DOSP Project 1

Components

project1.zip |- FindCoin.fsx |- Server.fsx |- Client.fsx |- README.md

How to run the code

1. Change the ip address to yours at the following place Server.fsx: line 32 Client.fsx line 22, line 28, line 60

2. Run server "dotnet fsi Server.fsx". Default setting: number of actors: 2 leading 0's: 5 prefix: xiongruoyang IP: 127.0.0.1:9001

3. Run client "dotnet fsi Client.fsx" （You should run server first) Default setting: number of actors: 3 leading 0's: based on the commmand of server prefix: based on the commmand of server IP: 127.0.0.1:2000

4. If you use Mac OS, please make sure that the port number shouldn't be too small, otherwise the binding request of network is likely to be rejected by your system.

5. If you meet any bugs when running the code, please let us know. We can show you a successful running sample of the code.

Size of the work unit that you determined results in the best performance for your implementation and an explanation of how you determined it.

In this project, we use the first character after the prefix as a reserved character. This character is associated with the actor's ID. For example, if the prefix is ​​"tom" and there are 20 actors with IDs 0, 1..19, then we will assign them 0x20 (the first visible character in the ascii table, which is a space), 0x21...0x33 as " The first character after "tom". Since then, the prefixes obtained by the 20 actors are no longer the same "tom", but a four-character string starting with "tom". In this project, theoretically we can assign tasks to 95 (95 represents the number of visible characters in the ascill table) actors. In a real scene, if thousands and millions of actors are needed to work together, we can easily use more bits as reserved characters.

In practice, we use 2-3 actors (the number depends on the OS and hardware). Under this condition, the mining efficiency is the highest, but the CPU usage is more than 250%. Of course, we can also add an appropriate sleep time to the "FindCoin" algorithm to reduce the CPU usage of a single actor mining, and then use multiple actors, so that we can get about 500% CPU usage.

The result of running your program for input 4

Some samples:

xiongruoyang# @M1 00008B4B52C983467F76EBC700052A83D3DB0893421221C7ACC637A8BE294E07 xiongruoyang HVy 0000AA1CAA6A334C4801C9AFD148FF75C5136E466A78F0457D6545AF50BF53F1 xiongruoyang I7{ 00002D6B235BDCD278D668DCAAEF56353213BD01ED302617AB81CA20236D0588 xiongruoyang JLD 00000C10FE92B2F3601111987F8F533936DF9F4D8B60699A3B75AC0A1035D821 xiongruoyang! 0t$ 000062D0F3627BD4A761C5E37C8E3DA12C6D3E2D068102445862DF393BE603FD xiongruoyang" k^# 00002CCF8BFBAA49A7043FEBCDAB21B3F6466DE062CE9E009362136FDE0FF8A5 xiongruoyang! 3]y 000041D18AE1637EC934D74EB5558BF2D0D3A2023D1AB8F9C823777209D557BA xiongruoyang# MH3 0000144F875BFD7001B261CB839ACCC8BBEA8EE9EE9373E55E42DB931F5DBE0A xiongruoyang# NQ0 0000343F73D10D29A6B95895C5B9787DF2982DB559711BC1C78EABD61FD63123 xiongruoyang" tHg 0000810AD598FFA6AC99396C5FF5D832143533E967571A1AC419A41D47FF8F80 xiongruoyang! :`2 0000F7DF6CA3EAD4D3278007B34BC9B4A3405DE34F96300A12BB8A1E36BBA1F3

The running time for the above

The Server is tested on Window system with 4 cpu cores. In this experiment, we create a monitor with 3 workers. we print the real time and cpu time as following:

CPU time = 49640 ms Real time = 19201 ms CPU time/Real time = 2.585315 CPU time = 99046 ms Real time = 40997 ms CPU time/Real time = 2.415954 CPU time = 172671 ms Real time = 71498 ms CPU time/Real time = 2.415059 CPU time = 217031 ms Real time = 90458 ms CPU time/Real time = 2.399249 CPU time = 298640 ms Real time =125232 ms CPU time/Real time = 2.384699 CPU time = 354640 ms Real time =146412 ms CPU time/Real time = 2.422210 CPU time = 391281 ms Real time =162018 ms CPU time/Real time = 2.415048

The coin with the most 0s you managed to find.

The coin with the most 0s we find is the following coin, which has 8 leading 0's.

xiongruoyang! !P]VB 00000000CA1581573DC08B5E1D5E72F4220C3DD464F6E36711A6E0F1AF992DE6 xiongruoyang% #7Kx; 00000000D2D8E516FAE700D57D02C4E7DFDD0AFDBC5EFF57CDBE2153B62E5A2B xiongruoyang$ 0ZYz] 00000000ECF20896E2FE42FA5C1AF364A1070EF121BB9819D8D0105B380A9541

The largest number of working machines you were able to run your code with

We test our code on two machines currently. However, our code can be adapted more machines if necessary.

Assuming that we only deploy one actor on each machine, the current largest number of working machines is 95. This depends on how we split a task into more subtask. In our code, if necessary, we can easily change the number to larger one. In theory, it can accommodate any number of machines.