German Aerospace Center (DLR) Institute of Software Methods for Product Virtualization VECTORIZING A CFD CODE WITH STD::SIMD SUPPLEMENTED BY (ALMOST) TRANSPARENT LOADING AND STORINGMotivation: The Origin simd_value(y[i]); }▪ Requirement to fundamentally re-engineer CFD software for extreme-scale parallel computing platforms was reported on a world-wide basis (cf. CFD Vision 2030): ▪ This necessity was also observed and Airbus ▪ ONERA, DLR and Airbus decided in 2017 to jointly develop the new next generation CFD code CODA with a focus on: ▪ modern algorithms ▪ exploiting current and future HPC hardware architectures

365c1cfd14de5b0e3b85e0fc815b0060f36ab955 3.2.0 31d 00-worker 365c1cfd14de5b0e3b85e0fc815b0060f36ab955 3 365c1cfd14de5b0e3b85e0fc815b0060f36ab955 3.2.0 31d 01-master-kubelet 365c1cfd14de5b0e3b85e0fc815b0060f36ab955 3 365c1cfd14de5b0e3b85e0fc815b0060f36ab955 3.2.0 31d 01-worker-kubelet 365c1cfd14de5b0e3b85e0fc815b0060f36ab955 3

https://github.com/parallel101/course 本课涵盖：稀疏矩阵、 unordered_map 、空间稀 疏网格、位运算、浮点的二进制格式、内存带宽优 化 面向人群：图形学、 CFD 仿真、深度学习编程人 员 第 0 章：稀疏矩阵 稠密数组存储矩阵 用 foreach 包装一下枚举的过程 改用 map 来存储 分离 read/write/create 三种访问模式 foreach 如果这里的 x 是负数，则 x % B 也是负数，会造成对 m_block 的越界访问。 • 因此 % 会返回负数对 CFD 用户来说是个很大的坑点，很多人想当然地用 % 做循环边界， 然而这对负方向会不起作用。 解决： (a % b + b) % b • 我看一些 CFD 用户喜欢写 (a + b) % b 做循环边界 ，从而避免负方向上出错。然而这还是避免不了 a < -b 时的出错。 而是哪怕搞另一个名字，比如 std::bit_vector 之类的都好 。 • 包括他的 data() 也是不对的。 第 6 章：量化浮点类型 使用 double ：每个占据 8 字节 • 很多 CFD 玩家喜欢用 double 表示浮点 数。 • 然而 double 是双精度浮点数，会占据 8 字节！虽然精度更高，但是在不需要精度 的图形学应用中，就非常浪费内存带宽。 使用 float

version=0.1 \ -d url=https://github.com/conan-io/libhello/archive/refs/tags/0.0.1.zip \ -d sha256=1dfb66cfd1e2fb7640c88cc4798fe25853a51b628ed9372ffc0ca285fe5be16b $ cd .. The conan new local_recipes_index command from source Requirements zlib/1.3.1#e20364c96c45455608a72543f3a53133:b647c43bfefae3f830561ca202b6cfd935b56205␣ ˓→- Build ======== Installing packages ======== zlib/1.3.1: Calling source() in /root/ -------- zlib/1.3.1: Building from source zlib/1.3.1: Package zlib/1.3.1:b647c43bfefae3f830561ca202b6cfd935b56205 zlib/1.3.1: Copying sources to build folder zlib/1.3.1: Building your package in /root/.c

version=0.1 \ -d url=https://github.com/conan-io/libhello/archive/refs/tags/0.0.1.zip \ -d sha256=1dfb66cfd1e2fb7640c88cc4798fe25853a51b628ed9372ffc0ca285fe5be16b $ cd .. The conan new local_recipes_index command from source Requirements zlib/1.3.1#e20364c96c45455608a72543f3a53133:b647c43bfefae3f830561ca202b6cfd935b56205␣ ˓→- Build ======== Installing packages ======== zlib/1.3.1: Calling source() in /root/ -------- zlib/1.3.1: Building from source zlib/1.3.1: Package zlib/1.3.1:b647c43bfefae3f830561ca202b6cfd935b56205 zlib/1.3.1: Copying sources to build folder zlib/1.3.1: Building your package in /root/.c

version=0.1 \ -d url=https://github.com/conan-io/libhello/archive/refs/tags/0.0.1.zip \ -d sha256=1dfb66cfd1e2fb7640c88cc4798fe25853a51b628ed9372ffc0ca285fe5be16b $ cd .. The conan new local_recipes_index command from source Requirements zlib/1.3.1#e20364c96c45455608a72543f3a53133:b647c43bfefae3f830561ca202b6cfd935b56205␣ ˓→- Build ======== Installing packages ======== zlib/1.3.1: Calling source() in /root/ -------- zlib/1.3.1: Building from source zlib/1.3.1: Package zlib/1.3.1:b647c43bfefae3f830561ca202b6cfd935b56205 zlib/1.3.1: Copying sources to build folder zlib/1.3.1: Building your package in /root/.c

version=0.1 \ -d url=https://github.com/conan-io/libhello/archive/refs/tags/0.0.1.zip \ -d sha256=1dfb66cfd1e2fb7640c88cc4798fe25853a51b628ed9372ffc0ca285fe5be16b $ cd .. The conan new local_recipes_index command from source Requirements zlib/1.3.1#e20364c96c45455608a72543f3a53133:b647c43bfefae3f830561ca202b6cfd935b56205␣ ˓→- Build ======== Installing packages ======== zlib/1.3.1: Calling source() in /root/ -------- zlib/1.3.1: Building from source zlib/1.3.1: Package zlib/1.3.1:b647c43bfefae3f830561ca202b6cfd935b56205 zlib/1.3.1: Copying sources to build folder zlib/1.3.1: Building your package in /root/.c

2019 17:32:18 GMT location: /images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg server: uvicorn { "id": "5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c", "image": "/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg" GMT", "size": [ 462, 462 ] } Next, try retrieving the uploaded image: $ http localhost:8000/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg HTTP/1.1 200 OK content-type: image/jpeg date: Tue, 24 Dec 2019 application/json date: Tue, 24 Dec 2019 17:36:31 GMT server: uvicorn [ { "id": "5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c", "image": "/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg", "modified": "Tue, 24 Dec 2019 17:32:19

2019 17:32:18 GMT location: /images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg server: uvicorn { "id": "5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c", "image": "/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg" GMT", "size": [ 462, 462 ] } Next, try retrieving the uploaded image: $ http localhost:8000/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg HTTP/1.1 200 OK content-type: image/jpeg date: Tue, 24 Dec 2019 application/json date: Tue, 24 Dec 2019 17:36:31 GMT server: uvicorn [ { "id": "5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c", "image": "/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg", "modified": "Tue, 24 Dec 2019 17:32:19

2019 17:32:18 GMT location: /images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg server: uvicorn { "id": "5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c", "image": "/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg" GMT", "size": [ 462, 462 ] } Next, try retrieving the uploaded image: $ http localhost:8000/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg HTTP/1.1 200 OK content-type: image/jpeg date: Tue, 24 Dec 2019 application/json date: Tue, 24 Dec 2019 17:36:31 GMT server: uvicorn [ { "id": "5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c", "image": "/images/5cfd9fb6-259a-4c72-b8b0-5f4c35edcd3c.jpeg", "modified": "Tue, 24 Dec 2019 17:32:19