CPU Performing better than GPU for Yolo Predictions

[iravikiran]

Hello Team,

I'm having a RISC-V Dev platform which has a IMG-GPU, and I'm able to successfully build the Vulkan-NCNN Framework and performing Yolo object detection what we noticed is We're observing better CPU is taking around ~3.0 seconds for performing Yolo v8 objection detection, while GPU is taking ~6.2 seconds for performing the same object detection.

Can you please help me on this, If there's a way to optimise or improvise the performance rates on GPU over CPU. Also it'll be highly appreciated if you can help me with sharing more details on this issue.

---

我有一个带有 IMG-GPU 的 RISC-V 开发平台，我能够成功构建 Vulkan-NCNN 框架并执行 Yolo 对象检测，我们注意到我们观察到更好的 CPU 执行 Yolo v8 对象检测大约需要约 3.0 秒，而 GPU 执行相同的对象检测大约需要约 6.2 秒。

你能帮我解决这个问题吗，如果有办法优化或提高 GPU 而不是 CPU 的性能。如果您能帮助我分享有关此问题的更多详细信息，我将不胜感激。

Regards, Ravi Kiran

[nihui]

The low performance of IMG-GPU is a known IMG driver optimization issue. I tried to measure the peak performance of IMG-GPU in vkpeak but was unable to do so. This needs to be optimized by the driver.