• 목록
• 답변
• 글쓰기
• 게시판 리스트 옵션
  • 수정
  • 삭제

", introduced the basics of CUDA programming by exhibiting how to put in writing a easy program that allotted two arrays of numbers in memory accessible to the GPU and then added them together on the GPU. To do that, I introduced you to Unified Memory, which makes it very easy to allocate and access knowledge that may be used by code running on any processor in the system, CPU or GPU. I finished that post with a number of simple "exercises", considered one of which encouraged you to run on a latest Pascal-based mostly GPU to see what happens. I was hoping that readers would attempt it and touch upon the results, and a few of you probably did! I advised this for 2 causes. First, as a result of Pascal GPUs such as the NVIDIA Titan X and the NVIDIA Tesla P100 are the first GPUs to include the Page Migration Engine, which is hardware help for Unified Memory web page faulting and migration.

The second motive is that it gives an ideal alternative to study more about Unified Memory. Quick GPU, Fast Reminiscence… Right! However let’s see. First, I’ll reprint the results of running on two NVIDIA Kepler GPUs (one in my laptop computer and one in a server). Now let’s attempt operating on a very quick Tesla P100 accelerator, based mostly on the Pascal GP100 GPU. Hmmmm, that’s beneath 6 GB/s: slower than operating on my laptop’s Kepler-primarily based GeForce GPU. Don’t be discouraged, though; we will repair this. To know how, I’ll must let you know a bit more about Unified Memory. What's Unified Memory? Unified Memory is a single memory tackle area accessible from any processor in a system (see Figure 1). This hardware/software program technology allows applications to allocate knowledge that can be read or written from code operating on either CPUs or GPUs. Allocating Unified Memory is so simple as replacing calls to malloc() or new with calls to cudaMallocManaged(), an allocation perform that returns a pointer accessible from any processor (ptr in the next).

When code running on a CPU or GPU accesses knowledge allotted this way (usually called CUDA managed knowledge), the CUDA system software program and/or the hardware takes care of migrating memory pages to the memory of the accessing processor. The necessary level here is that the Pascal GPU structure is the primary with hardware help for virtual memory web page faulting and migration, via its Web page Migration Engine. Older GPUs based mostly on the Kepler and Maxwell architectures also support a more restricted form of Unified Memory. What Occurs on Kepler When i call cudaMallocManaged()? On methods with pre-Pascal GPUs like the Tesla K80, calling cudaMallocManaged() allocates dimension bytes of managed memory on the GPU system that is lively when the call is made1. Internally, the driver also units up page table entries for all pages lined by the allocation, in order that the system knows that the pages are resident on that GPU. So, in our example, running on a Tesla K80 GPU (Kepler structure), x and y are each initially totally resident in GPU Memory Wave.

Then within the loop starting on line 6, the CPU steps by way of each arrays, initializing their parts to 1.0f and 2.0f, respectively. Because the pages are initially resident in machine memory, a page fault happens on the CPU for every array web page to which it writes, and the GPU driver migrates the web page from device memory to CPU memory. After the loop, all pages of the 2 arrays are resident in CPU memory. After initializing the information on the CPU, the program launches the add() kernel so as to add the elements of x to the weather of y. On pre-Pascal GPUs, upon launching a kernel, the CUDA runtime should migrate all pages previously migrated to host memory or to another GPU back to the device memory of the system operating the kernel2. Since these older GPUs can’t page fault, all information have to be resident on the GPU simply in case the kernel accesses it (even when it won’t).