莊雲凱
    • Create new note
    • Create a note from template
      • Sharing URL Link copied
      • /edit
      • View mode
        • Edit mode
        • View mode
        • Book mode
        • Slide mode
        Edit mode View mode Book mode Slide mode
      • Customize slides
      • Note Permission
      • Read
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Write
        • Only me
        • Signed-in users
        • Everyone
        Only me Signed-in users Everyone
      • Engagement control Commenting, Suggest edit, Emoji Reply
    • Invite by email
      Invitee

      This note has no invitees

    • Publish Note

      Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

      Your note will be visible on your profile and discoverable by anyone.
      Your note is now live.
      This note is visible on your profile and discoverable online.
      Everyone on the web can find and read all notes of this public team.
      See published notes
      Unpublish note
      Please check the box to agree to the Community Guidelines.
      View profile
    • Commenting
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
      • Everyone
    • Suggest edit
      Permission
      Disabled Forbidden Owners Signed-in users Everyone
    • Enable
    • Permission
      • Forbidden
      • Owners
      • Signed-in users
    • Emoji Reply
    • Enable
    • Versions and GitHub Sync
    • Note settings
    • Note Insights
    • Engagement control
    • Transfer ownership
    • Delete this note
    • Save as template
    • Insert from template
    • Import from
      • Dropbox
      • Google Drive
      • Gist
      • Clipboard
    • Export to
      • Dropbox
      • Google Drive
      • Gist
    • Download
      • Markdown
      • HTML
      • Raw HTML
Menu Note settings Versions and GitHub Sync Note Insights Sharing URL Create Help
Create Create new note Create a note from template
Menu
Options
Engagement control Transfer ownership Delete this note
Import from
Dropbox Google Drive Gist Clipboard
Export to
Dropbox Google Drive Gist
Download
Markdown HTML Raw HTML
Back
Sharing URL Link copied
/edit
View mode
  • Edit mode
  • View mode
  • Book mode
  • Slide mode
Edit mode View mode Book mode Slide mode
Customize slides
Note Permission
Read
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Write
Only me
  • Only me
  • Signed-in users
  • Everyone
Only me Signed-in users Everyone
Engagement control Commenting, Suggest edit, Emoji Reply
  • Invite by email
    Invitee

    This note has no invitees

  • Publish Note

    Share your work with the world Congratulations! 🎉 Your note is out in the world Publish Note

    Your note will be visible on your profile and discoverable by anyone.
    Your note is now live.
    This note is visible on your profile and discoverable online.
    Everyone on the web can find and read all notes of this public team.
    See published notes
    Unpublish note
    Please check the box to agree to the Community Guidelines.
    View profile
    Engagement control
    Commenting
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    • Everyone
    Suggest edit
    Permission
    Disabled Forbidden Owners Signed-in users Everyone
    Enable
    Permission
    • Forbidden
    • Owners
    • Signed-in users
    Emoji Reply
    Enable
    Import from Dropbox Google Drive Gist Clipboard
       owned this note    owned this note      
    Published Linked with GitHub
    Subscribed
    • Any changes
      Be notified of any changes
    • Mention me
      Be notified of mention me
    • Unsubscribe
    Subscribe
    # 2017q3 Homework1 (phonebook) contributed by <`amikai`> ###### tags: `sysprog2017` ### Reviewed by `yusihlin` * 參考資料可以在內容中直接以超連結或索引的方式指向,放在多處顯得有些雜亂 * 使用 BKRD HASH 缺乏動機、說明,應對程式運行的結果加以解釋 * chunking 部分的解說不夠詳盡,可以的話列出部分程式碼,直接貼出分析圖有點突兀 --- - [ ] 準備 GNU/Linux 開發工具 - [ ] 學習使用 Git 與 GitHub - [ ] 學習效能分析工具 - [x] 研究軟體最佳化 - [x] 技術報告寫作練習 # 開發環境 cpu 資訊: ``` $ lscpu Architecture: x86_64 CPU op-mode(s): 32-bit, 64-bit Byte Order: Little Endian CPU(s): 8 On-line CPU(s) list: 0-7 Thread(s) per core: 2 Core(s) per socket: 4 Socket(s): 1 NUMA node(s): 1 Vendor ID: GenuineIntel CPU family: 6 Model: 26 Model name: Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz Stepping: 5 CPU MHz: 1600.000 CPU max MHz: 2668.0000 CPU min MHz: 1600.0000 BogoMIPS: 5332.22 Virtualisation: VT-x L1d cache: 32K L1i cache: 32K L2 cache: 256K L3 cache: 8192K NUMA node0 CPU(s): 0-7 ``` 會在想要用底下這個指令得原因是可以的到 cache line 大小, 我可以針對 cache line 大小做設計 ``` $ x86info -c x86info v1.31pre Dave Jones 2001-2011 Feedback to <davej@redhat.com>. Found 8 identical CPUs Extended Family: 0 Extended Model: 1 Family: 6 Model: 26 Stepping: 5 Type: 0 (Original OEM) CPU Model (x86info's best guess): Core i7 (Nehalem) [bloomfield/gainestown] Processor name string (BIOS programmed): Intel(R) Core(TM) i7 CPU 920 @ 2.67GHz Cache info L1 Instruction cache: 32KB, 4-way associative. 64 byte line size. L1 Data cache: 32KB, 8-way associative. 64 byte line size. L2 (MLC): 256KB, 8-way associative. 64 byte line size. TLB info Instruction TLB: 2MB or 4MB pages, fully associative, 7 entries Instruction TLB: 4K pages, 4-way associative, 64 entries. Data TLB: 4KB or 4MB pages, fully associative, 32 entries. Data TLB: 4KB pages, 4-way associative, 64 entries Data TLB: 4K pages, 4-way associative, 512 entries. Data TLB: 4KB or 4MB pages, fully associative, 32 entries. Data TLB: 4KB pages, 4-way associative, 64 entries 64 byte prefetching. Data TLB: 4K pages, 4-way associative, 512 entries. Found unknown cache descriptors: e4 Total processor threads: 8 This system has 1 quad-core processor with hyper-threading (2 threads per core) running at an estimated 2.70GHz ``` kernel 資訊: ``` $ uname -a Linux ruei-pc 4.4.0-96-generic #119-Ubuntu SMP Tue Sep 12 14:59:54 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux ``` # 開發使用工具 ## git TODO ## 效能分析工具 再往路上看到一張圖, 原來我們用的 perf 只是眾多效能工具一個而已 ![](http://www.brendangregg.com/Perf/linux_observability_tools.png) 參考資料: - http://www.brendangregg.com/linuxperf.html ## gnuplot TODO # 參考資料: - http://www.phyast.pitt.edu/~zov1/gnuplot/html/histogram.html ## Makefile TODO # phonebook ## 程式研讀 ### Defined `if` 和 `#elif` 表達式用來測試某個 `macro` 是否被定義, 舉例: 當 `name` 有被定義在程式裡時, `defined name` 和 `defined (name)` 這兩個表達式的值會為 1, 否則為 0. 因此 `# if defined MACRO` 和 `#ifdef MACRO` 是相同的. 當你想要測試不只一個 `macro` 是否背定義時, `defined` 就很實用, 舉例: ``` #if defined (__vax__) || defined (__ns16000__) ``` 若 `__vax__` 和 `__ns16000__` 皆被定義時, 此條 if 將會成功. 原本的程式有用到 `__GNUC__` 這個 macro, 我去翻了一下 gcc doc, gcc 定義了 `__GNUC__` `__GNUC_MINOR__1` `__GNUC_PATCHLEVEL__` 3個 macro, 用 gcc -v 會看到 gcc 的版本, 若版本為 x.y.z 則 `__GNUC__` 為 x  `__GNUC_MINOR__1` 為 y `__GNUC_PATCHLEVEL__` 為 z, 你可以使用此 3 個 macro 限制某些版本只編譯哪些 code. 以下式我實驗的 code. ``` $ gcc -v 2>&1 | tail -n 1 gcc version 5.4.0 20160609 (Ubuntu 5.4.0-6ubuntu1~16.04.4) $ cat testVersion.c #include <stdio.h> int main() { #if __GNUC__ == 5 && __GNUC_MINOR__ == 4 && __GNUC_PATCHLEVEL__ == 0 printf("%d.%d.%d\n", __GNUC__, __GNUC_MINOR__, __GNUC_PATCHLEVEL__); #endif } $ gcc testVersion.c -o test && ./test 5.4.0 ``` 參考: - [Defined - gcc doc ](https://gcc.gnu.org/onlinedocs/cpp/Defined.html) - [Common Predefined Macros - gcc doc](https://gcc.gnu.org/onlinedocs/cpp/Common-Predefined-Macros.html) ### __builtin___clear_cache TODO ### 時間偵測 TODO 參考: - [Other Built-in Functions Provided by GCC - gcc doc](https://gcc.gnu.org/onlinedocs/gcc/Other-Builtins.html) ## 改進方法 ### 準備工作 看到原始的版本沒有將記憶體完整釋放, 編譯完後, 我們可以再寫程式的時候用 valgrind 檢查: ``` $ valgrind --leak-check=full ./phonebook_orig ``` 我將有用的資訊擷取出來, 首先以下這些: ``` LEAK SUMMARY: definitely lost: 136 bytes in 1 blocks indirectly lost: 16,034,808 bytes in 117,903 blocks possibly lost: 31,551,320 bytes in 231,995 blocks ``` 滿滿的 memory leak, 所以只要把以下程式更改即可: ``` c /*-------------- before fix --------------*/ if (pHead->pNext) free(pHead->pNext); free(pHead); /*-------------- after fix --------------*/ entry *temp = pHead; while(temp) { pHead = pHead->pNext; free(temp); temp = pHead; } ``` valgrind 噴出現了一個錯誤: ``` Use of uninitialised value of size 8 ``` 最後發現是原本的程式都棄用第一個 entry, 只有配置記憶體, 但是沒有放任何資訊, 我重寫了 append function 和部份的 main.c, 全部貼上來會有些雜亂, 所以我放了更改的 [commit](https://github.com/as23041248/phonebook/commit/465211f4c4cc6a5791df3493266c5c8b51674d1b) 參考資料: - [0140454 共筆](https://hackmd.io/EwMwhgDBDGBsYFoDsBGWAOBAWdSCsCAnCIQEYLAoRgAmApvHiAMylA==?view) ### 縮小 struct 原本的程式只要比對 lastName, 所以將次要的資訊使用指標紀錄就好, 這樣可以增加放在 struct 資料放在 cache 裡的數量, 進而達到降低 cache miss 的目的 ``` c typedef struct __PHONE_BOOK_ENTRY { char lastName[MAX_LAST_NAME_SIZE]; struct __PHONE_BOOK_ENTRY *pNext; } entry; typedef struct _PERSONAL_INFO { char firstName[16]; char email[16]; char phone[10]; char cell[10]; char addr1[16]; char addr2[16]; char city[16]; char state[2]; char zip[5]; } _PERSONAL_INFO; ``` 以下是實驗數據: - cache misses 由原本的 92% 下降到 75% ``` ----------------------- before reduce struct size ----------------------- 100 runs 1,224,244 cache-misses # 92.509 % of all cache refs ( +- 0.60% ) 1,323,384 cache-references ( +- 0.75% ) 328,196,229 instructions # 1.05 insns per cycle ( +- 0.01% ) 312,636,095 cycles ( +- 0.81% ) 0.114970283 seconds time elapsed ( +- 0.88% ) ----------------------- after reduce struct size ----------------------- 100 runs 225,574 cache-misses # 75.978 % of all cache refs ( +- 0.57% ) 296,893 cache-references ( +- 0.99% ) 286,371,069 instructions # 1.54 insns per cycle ( +- 0.02% ) 186,522,979 cycles ( +- 0.37% ) 0.070776705 seconds time elapsed ( +- 0.49% ) ``` ![](https://i.imgur.com/PiQtwDG.png) 注意: 從此這裡開始之後都是從 reduce size 版本, 繼續建構的, 之後將不會特別提到 ### BKRD Hash ``` c #define HASH_TABLE_SIZE 1003 // BKDR Hash Function unsigned int BKDRHash(char *str) { unsigned int seed = 131; // 31 131 1313 13131 131313 etc.. unsigned int hash = 0; while (*str) { hash = hash * seed + (*str++); } return (hash & 0x7FFFFFFF) % HASH_TABLE_SIZE; } ``` ``` ----------------------- BKDR Hash ----------------------- 100 runs 585,165 cache-misses # 51.258 % of all cache refs ( +- 1.16% ) 1,141,606 cache-references ( +- 0.51% ) 300,929,371 instructions # 1.06 insns per cycle ( +- 0.01% ) 284,154,565 cycles ( +- 0.49% ) 0.104957176 seconds time elapsed ( +- 0.51% ) ``` ![](https://i.imgur.com/2OFzod0.png) 從此資料看來 hash 在 append 上的效能需要改善, 但是 findName 的處理是相當快的 參考資料: - http://www.jianshu.com/p/0be67bf8887e - http://www.cse.yorku.ca/~oz/hash.html - https://stackoverflow.com/questions/7666509/hash-function-for-string - http://www.eternallyconfuzzled.com/tuts/algorithms/jsw_tut_hashing.aspx ### MMAP + BKDR Hash 看了 [yenWu](https://hackmd.io/s/BkN1JNQp#) 學長的筆記才知道 linux 還有 mmap 這種函數可以用, memory map i/o 比起 read() 和 write() 效能來的好, 因為 read() 和 write() 會有兩次傳輸: 檔案與和 kernrl buffer cache 之間, 而另一個則是 kernel buffer cache 和 user space 之間. 而我們每次 malloc 一次都是一個 system call, 若直接使用 mmap 將檔案映射到記憶體上, 就可以直接存取整個檔案了. 而且這麼大量的存取如果使用 malloc 一小塊一小塊的分配則會造成記憶體不續,記憶體不連續就無法達到 locality 的優點 ![](https://images.devshed.com/ds/stories/Advanced_File_IO/image_1.JPG ) 我使用 hash 是為了在 findName 上降低查找時間, 而 mmap 則會改善append 的效能 (替換 malloc) 那麼 mmap 要怎麼跟 hash 結合呢, 我先將原本的 phonebook 資料讀入之後, 把資料放在 struct 裡, 再以寫入二進制檔的方式, 寫到另一個 file 裡, 而這是我用的 struct ```c typedef struct __PHONE_BOOK_ENTRY { char lastName[MAX_LAST_NAME_SIZE]; struct __PHONE_BOOK_ENTRY *pNext; _PERSONAL_INFO *detail; } entry; typedef struct _PERSONAL_INFO { char firstName[16]; char email[16]; char phone[10]; char cell[10]; char addr1[16]; char addr2[16]; char city[16]; char state[2]; char zip[5]; } _PERSONAL_INFO; ``` 為什麼不寫入 lastName 而寫入整個 struct, 以下有兩個好處: - 直接寫入 struct 則大小就一定是一致的, 在使用記憶體映射時方便迭代讀取 - pNext 的大小也映射再記憶體上了, 這樣我就直接對 pNext 存取就好而不是再另外再配置一次 然後我將此二進制檔使用 mmap 函數映設到記憶體上, 並直接對記憶體存取, 之後我就可以做使用了. 接下來就是 hash 的部份了: `append` 只要迭代記憶體上的每一個 struct 並做 hash function 再放到 hash table 相對應的 slot 即可, 以下是片斷程式碼 ``` c entry *pHead[HASH_TABLE_SIZE], *tail[HASH_TABLE_SIZE]; void *block_start = mmap(NULL, size_of_file, PROT_WRITE, MAP_PRIVATE, fd, 0); void *iterate = block_start; while(iterate < block_start + size_of_file) { append(iterate, pHead, tail); iterate += size_of_entry; } ----------------------- append ----------------------- void append(void *addr, entry *pHead[], entry *tail[]) { entry *e = (entry*) addr ; e->pNext = NULL; unsigned int hashVal = BKDRHash(e->lastName); if(pHead[hashVal] == NULL) { pHead[hashVal] = tail[hashVal] = e; return; } tail[hashVal]->pNext = e; tail[hashVal] = e; } ``` 想法非常的簡單, 但是做抽象化的時候想的好久呀 ``` ----------------------- BKDR Hash + mmap ----------------------- 100 runs 386,254 cache-misses # 54.674 % of all cache refs ( +- 0.36% ) 706,469 cache-references ( +- 3.23% ) 246,095,842 instructions # 1.16 insns per cycle ( +- 0.03% ) 211,475,364 cycles ( +- 1.11% ) 0.158217696 seconds time elapsed ( +- 3.03% ) ``` ### MMAP + BKDR Hash + Chuinking 老師在上課時給了一個提示就是以下這[網址](https://www.cs.cornell.edu/courses/cs3110/2010sp/lectures/lec24.html) 使用 Hash Table 加上 Chaning 的方法,每次存取的資料在記憶體上都不是連續的,為了增加locality 我嘗試此用 Chunking 的方法, 在最上面有提到我的 cache line 有 64 bytes 所以我會將我的 phonebook 裡的每個 entry 塞滿到 64bytes: TODO ![](https://i.imgur.com/pbzM8FR.png) 另外我做了一個無聊的實驗去測試 grep 有多快: ``` bash $ time grep zyxel dictionary/words.txtzyxel zyxels real 0m0.007s user 0m0.004s sys 0m0.000sqj ``` TODO 解釋此數值 參考: - [yenWu 共筆](https://hackmd.io/s/BkN1JNQp#) - [The Linux Programmin Interface](http://man7.org/tlpi/) - [Linux System Programming, Second Edition](http://shop.oreilly.com/product/0636920026891.do)] # 問題思考 - Q: 有代表性嗎?跟真實英文姓氏差異大嗎? A: 有沒有代表性我真的想不到, 若跟真實英文姓名差很多的話, 那使用 hash table 過後的結果也不太一樣 TODO: 使用不同的 dataset - Q: 資料難道「數大就是美」嗎? A: 自然風景數大就是美, 但是資料一大就難搞了, 而人類也因為資料的成長發展了許多技術, 像是 Hadoop, Spark 等等 TODO: 換大一點的 dataset - Q: 如果我們要建立符合真實電話簿情境的輸入,該怎麼作? A: 真實的電話簿情境, 資料量可能不是只有 3 萬多筆, 我會先去學習已知的技術, 像是 MogoDB 之類的東西 TODO: 可以嘗試使用真實的資料做分析 # 補強工作 - mmap 在 locality 上的分析 - 可以使用計算機組織的觀點分析, 像是 asscoiative, locality, branch prediction 等等觀點分析 (perf) - 可以使用標準差, 平均, CDF, PDF 觀點 - 複習 big-o 的前提 - malloc 的位置是否連續 - ptmalloc - 看一下技術報告撰寫方式 - 研讀 https://www.cs.cornell.edu/courses/cs3110/2010sp/lectures/lec24.html

    Import from clipboard

    Paste your markdown or webpage here...

    Advanced permission required

    Your current role can only read. Ask the system administrator to acquire write and comment permission.

    This team is disabled

    Sorry, this team is disabled. You can't edit this note.

    This note is locked

    Sorry, only owner can edit this note.

    Reach the limit

    Sorry, you've reached the max length this note can be.
    Please reduce the content or divide it to more notes, thank you!

    Import from Gist

    Import from Snippet

    or

    Export to Snippet

    Are you sure?

    Do you really want to delete this note?
    All users will lose their connection.

    Create a note from template

    Create a note from template

    Oops...
    This template has been removed or transferred.
    Upgrade
    All
    • All
    • Team
    No template.

    Create a template

    Upgrade

    Delete template

    Do you really want to delete this template?
    Turn this template into a regular note and keep its content, versions, and comments.

    This page need refresh

    You have an incompatible client version.
    Refresh to update.
    New version available!
    See releases notes here
    Refresh to enjoy new features.
    Your user state has changed.
    Refresh to load new user state.

    Sign in

    Forgot password

    or

    By clicking below, you agree to our terms of service.

    Sign in via Facebook Sign in via Twitter Sign in via GitHub Sign in via Dropbox Sign in with Wallet
    Wallet ( )
    Connect another wallet

    New to HackMD? Sign up

    Help

    • English
    • 中文
    • Français
    • Deutsch
    • 日本語
    • Español
    • Català
    • Ελληνικά
    • Português
    • italiano
    • Türkçe
    • Русский
    • Nederlands
    • hrvatski jezik
    • język polski
    • Українська
    • हिन्दी
    • svenska
    • Esperanto
    • dansk

    Documents

    Help & Tutorial

    How to use Book mode

    Slide Example

    API Docs

    Edit in VSCode

    Install browser extension

    Contacts

    Feedback

    Discord

    Send us email

    Resources

    Releases

    Pricing

    Blog

    Policy

    Terms

    Privacy

    Cheatsheet

    Syntax Example Reference
    # Header Header 基本排版
    - Unordered List
    • Unordered List
    1. Ordered List
    1. Ordered List
    - [ ] Todo List
    • Todo List
    > Blockquote
    Blockquote
    **Bold font** Bold font
    *Italics font* Italics font
    ~~Strikethrough~~ Strikethrough
    19^th^ 19th
    H~2~O H2O
    ++Inserted text++ Inserted text
    ==Marked text== Marked text
    [link text](https:// "title") Link
    ![image alt](https:// "title") Image
    `Code` Code 在筆記中貼入程式碼
    ```javascript
    var i = 0;
    ```
    var i = 0;
    :smile: :smile: Emoji list
    {%youtube youtube_id %} Externals
    $L^aT_eX$ LaTeX
    :::info
    This is a alert area.
    :::

    This is a alert area.

    Versions and GitHub Sync
    Get Full History Access

    • Edit version name
    • Delete

    revision author avatar     named on  

    More Less

    Note content is identical to the latest version.
    Compare
      Choose a version
      No search result
      Version not found
    Sign in to link this note to GitHub
    Learn more
    This note is not linked with GitHub
     

    Feedback

    Submission failed, please try again

    Thanks for your support.

    On a scale of 0-10, how likely is it that you would recommend HackMD to your friends, family or business associates?

    Please give us some advice and help us improve HackMD.

     

    Thanks for your feedback

    Remove version name

    Do you want to remove this version name and description?

    Transfer ownership

    Transfer to
      Warning: is a public team. If you transfer note to this team, everyone on the web can find and read this note.

        Link with GitHub

        Please authorize HackMD on GitHub
        • Please sign in to GitHub and install the HackMD app on your GitHub repo.
        • HackMD links with GitHub through a GitHub App. You can choose which repo to install our App.
        Learn more  Sign in to GitHub

        Push the note to GitHub Push to GitHub Pull a file from GitHub

          Authorize again
         

        Choose which file to push to

        Select repo
        Refresh Authorize more repos
        Select branch
        Select file
        Select branch
        Choose version(s) to push
        • Save a new version and push
        • Choose from existing versions
        Include title and tags
        Available push count

        Pull from GitHub

         
        File from GitHub
        File from HackMD

        GitHub Link Settings

        File linked

        Linked by
        File path
        Last synced branch
        Available push count

        Danger Zone

        Unlink
        You will no longer receive notification when GitHub file changes after unlink.

        Syncing

        Push failed

        Push successfully