Must-read list on GPU architecture


I’d like to read some good (very precise and accurate) articles on GPU architecture in general in order to achieve better performance through better understanding on what is really going on inside. It would be also great if there were listed some differences between CPU and GPU.
Of course, I can find some links but I want an advise from someone who has experience in GPGPU.

Could you give me some links?

Any help would be greatly appreciated.