I am not very familiar with GPU programming, and I have just started using ILGPU. When I tried to migrate my existing computational code to ILGPU, I found that the results on CPU and GPU were completely different. I suspect it may be a precision issue. Therefore, I was wondering if there is any configuration in this library that can improve the calculation precision without making too many changes to the code.
Question
I am not very familiar with GPU programming, and I have just started using ILGPU. When I tried to migrate my existing computational code to ILGPU, I found that the results on CPU and GPU were completely different. I suspect it may be a precision issue. Therefore, I was wondering if there is any configuration in this library that can improve the calculation precision without making too many changes to the code.
Environment
No response
Additional context
No response