Now it's going to be interesting! We test a GeForce GTX 1060 6GB and Radeon RX 580 8GB to document the respective scaling over the maximum possible threads. The averaged Load Pro core is first seen in the summary graph. At first, we deliberately chose the resolution, where both cards run equally into the limiter. This allows us to perfectly control how both cards behave under very similar conditions and how the drivers react.
However, since the component lacks time, we have added the timeline for both graphics cards in the next two single graphics. The Linux representation with the 1200% load refers as a sum to the maximum possible 12 threads with a maximum load of 100% each. So please don't let it get confused:
What is interesting now is that the Radeon RX 580 8GB not only generates more processor load on some of the threads, but also slightly increases the overall load of all threads compared to the Geforce card.
Finally, let's consider the CPU and the clock of all six physical cores. We start with the GeForce and see a relatively balanced behavior at the maximum clock.
The Radeon RX 580 8GB generates a higher processor load in part, but in return also shows slightly stronger clock jumps towards low utilization, which nevertheless do not prevent the Radeon card from running a little more performantly here.
It's not very much, but the Radeon RX 580 uses more than 200 MB more of its own video memory. This, too, is more likely to speak for the AMD card…
while the GeForce, interestingly, needs almost 300 MB more memory. This behavior is also stable reproducible over all test runs.
- 1 - Einführung und Testsystem
- 2 - Gaming-Performance in FHD mit 1920 x 1080 Pixeln
- 3 - Gaming-Performance in WQHD mit 2560 x 1440 Pixeln
- 4 - Gaming-Performance in UHD mit 3840 x 2160 Pixeln
- 5 - Skalierung über Kerne und Threads
- 6 - CPU-Auslastung, Grafik- und Arbeitsspeicherbelegung in Full-HD (1080p)
- 7 - CPU-Auslastung, Grafik- und Arbeitsspeicherbelegung in WQHD (1440p)
- 8 - Zusammenfassung und Fazit
