Compute Benchmarks

Now I couldn’t forget about compute benchmarks. The games get all of the glory, but compute performance is what helps build the games and edit and render the videos. I started off with Passmark’s Performance Test with their GPU Compute benchmark. This one runs through a series of tests to give an overall score. The RTX 2080 FE didn’t do very well on this one. The 2080 Ti was barely ahead of the 1080 Ti and the 2080 was within a point of the original GTX 1080 FE.

graph30

In the V-Ray benchmark performance was better though. This time the RTX 2080 FE was tied with the GTX 1080 Ti. I should point out that while this is a ray tracing test, it isn’t using the new RT cores, hopefully, V-ray introduces support for those later.

graph31

Next, we have Cinegy Cinescore, this test had some issues. As it turns out both RTX cards had some of the Nvidia specific formats fail. I included a picture below from when it happened on the RTX 2080 Ti. Because of the failed results, the score was a little lower. The Vega 64 was even worse, it had enough failed that the benchmark wouldn’t give a score at all.

graph32

cinescoreproblem

My last test was going to be taking a look at one of the most popular modeling programs with their recently introduced benchmark. But both RTX cards crashed. I spoke with Nvidia about it and right now they are waiting for blender to add in CUDA 10 support. I left the results though for comparison with the other cards. The Vega 64 did extremely well here.

graph33

 

Log in to comment

We have 2022 guests and no members online

supportus