Synthetic Benchmarks

As always for the first batch of testing, I ran the two cards in Crossfire through a few synthetic benchmarks. These are typically more likely to support multi-card configurations and are great for comparing the difference between cards. The first tests were in 3DMark where I tested using the Fire Strike benchmark at 1080p, 1440p, and 4k then I also used Time Spy to check out DX12 performance. At 1080p the 570’s came in below the 1080 Ti but above the new 11Gbps GTX 1080’s. Turning the resolution up had the pair dropping down some at 1440p and even more at 4k where single cards with more VRAM and higher memory bandwidth have advantages. In Time Spy there is a noticeable different between the RX 570’s in Crossfire compared to the RX 470’s in Crossfire and the 570’s came in just below the 11Gbps GTX 1080.

Unigine Valley had Crossfire support and the cards were up in the GTX 1070 range, about 40 FPS above a single RX 570. In the new Unigine Superposition benchmark, there wasn’t any support and the two cards running together basically matched the single card performance of the Asus RX 570 that was the main card anyhow so at least there wasn’t a performance drop but this is an example of what happens when games aren’t supported.

Catzilla was the same, all three tests came in just below the Strix RX 570 with no Crossfire support and a few point loss.

I also slipped in a little VR testing and there was an improvement in the SteamVR benchmark from 6.5 up to 7.9, both are in the green but it looks like if the game supports Crossfire you will see much smoother VR performance. The same happened in VRMark where we gained 23 FPS on the orange room and about 9 on the always demanding blue room test.