AMD released some information about the upcoming AMD’s next generation APU codenamed ‘Kaveri’ which is based on the Steamroller architecture which will roll out in the coming months. The APU will not only the first to use a fully shared memory, but will also be using a new uniform memory architecture design: heterogeneous uniform memory access- hUMA.
In the current gen APUs, they use the system memory with CPU cores and the IGP, but each comes with their individual memory pool, which is basically termed as a ‘headache’ for developers as the CPU needs to jump through various points before accessing the memory which is also being used by the graphics component on the architecture, therefore the data is copied to and from the CPU and IGP pools whenever the data needs to be shared.
hUMA on the other hand uses both physical and virtual memory along with the shared address space. It also allowed the discrete graphic cards to access GPU or APU memory space, followed by higher performance throughput with lower power utilization during this process.
hUMA is said to have great benefits for 3D rendering performance as it allows partial texture pages to load instead of the entire large texture. hUMA will also be having the following key features:
- Bi-Directional Coherent Memory: This means that any updates made by one processing element will be seen by all other processing elements such as the CPU or GPU.
- Pageable Memory: This allows the GPU to handle page faults the same way that the CPU does. This also removes restricted page locked memory.
- Entire Memory Space: Both the CPU and GPU processes can dynamically allocate memory from the entire memory space. This means it has access to no only the physical memory, but it also has access to the entire virtual memory address space.