allocator initial cpu memory usage is too large
Created by: tensor-tang
This is an issue related with #11185 (closed)
Even we can use fraction_of_cpu_memory_to_use to work around, but it's still tricky and does not make sense that init memory with max_chunk_size_ which it 3% memory.
BuddyAllocator::PoolSet::iterator BuddyAllocator::RefillPool() :
https://github.com/PaddlePaddle/Paddle/blob/6d80dd5a50f32db7c4353e52334bca7e8179d785/paddle/fluid/memory/detail/buddy_allocator.cc#L189-L192
void* BuddyAllocator::Alloc(size_t unaligned_size) :
https://github.com/PaddlePaddle/Paddle/blob/6d80dd5a50f32db7c4353e52334bca7e8179d785/paddle/fluid/memory/detail/buddy_allocator.cc#L67-L69