2018 Huawei Software Elite Challenge - Rematch Questions

The following descriptions are mainly the changes from the preliminary questions, and other descriptions and conditions are the same:
Generic description change points:
Physical server: In order to meet the requirements of different virtual machine specifications, there are also various physical server specifications. It is assumed that the cloud platform has three types of physical servers: general, high-performance, and memory. For Large-Memory, each type of physical server has different CPU counts and memory sizes.
Resource dimension: In order to balance the use of various resources, it is necessary to consider the simultaneous optimization of the two resource dimensions of CPU and memory. It is assumed that the priority of each dimension resource is equivalent, that is, the weight of CPU and memory is 0.5 each.
Supplementary explanation of changes:
1. The fragmentation rate of CPU and memory needs to be considered at the same time, that is, the final resource utilization score is the weighted score of the two;
2. The time span to be predicted is 1 to 4 weeks, and the start time to be predicted and the end time of the training data set are not necessarily continuous, and the prediction will start after 0 to 15 days;
3. The maximum number of training data sets for each test case in the semi-finals is 20,000;
4. The running time of the program does not exceed 90s (single use case);
5. Need to predict the type and quantity of virtual machine specifications to expand to the following 18 types:
flavor1  1  1024
flavor2  1  2048
flavor3  1  4096
flavor4  2  2048
flavor5  2  4096
flavor6  2  8192
flavor7  4  4096
flavor8  4  8192
flavor9  4  16384
flavor10  8  8192
flavor11  8  16384
flavor12  8  32768
flavor13  16  16384
flavor14  16  32768
flavor15  16  65536
flavor16  32  32768
flavor17  32  65536
flavor18  32  131072
Remarks: flavor name CPU core memory size (MB)
Changes in the winning and losing rules of the game:
Compare the product of the prediction accuracy output by the participating teams' programs and the resource utilization rate (the utilization rate is the weight of CPU resources and memory resources), and the larger one wins. If the scores are the same, the running times of the programs are compared, and the one with the shorter time wins. If the running time is also the same, the ranking will be distinguished according to the time of submission. If the output result does not meet the constraints, the score is zero.
Program input and output change points:
input file format
Program input is a space-delimited text file, each line of the file ends with a newline character ('\r\n').
The file format is:
Number of physical server types (default 3)
Type name 1 Number of CPU cores of physical server Memory size (GB) Hard disk size (GB)
Type name 2 Number of physical server CPU cores Memory size (GB) Hard disk size (GB)
Type name 3 Number of CPU cores of physical server Memory size (GB) Hard disk size (GB)
(Blank line)
Number of virtual machine specifications
Virtual machine specification name 1 Number of CPU cores Memory size (MB)
Virtual machine specification name 2 CPU cores Memory size (MB)
. . . (Several lines of virtual machine specification information above)
(Blank line)
Forecast start time
Forecast end time (time span unit: days)
(end of file)
示例:
3
General  56  128  1200
Large-Memory  84  256  2400
High-Performance  112  192  3600
(备注:物理服务器类型名称不超过16个字符,CPU核数及内存大小均为不超过3位的整数,硬盘大小为不超过4位的整数。1GB=1024MB)
3
flavor5  2  4096
flavor10  8  8192
flavor15  16  65536
(备注:虚拟机规格名称最大不超过10位字符,CPU核数最大不超过2位字符,内存大小最大不超过6位字符。)
2017-01-09 00:00:00
2017-01-16 00:00:00 //注:即7天的时间跨度
(备注:日期与具体时间之间为空格符。需要预测的时间跨度为1~4个星期,且要预测的开始时间与训练数据集的结束时间不一定是连续的,会从0~15天后开始预测。)
(文件结束)
输出文件格式
程序输出为一个以空格分隔的文本文件,文件每行以换行符(’\n’)为结尾。
文件格式为:
预测的虚拟机总数
虚拟机规格名称1  虚拟机个数
虚拟机规格名称2  虚拟机个数
。。。(如上预测的不同规格虚拟机名称及数量若干行)
(空行)
物理服务器类型名称1  预测所需数量
名称1-服务器1  虚拟机规格名称1  能放置该类型虚拟机个数  虚拟机规格名称2  能放置该类型虚拟机个数 ……
名称1-服务器2  虚拟机规格名称1  能放置该类型虚拟机个数  虚拟机规格名称2  能放置该类型虚拟机个数 ……
……(如上每种类型物理服务器对应放置每种虚拟机规格数量若干行)
(空行)
物理服务器类型名称2  预测所需数量
名称2-服务器  虚拟机规格名称1  能放置该类型虚拟机个数  虚拟机规格名称2  能放置该类型虚拟机个数 ……
名称2-服务器2  虚拟机规格名称1  能放置该类型虚拟机个数  虚拟机规格名称2  能放置该类型虚拟机个数 ……
……(如上每种类型物理服务器的分别放置每种虚拟机规格数量若干行)
(空行)
物理服务器类型名称3  预测所需数量
名称3-服务器1  虚拟机规格名称  能放置该类型虚拟机个数  虚拟机规格名称2  能放置该类型虚拟机个数 ……
名称3-服务器2  虚拟机规格名称  能放置该类型虚拟机个数  虚拟机规格名称2  能放置该类型虚拟机个数 ……
……(如上每种类型物理服务器的分别放置每种虚拟机规格数量若干行)
(文件结束)
(备注:物理服务器名称可用“类型名称-阿拉伯数字”表示,如General-1)
示例:
6
flavor5  3
flavor10  2
flavor15  1
(备注:如果某种虚拟机规格的预测结果为零,即对应写0)
General  2
General-1  flavor5  2
General-2  flavor5  1  flavor10  1
Large-Memory  1
Large-Memory-1  flavor10  1
High-Performance  1
High-Performance-1  flavor15  1
(备注:每个字段之间用空格隔开即可。如果输出结果中某种类型的物理服务器所需个数为0,则不需要写出来,例如“General 0”这种情况可以直接省略不写。)
单个用例评分公式变化点:
:表示第i种虚拟机规格的实际数量;
:表示预测出来的第i种虚拟机规格数量;
:表示虚拟机规格的集合;
:表示预测到的第v个虚拟机的CPU资源大小;
:表示预测到的第v个虚拟机的内存资源大小;
:表示所放置的第h个物理主机的CPU资源容量;
:表示所放置的第h个物理主机的内存资源容量;
:表示预测出来的虚拟机集合;
:表示所需要的物理主机集合。
(备注:单个用例满分为100分。)
运行环境
开发语言支持:C/C++、Java 7/8、Python 2.7
CPU:Intel(R) Xeon(R) CPU E5-2680 V4 @ 2.40GHz
内存:2G
CPU核数:单核
编译器:gcc 4.8.4、java 1.8、python 2.7
操作系统:Ubuntu 14.04.4 LTS 64位,内核版本 Linux version 4.4.0-31-generic
SDK:为方便选手做题,分别提供c++(兼容c)、Java、Python的SDK包供参考(见DevCloud项目),详细描述信息请见SDK目录下的readme.txt。

Guess you like

Origin http://43.154.161.224:23101/article/api/json?id=325936957&siteId=291194637