-Lexman™ Posted February 20, 2021 Posted February 20, 2021 UL Benchmarks is getting us used to launching new small tests to check new functionalities that incorporate the new APIs and compatible graphics cards, so in recent times we have not seen a major change in the main performance test of the suite but we have specific tests such as DirectX Raytracing, Variable Rate Shading, PCI Express bus speed test or even a test for NVIDIA DLSS. Now we see a new free addition to owners of the 3DMark suite that covers another DirectX 12 Ultimate functionality: Mesh Shaders. It should be remembered that DX12 Ultimate has 4 great technologies compared to DX12 which are DX Raytracing, Sampler Feedback, Variable Rate Shading and Mesh Shaders, so the UL Benchmarks suite is already prepared to test almost all of them separately. They actually have tests for 3 of the 4 new technologies and UL Benchmarks promises that they will include the Sampler Feedback test in a few months. That 3DMark incorporates the new Mesh Shaders test is not a surprise either since at the beginning of the week Intel raised the hare showing that its future graphics were capable of executing this precise test and therefore it seems that they will reach the market supporting the latest version of the API from Microsoft for games. The new test is not visually attractive, but it can be very useful for the measurement of this new technology, they comment that it is very short and that two passes are made: one without activating the function and another with Mesh Shaders activated, thus showing the improvement of performance involving the use of this technology. The first summary of provisional performance by architecture is as follows: NVIDIA Ampere (RTX 30): 702% AMD RDNA2: 547% NVIDIA Turing (RTX 20): 409% NVIDIA Turing: 244% In VideoCardz they have compiled the results of several graphic cards that we can see in the following table although they recognize that they have chosen the best result of each one and it does not have to be 100% representative of the real performance, for example there could be overclock. As a curious note we see that the GeForce RTX 20 offer higher performance than the RTX 30 without activating Mesh Shaders, however after activating the technology the increase with the RTX 30 is the most notable of all. We are left wondering what consumers should look for in the test results, if the% improvement or overall performance, since in the results we do not see a big difference between an RTX 2080 Ti and an RTX 3080 or for example the GTX. 1650 Super achieves a% improvement similar to that of an RTX 2080. We also don't know if it would be wise to wait for new versions of the graphics drivers from AMD and NVIDIA to get more reliable results. After the update of this news below we see that it was wise to wait. Unfortunately, the UL Benchmarks YouTube channel seems to be quite abandoned and we could not find the official video that would show us how the test looks on our computer, so we have to resort to what they have published on the Hardwareluxx channel, which seems to be made with an RTX 3080: The official UL Benchmarks website gives us all the information about the new test as well as the requirements such as Windows 10 version 2004 and a graph with drivers that give support for DirectX 12 Ultimate. Updated February 12: After the arrival of the new AMD drivers, many users are reporting that the performance of this test has improved a lot with the Radeon RX 6000, we have examples in our forum and VideoCardz shows us an example of this:
Premium
Recommended Posts