performance) and E-Cores (efficiency). We leave it up to the Operating System to select the best Core for the job (which is usually the P-Core). 
• The workload and the measurement is done for this single thread. There's a second thread running for the User Interface output, but its workload is insignificant considering potential measurement errors and what else is happening in the Operating System while measuring (if the single thread would have to deal with User Interface too, then its render performance would be roughly 1-1.5% lower)
• More technical information about Cinebench can be found here.