What you'll notice is that Hawaii/Grenada and Polaris perform dramatically. Mind you that we cannot select OpenCL on the GeForce cards, ergo CUDA is selected. OpenCL is applied for Radeon and CUDA for GeForce. Here we can see twenty cards rendering the classroom scene. To address that we'll compare brand compatible API to the graphics cards tested, first Radeon OpenCL versus GeForce CUDA, let's re-visit and plot a chart:Īl GeForce cards of recent years can utilize the CUDA (closed API), which is fairly close to OpenCL (open API). So yes, albeit a valid chart, it will be too confusing. With three APIs for which some work on some cards but not on others, that's going to be a bit of a challenge to chart up, as literally you'd be getting a cluttered confusing chart like this: Score in second to render / lower is fasterĪs you can see the Optix code path is greatly faster if you have a compatible graphics card (RTX).
To show you the performance benefits between CUDA and the OptiX code path, well here is an example of an RTX 2060 Super in CUDA and Optix: CUDA is only proprietary towards Nvidia's GPU products. The computations are offloaded to the GPUs through either the low-level or the high-level API introduced with CUDA. It will prove to be (by far) the fastest renderer for compatible GeForce graphics cards. OptiX Application Acceleration Engine is a ray tracing API. OptiX only works on RTX cards that said however, not Turing architecture-based cards like the GTX 1660 series. That OptiX option you need to keep an eye on. NVIDIA GeForce RTX cards based on Turing can be assigned CUDA or OptiX - but not OpenCL.NVIDIA GeForce cards up to Pascal support CUDA - but not OpenCL or Optix.There are API related challenges to address with Blender though: We fire off a scene where we render a Classroom - we only allow the GPU to render in the benchmark application. Blender v2.81a has recently been updated towards build 2.82 and offers a wide variety of options and.
0 kommentar(er)
