Python image algorithm ported to CUDA C++ with no fixes

Vlad Mihalcea@vlad_mihalcea

Whenever I see people on Reddit or other social media platforms saying that AI cannot write complex code, I wonder what models they are using, as Opus 5.5 is really good.

For example, last week, after I gave my IEEE presentations, a professor from our department asked me to help one of her students with a problem he was facing: porting an image-processing algorithm from a Python library that can only emulate the processing on GPUs. The student wanted to process the images using CUDA from a C++ port of that Python lib. Not a very easy task.

So I told him it would be interesting to see whether Opus 5.5 could do it without any additional training material. My notebook uses an AMD Ryzen 9 CPU with an integrated Radeon 840M chip, so there's no way to test CUDA natively as I'd need an NVIDIA GPU for that. But Opus managed to emulate the lib on the CPU, and it said it worked. I have no idea how it did that.

After 20 minutes, I had the C++ lib in place and a nice step-by-step guide on how to use it, and I gave it to the student.

Two days later, he told me that it was working like a charm without any modifications. Frankly, I didn't think it would work out of the box, but it did.