Synthetic Benchmarks
To start off my testing I always like to run through a few synthetic benchmarks. They are consistent and typically very optimized on the driver side so they are great for getting a good comparison from card to card. So the RX 5700 XT is running $399 and the RX 5700 is $349 which puts the two cards right in line with the new RTX 2060 SUPER and the original RTX 2060 which have the same $399 and $349 MSRP’s. There are also a few Vega 64’s that come in at $399 as well. So I’m, really curious to see how those three cards compare to both of the RX 5700’s.
For tests I mostly focused on a few different 3DMark benchmarks, starting with the Fire Strike tests which are all DirectX 11 based and touch on three different detail and resolution settings. In all three tests, the order of the cards is the same with the RX 5700 outperforming the RTX 2070 SUPER and coming in between that and the RTX 2080. The RX 5700, on the other hand, was in between the Vega 64 and the RTX 2060 SUPER.
Then I moved over to the two 3DMark Time Spy tests which are DX12 and a little newer. Nvidia’s Turing based cards have really been doing well in DX12 applications and the RX 5700 XT was well below the 2070 SUPER and actually very close to the 2060 SUPER. The RX 5700 was running with the Vega 64 and the original RTX 2060 here. But when the detail was increased both dropped a little with the RX 5700 still outperforming the RTX 2060 but the RX 5700 XT was almost 100 points behind the RTX 2060 SUPER.
To mix things up I also tested with the Unigine engine based Superposition test. For this one, I tested at 1080p at both the extreme and medium detail settings under DirectX. The RX 5700 XT did really well here with a good margin between it and the RTX 2060 SUPER and the RX 5700 was ahead of its competition as well, but with less of an advantage. The RX 5700 also dropped below the RTX 2060 when the detail went up and the RX 5700 XT was similar barely squeaking ahead of the 2060 SUPER with just one point at the extreme detail setting.
