Our next test is GeekBench 5 OpenCL and Vulkan Compute Benchmark. This is the only test in our lineup here that can also test Vulkan, which is unique. The result is an overall score.
In this first graph, we are comparing OpenCL Compute. Once again we do see a large improvement from the Radeon RX 5700 XT. The Radeon RX 6700 XT is a whopping 242% faster than the Radeon RX 5700 XT. However, the GeForce RTX 3060 Ti and GeForce RTX 3070 are both faster still. The RTX 3060 Ti is 23% faster than the RX 6700 XT. The RTX 3070 is 37% faster.
In this graph above we are looking at Vulkan Compute performance. The results show that the Radeon RX 6700 XT isn’t much faster than the Radeon RX 5700 XT in this API. It’s only 2% faster, so that’s an interesting result. The GeForce RTX 3060 Ti is much faster at 64% over the RX 6700 XT. The GeForce RTX 3070 is 112% faster than the Radeon RX 6700 XT.
These Vulkan results are interesting because, in the past, AMD had a leg up on Vulkan performance given its evolution from Mantel. However, today, with the current generation, when it comes to Vulkan Compute performance, even the RTX 3060 Ti has a big leg up over the Radeon RX 6700 XT.
Aida64 GPGPU Benchmark
There are a few noteworthy differences here. Memory Copy performance is a lot faster, almost double, on the Radeon RX 6700 XT versus the other video cards. This could be the signs of the AMD Infinity Cache doing its job.
Looking at the Single-Precision FLOPS performance, the GeForce RTX video cards are much faster. However, the tables turn in the Double Precision FLOPS results, the Radeon RX video cards are faster, with the 6700 XT being the fastest.
Looking at the Integer IOPS it seems the Radeon RX 6700 XT is good at 24-bit and 32-bit. However, once it hits 64-bit it falls way behind, and RTX 3070 is much faster.
AES-256 seems to be much stronger on the GeForce RTX cards as does SHA-1 Hash. Single-Precision Julia is faster on the RTX 3070, but Double-Precision Mandel is faster on the Radeon RX 6700 XT.