[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$fjqGjzaZa9nVJTtpgoXnydqwCOikTEU5mks6rdYfCgbA":3},{"answer":4,"createTime":5,"id":6,"options":7,"origin":12,"question":16,"related":17,"source":28,"type":29},[],"2023-12-25 21:43:21",121568610,[8,9,10,11],"①②③","②③④","①②③④","①③④",{"courseId":13,"courseImg":14,"courseName":15},"f3216a940ddb2a48ef903cdf7d98732e","https:\u002F\u002Ftihai-oss-cloud.itihey.com\u002Fimg\u002F7e2d5ad53be3d12aacc56ee37db405b7.jpg","并行处理及分布式系统","研究并行算法的可扩展性的目的是 ①确定某类问题的何种并行算法与何种并行系统的组合,可有效的利用系统大量处理机;②算法在小规模并行机上的运行性能来预测其移植到大规模并行机上后的运行性能;③对一固定规模的问题,能确定其运行在某类并行系统上时,所需的最多处理机台数和获得的最大加速比;④指导并行算法和并行机体系结构的改进.( )",[18,30,39,49,58,67,75,83,92,101],{"answer":19,"createTime":20,"id":21,"options":22,"question":27,"source":28,"type":29},[],"2023-12-25 21:43:17",121568371,[23,24,25,26],"是多源状态","加锁解锁开销低","条件不成立时阻塞线程并自动解锁加锁","以上皆错","条件变量相对于互斥量的优点是_____.( )","v1",0,{"answer":31,"createTime":20,"id":32,"options":33,"question":38,"source":28,"type":29},[],121568388,[34,35,36,37],"其结束时需程序员人为添加同步","其结束时OpenMP已默认添加隐式同步","其结束时完全不需要同步","其结束时必须进行同步","关于OpenMP程序中的并行结构,以下说法正确的是 .( )",{"answer":40,"createTime":41,"id":42,"options":43,"question":48,"source":28,"type":29},[],"2023-12-25 21:43:18",121568397,[44,45,46,47],"可能导致负载不均","通信开销高","任务粒度粗","计算复杂度高","动态任务划分相对于静态任务划分的缺点是____.( )",{"answer":50,"createTime":41,"id":51,"options":52,"question":57,"source":28,"type":29},[],121568404,[53,54,55,56],"外层存在,内层不存在","外层不存在,内层存在","两层都不存在","两层都存在","将冒泡排序程序改写为奇偶交换排序关于其两层循环是否存在数据依赖,下面说法正确的是()",{"answer":59,"createTime":41,"id":60,"options":61,"question":66,"source":28,"type":29},[],121568412,[62,63,64,65],"成本太高","结果不准确","时间太长","太危险","传统科学和工程实验相对于计算机并行仿真的劣势不包括()",{"answer":68,"createTime":41,"id":69,"options":70,"question":74,"source":28,"type":29},[],121568427,[71,72,73,26],"存在","不存在","不确定","for (i=2; i<10; i++) A[i] = A[i-2]+1; 此循环_____数据依赖.( )",{"answer":76,"createTime":41,"id":77,"options":78,"question":82,"source":28,"type":29},[],121568437,[79,80,81,26],"肯定快于","肯定慢于","可能快于","在编写使用n个线程进行n个数求和的OpenMP的程序时采用临界区技术,则多线程版本_____串行版本.( )",{"answer":84,"createTime":41,"id":85,"options":86,"question":91,"source":28,"type":29},[],121568448,[87,88,89,90],"排列","交换","广播","归约","对矩阵乘法串行程序主体三重循环的最内层循环进行向量化,则该循环执行完毕后,还需进行SIMD寄存器中几个元素的 操作才能得到结果矩阵的一个元素.( )",{"answer":93,"createTime":41,"id":94,"options":95,"question":100,"source":28,"type":29},[],121568454,[96,97,98,99],"读取x存在数据依赖,不能并发进行","累加顺序被改变,结果是错误的","加法操作是简单运算,无需同步","加法操作不是原子操作,需要同步保证数据依赖","多个线程进行并行求和,每个线程将自己负责的值依次读入局部变量x,累加到全局变量sum上,sum+=x,对此,下面说法正确的是 .( )",{"answer":102,"createTime":41,"id":103,"options":104,"question":109,"source":28,"type":29},[],121568460,[105,106,107,108],"MPI_Sendrecv","MPI_Send","MPI_Recv","MPI_Sendrecv_Replace","MPI原址收发各一条消息的API是()"]