I see, you present the illusion of having a contiguous block of free space by remapping fragmented space using fieldPtr. I suppose you don’t care about the amortized cost of delete because they’re so few.
Related topics
Topic | Replies | Views | Activity | |
---|---|---|---|---|
Data structures on the GPU (heap) | 4 | 4265 | April 3, 2012 | |
Dynamic Heap initialization | 12 | 412 | June 24, 2024 | |
Build Tree on Device How to create a tree parallely on CUDA | 1 | 900 | January 30, 2012 | |
Fermi memory management different? | 10 | 1330 | April 17, 2011 | |
Dynamic memory allocation inside the kernel | 2 | 17682 | March 31, 2015 | |
How to allocate dynamic memory on device? | 3 | 1651 | June 7, 2010 | |
Reallocating memory on CUDA? | 2 | 4831 | January 17, 2011 | |
dynamic allocation using new/delete with existing CUDA 3.0 | 0 | 1519 | April 24, 2010 | |
Question Dynamic Memory Allocation in the kernel function | 2 | 3641 | November 30, 2009 | |
Shared memory - dynamic allocation | 3 | 1242 | November 28, 2017 |