AWS EC2 p2.xlarge should have 61GiB memory but my code run out of RAM at about 11GB

Question

In the terminal I found some error like these:

I tensorflow/core/common_runtime/bfc_allocator.cc:702] Stats: 
Limit:                 11332668621
InUse:                 11332668416
MaxInUse:              11332668416
NumAllocs:                      13
MaxAllocSize:          11332445440

But the p2.xlarge instance should have 61GiB memory... What is wrong here? Please help!

Answer 1

From New P2 Instance Type

The host memory is 61GB for p2.xlarge instance. The GPU memory is 12GB and you exhausted your GPU memory.