Compute Performance
There is another aspect to GPU performance, beyond gaming, and that is general-purpose compute performance or content creation workloads utilizing the GPU. Just because these GPUs are integrated doesn’t mean they don’t support features like OpenCL for compute workloads, or hardware encoding engines for video encoding and transcoding. Therefore we wanted to test both, and compare them and see which one is actually better at other things than just gaming.
HandBrake Video Encoding
Let’s start with video encoding performance, which is something a lot more people are doing with their computers these days. If you are outputting video from editing software utilizing your GPU to accelerate the process can be an important part of speeding up the process, which can take a long time on the CPU alone. In this test, we have taken a 10-Minute video that we recorded at 4K in H.264. We have pulled it into HandBrake and are encoding it using the Fast 1080p30 preset utilizing the GPU video encoder in H.264. We are looking for the quickest time to encode, the least amount of time is the winner.
The clear winner is the Intel Core i9-12900K with its Intel UHD Graphics 770 using Intel QuickSync (QSV). The UHD 770 uses the latest Version 8 of Intel QuickSync, and this is a huge advantage toward the UHD 770 for video encoding. Intel QuickSync is the clear leader here in performance. It encoded our video in 6 minutes and 51 seconds, while the Ryzen 7 5700G Vega 8 AMD encoder took 10 minutes and 15 seconds. The Intel UHD 770 with Intel QuickSync was 36% faster at encoding this video and shaved off three minutes from the encode time.
Geekbench 5 Compute Benchmark
Next up is GeekBench 5’s Compute Benchmark which can test OpenCL and Vulkan performance and spits out an overall benchmark result.
Testing OpenCL performance the AMD Ryzen 7 5700G with Vega 8 is a whopping 92% faster in OpenCL performance compared to the 12900K with Intel UHD 770. We don’t however know what version of OpenCL is being used and tested, the UHD 770 supports OpenCL 3.0 while the Vega 8 does not, and that could change the results.
Testing Vulkan performance the 5700G with Vega 8 is 126% faster than the 12900K with UHD 770. We did observe the 5700G using a newer version of Vulkan.
LuxMark
LuxMark is a popular OpenCL benchmark that can test performance with Ray Tracing and Path Tracing. We are using the latest LuxMark 4.0 version, which includes path trace testing. The Hall Bench utilizes Path Tracing plus a Global Illumination cache. The Food scene utilizes brute force Path Tracing. We also observed that LuxMark recognized the OpenCL 3.0 support on the Intel UHD Graphics 770, and may be using OpenCL 3.0 with it.
The tables have turned in LuxMark, the Intel Core i9-12900K with Intel UHD 770 is now faster than the 5700G with Vega 8, here it is 24% faster. Hall Bench tests path tracing with global illumination, but we noticed that Intel UHD 770 was using OpenCL 3.0 versus 2.1 on Vega 8.
The same is true in the Food benchmark, where Intel UHD 770 is 50% faster than the Vega 8 here. Food utilizes a brute force path tracing. Does it seem OpenCL 3.0 is resulting in an advantage for the Intel UHD 770?
AIDA64 GPGPU Benchmark
Lastly, we have AIDA64 GPGPU Benchmark. This benchmark tests several areas of the GPU for compute performance, including single-precision, double-precision, integer performance, AES, and SHA performance.
You can compare the data side-by-side above. From looking at the performance above it seems the single-precision performance is much greater with Vega 8, and double-precision is not supported on Intel UHD 770? In addition, 24-bit and 32-bit and 64-bit Integer performance is also much greater on Vega 8. We also find AES-256 and SHA-1 Hash to be much faster on Vega 8. Just all around Vega 8 is the winner on compute performance.