Datenquelle: TechPowerUp GPU-Datenbank (über die dbgpu-Python-Bibliothek)
Werkzeuge: Python (Pandas, Plotly)
GitHub: https://github.com/BryceDonston/gpu-trend-data
Diagramm 1 konzentriert sich auf die Moderne (2015–2025) mit einer hervorzuhebenden linearen Skala
die dramatische Explosion der Transistorzahlen. Diagramm 2 zeigt die gesamten 30 Jahre
Geschichte mit einer logarithmischen Skala, um den Verlauf des Mooreschen Gesetzes zu veranschaulichen
alle GPU-Generationen.
Von Low_Feed2001
3 Kommentare
The last chart could use showing which GPUs are counted, because it seems very weird that transistor count would drop by 3-5x unless you are doing something weird like only picking biggest GPU released that year or something
I almost forgot Matrox. Good old times
Transistors need to be in the same area for this to be Moores law. Otherwise you could just add more chips to get more transistors.
Moores law ended half a decade ago as we are now limited by physical restraints on how small transistors can be.