Against bias in Hardware Benchmark Reviews
Bias in Benchmark Reviews
There are multiple problems with lots of popular Benchmark reviews and people who are doing or posting them.
- They get paid or directly their products from the corpo they review. Sponsoring, advertisement, marketing, ...
- There is an NDA that needs to be signed.
- There is a code of conduct which disallows saying X.
- There is a code of conduct disallowing them to post things on platform X.
Results can be misinterpreted
Does it apply in the real-world or is it only a synthetic Benchmark. Most people do not give much about theoretical scores.
What should a Benchmark review minimum include
I make it quick and list specific categories that needs to be included to get the whole picture.
- Objective title
- Architecture analysis, are there new changes compared to the old one
- Core analysis
- Performance per Watt analysts
- OS power profiles and the influence
- Synthetic benchmarks
- Real-world in-game benchmarks
- Application performance with real-world apps
- Compression and decompression
- Turbo, boost and maximum clock
- RAM configuration as well as impact of RAM
- OS tests across different platforms, Windows and Linux
- Bottleneck benchmarks
- Security changes and the impact on performance
- Comparison to previous generation
- Conclusion if it is worth to update when you posses generation x on hardware already
- Overclocking aspect
- Usability like drivers Drivers
- Maintenance cost as well as platform switching in consideration
This is currently a draft I am working on and not the final version, please do not link it.