NVIDIA's Open64 paper and an analytical model of GPU.

[Zhenyu Ye]

Hi, Gert-Jan and Cedric,

I attached a paper describing an analytical model for GPUs:

Sunpyo Hong, Hyesoon Kim, "An Analytical Model for a GPU Architecture
with Memory-level and Thread-level Parallelism Awareness", to appear in
Proceedings of the 36th International Symposium on Computer Architecture
(ISCA), Austin, TX, June 2009.

"NVIDIA's Experience with Open64":
http://www.capsl.udel.edu/conferences/open64/2008/

Hope this information is useful.

Best regards,

Zhenyu