Compute Benchmarks

Now some people don’t need a video card for gaming, they need the processing power for rendering or 2D/3D production or in some cases people who game also do work on the side. So it is also important to check out the compute performance on all of the video cards that come in. That includes doing a few different tests. My first test was a simple GPU Compute benchmark using Passmark’s Performance Test 9 and the RTX 2060 KO fell right in between the 2060 Founders Edition and the overclocked RX 5600 XT. Leaning closer to the 5600 XT.

graph29

Next, I took a look at performance using Basemark. With this test, I took a look at DirectX12 and OpenGL performance. The AMD cards have fallen behind with the OpenGL results across the board so it wasn’t a shock when the RX 5600 XT performed poorly. But even the DirectX12 result was much lower than both RTX 2060s. The 2060 KO did still have a nice edge over the Founders Edition as well.

graph31

Geekbench 5 is similar, but taking a look at Vulkan and OpenCL performance. Oddly enough Vulkan compute performance on the RX 5600 XT was much lower. Given that Vulkan was AMDs thing, that is interesting. But the OpenCL results being the highest of the two on all of the cards I used those to compare and the 2060 KO came in at the top of the chart.

graph32

My last Compute test is most likely the one everyone is curious about. Both because it is a real world test, not a synthetic benchmark and also because the RTX 2060 KO already has a reputation for this specific test. I actually ran into this result two weeks ago and this was the entire reason for this review being delayed. I reached out to EVGA and Nvidia and at the time no one understood my issue or had heard about it. So I was waiting to hear back on it to see if it was a testing issue on my end. But after seeing our friends over at Gamers Nexus publish a video on the 2060 KO in Blender it confirmed exactly what I was seeing and even more. So the short version is for some reason the 2060 KO with its unique cut down TU104 GPU performs WAY better than it should in Blender and a few other compute tests. Blender is the only one in our test suite that it does this. But our 2060 KO came in at 291 seconds which is a significant jump over the 2060 SUPER that it should be behind and a 100 seconds faster than the RX 5600 XT for example and 80 seconds faster than the stock RTX 2060 FE that runs at the same clock speeds. I still don’t know if this is a weird bug that will get fixed later or what causes it, but both EVGA and Nvidia have been looking at the issue. The end result makes the 2060 KO a nice little mid-range card for Blender users who can’t afford cards like the 2070 and 2080.

graph30

 

Log in to comment

enterthemax's Avatar
enterthemax replied the topic: #38746 20 Mar 2020 12:35
Hello, may I know which CPU have you used for the tests?
garfi3ld's Avatar
garfi3ld replied the topic: #38747 20 Mar 2020 14:07
No problem, it was tested with the AMD Ryzen 3900X. Not sure why that was no longer on the list of components in the test bench list but I updated that as well

We have 672 guests and one member online

supportus