文件列表:

.clang-format (2011, 2023-07-14)
.editorconfig (40, 2023-07-14)
CMakeLists.txt (4533, 2023-07-14)
LICENSE (11357, 2023-07-14)
doc (0, 2023-07-14)
doc\dev_gui_guide.md (370, 2023-07-14)
doc\developer_zh.md (14383, 2023-07-14)
doc\implement.md (2451, 2023-07-14)
doc\import_guide.md (1095, 2023-07-14)
doc\profile_muda_zh.assets (0, 2023-07-14)
doc\profile_muda_zh.assets\profile_result.png (64662, 2023-07-14)
doc\profile_muda_zh.md (1532, 2023-07-14)
doc\tutorial_zh.md (13213, 2023-07-14)
example (0, 2023-07-14)
example\cooperative_groups (0, 2023-07-14)
example\cooperative_groups\async_transfer.cu (1508, 2023-07-14)
example\device_query (0, 2023-07-14)
example\device_query\device_query.cu (14062, 2023-07-14)
example\device_query\log.txt (12710, 2023-07-14)
example\event (0, 2023-07-14)
example\event\event.cu (1277, 2023-07-14)
example\example_common.h (2991, 2023-07-14)
example\graph (0, 2023-07-14)
example\graph\graph_bindless.cu (4829, 2023-07-14)
example\graph\graph_manager.cu (3804, 2023-07-14)
example\graph\graph_quick_start.cu (1991, 2023-07-14)
example\hello_muda (0, 2023-07-14)
example\hello_muda\hello_muda.cu (621, 2023-07-14)
example\hello_muda\muda_vs_cuda.cu (1008, 2023-07-14)
example\hello_muda\vector_add.cu (1853, 2023-07-14)
example\launch (0, 2023-07-14)
example\launch\parallel_for.cu (2473, 2023-07-14)
example\main.cpp (53, 2023-07-14)
example\mk.py (668, 2023-07-14)
example\pba (0, 2023-07-14)
example\pba\sph2d.cu (10841, 2023-07-14)
... ...

# muda MUDA is **-CUDA**, yet another painless CUDA programming **paradigm**. > stop bothering yourself using raw CUDA, just stay elegant and clear-minded. ## overview easy launch: ```c++ #include using namespace muda; int main() { launch(1, 1) .apply( [] __device__() { print("hello muda!\n"); }).wait(); } ``` muda vs cuda: ```c++ /* * muda style */ void muda() { device_vector dv(***, 1); stream s; parallel_for(2, 16, 0, s) // parallel-semantic .apply(***, //automatically cover the range using (gridim=2, blockdim=16) [ // mapping from the device_vector to a proper viewer // which can be trivially copy through device and host dv = make_viewer(dv) ] __device__(int i) mutable { dv(i) *= 2; // safe, the viewer check the boundary automatically }) .wait();// happy waiting, muda remember the stream. //.apply(...) //if you want to go forward with the same config, just call .apply() again. } /* * cuda style */ // manually create kernel __global__ void times2(int* i, int N) // modifying parameters is horrible { auto tid = threadIdx.x; if(tid < N) // check corner case manaully { i[tid] *= 2;// unsafe: no boundary check at all } } void muda_vs_cuda() { // to be brief, we just use thrust to allocate memory thrust::device_vector dv(***, 1); // cast to raw pointer auto dvptr = thrust::raw_pointer_cast(dv.data()); // create stream and check error cudaStream_t s; checkCudaErrors(cudaStreamCreate(&s)); // call the kernel (which always ruins the Intellisense, if you use VS.) times2<<<1, ***, 0, s>>>(dvptr, dv.size()); // boring waiting and error checking checkCudaErrors(cudaStreamSynchronize(s)); } ``` ## quick start run example: ```shell $ xmake f --example=true $ xmake $ xmake run muda_example hello_muda ``` to show all examples: ```shell $ xmake run muda_example -l ``` play all examples: ```shell $ xmake run muda_example ``` ## tutorial - [tutorial_zh](https://github.com/MuGdxy/muda/blob/master/./doc/tutorial_zh.md) ## features ### thread_only container allow you to use STL-like containers and algorithms in one thread. ```cpp #include #include using namespace muda; namespace to = muda::thread_only; int main() { launch(1, 1) .apply( [] __device__() mutable { to::priority_queue queue; auto& container = queue.get_container(); container.reserve(16); queue.push(4);queue.push(5);queue.push(6); queue.push(7);queue.push(8);queue.push(9); while(!queue.empty()) { print("%d ", queue.top()); queue.pop(); } //result: 9 8 7 6 5 4 }) .wait(); } ``` ### graph support a friendly way to create cuda graph. ```cpp void graph_example() { device_var value = 1; // create a graph auto graph = graph::create(); // create kernel as graph node parameters(we switch from .apply to .asNodeParms) auto pA = parallel_for(1).asNodeParms(1, [] __device__(int i) mutable { print("A\n"); }); auto pB = parallel_for(1).asNodeParms(1, [] __device__(int i) mutable { print("B\n"); }); auto phB = host_call().asNodeParms( [] __host__() { std::cout << "host" << std::endl; }); auto pC = parallel_for(1).asNodeParms(1, [value = make_viewer(value)] __device__(int i) mutable { print("C, value=%d\n", value); value = 2; }); auto pD = launch(1, 1).asNodeParms( [value = make_viewer(value)] __device__() mutable { print("D, value=%d\n", value); }); // create nodes and dependencies auto kA = graph->addKernelNode(pA); auto kB = graph->addKernelNode(pB); // hB deps on kA and kB auto hB = graph->addHostNode(phB, {kA, kB}); // kC deps on hB auto kC = graph->addKernelNode(pC, {hB}); // kD deps on kC auto kD = graph->addKernelNode(pD, {kC}); // create instance (or say graphExec) auto instance = graph->instantiate(); //launch on the default stream instance->launch(); } ``` ## Contribute Go to [developer_zh.md](https://github.com/MuGdxy/muda/blob/master/./doc/developer_zh.md) and [zhihu-ZH](https://github.com/MuGdxy/muda/blob/master/https://zhuanlan.zhihu.com/p/592439225) for further info. ## More Info [zhihu-ZH](https://github.com/MuGdxy/muda/blob/master/https://zhuanlan.zhihu.com/p/592439225) : description of muda.

近期下载者：

相关文件：

评论：[我要评论] [举报此文件]

收藏者：