AMD Radeon RX 5700 faster than NVIDIA RTX 2070 Super in OpenCL

    2
    2002
    AMD Radeon

    We’ve been seeing a significant number of benchmarks related to the AMD Navi GPUs as well as the NVIDIA GeForce RTX Super cards, but the data is still far from enough to come to a conclusive result. Continuing his benchmark spree, serial leaker TUM_APISAK has now shared the OpenCL score of the Navi based Radeon RX 5700 which surprisingly appears to be faster than the NVIDIA GeForce RTX 2070 Super. A bit hard to buy, innit? Let’s dig in:

    And there you have it, although the delta isn’t that wide, it’s still rather glaring considering that the RX 5700 will cost $379 while the NVIDIA RTX 2070 Super will retail for $499. However, on a closer look, it becomes clear that these differences won’t be there in real-world applications, including gaming. The Navi card is mainly faster in the particle physics test (by a lot), while in the rest of the tests the 2070 Super comes out on top. So normally, I’d say that AMD’s cards are probably optimized much better for particle physics, but that’s not really the case.

    Note the platforms on which these two tests were run. Yes, they are different. The Turing card leverages NVIDIA’s popular CUDA framework while the AMD card runs on the company’s Accelerated Parallel Processing platform. So, I can say with a good deal of certainty that this performance deficit is due to the use of different SDKs. If both were using the standard OpenCL platform, the RTX 2070 Super would no doubt have been faster by a notable margin. Take from this what you will, but it’s *NOT* representative of real-world performance. Cheers!

    Further Reading:

    Source

    2 COMMENTS

    1. Open CL awesome!

      Wait. Who uses OpenCL?

      Windows – Direct X

      Apple – Metal

      ????? – Open CL

      ????? = Linux

      So this is only great news for Linux users . . .

    Leave a Reply