sysprog
      • 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
        • Owners
        • Signed-in users
        • Everyone
        Owners Signed-in users Everyone
      • Write
        • Owners
        • Signed-in users
        • Everyone
        Owners Signed-in users Everyone
      • Engagement control Commenting, Suggest edit, Emoji Reply
      • Invitee
    • Publish Note

      Publish Note

      Everyone on the web can find and read all notes of this public team.
      Once published, notes can be searched and viewed by anyone online.
      See published notes
      Please check the box to agree to the Community Guidelines.
    • 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
    • Engagement control
    • Transfer ownership
    • Delete this note
    • Insert from template
    • Import from
      • Dropbox
      • Google Drive
      • Gist
      • Clipboard
    • Export to
      • Dropbox
      • Google Drive
      • Gist
    • Download
      • Markdown
      • HTML
      • Raw HTML
Menu Note settings Sharing URL Help
Menu
Options
Versions and GitHub Sync 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
Owners
  • Owners
  • Signed-in users
  • Everyone
Owners Signed-in users Everyone
Write
Owners
  • Owners
  • Signed-in users
  • Everyone
Owners Signed-in users Everyone
Engagement control Commenting, Suggest edit, Emoji Reply
Invitee
Publish Note

Publish Note

Everyone on the web can find and read all notes of this public team.
Once published, notes can be searched and viewed by anyone online.
See published notes
Please check the box to agree to the Community Guidelines.
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
# Linux 核心專題: 重做 fibdrv > 執行人: seasonwang0905 > [期末專題解說影片](https://youtu.be/hFbArw7STjk) :::success :question: 提問清單 * ? ::: ## 任務簡述 依據 [fibdrv](https://hackmd.io/@sysprog/linux2023-fibdrv) 作業規範,繼續投入 Linux 核心模組和相關程式的開發。 ## TODO: 紀錄閱讀作業說明中所有的疑惑 閱讀 [fibdrv](https://hackmd.io/@sysprog/linux2023-fibdrv) 作業規範,包含「作業說明錄影」和「Code Review 錄影」,本著「誠實面對自己」的心態,在本頁紀錄所有的疑惑,並與授課教師預約討論。 過程中,彙整 [Homework3](https://hackmd.io/@sysprog/linux2023-homework3) 學員的成果,挑選至少三份開發紀錄,提出值得借鏡之處,並重現相關實驗。 ## TODO: 回覆「自我檢查清單」 回答「自我檢查清單」的所有問題,需要附上對應的參考資料和必要的程式碼,以第一手材料 (包含自己設計的實驗) 為佳 ## TODO: 以 [sysprog21/bignum](https://github.com/sysprog21/bignum) 為範本,實作有效的大數運算 理解其中的技巧並導入到 fibdrv 中,並留意以下: * 在 Linux 核心模組中,可用 ktime 系列的 API; * 在 userspace 可用 clock_gettime 相關 API; * 善用統計模型,除去極端數值,過程中應詳述你的手法 * 分別用 gnuplot 製圖,分析 Fibonacci 數列在核心計算和傳遞到 userspace 的時間開銷,單位需要用 us 或 ns (自行斟酌) ## 彙整及重現加速 Fibonacci 運算 參考 [eecheng87](https://hackmd.io/@eecheng/H18gYdsEI) 的實作,他在 fast doubling 的基礎上加入了 `__builtin_clz()` 來加速 Fibonacci 的運算。 Fast doubling 無 `__builtin_clz()` 版本: ```c unsigned long long double_fib1(int n) { unsigned long long f1 = 0, f2 = 1, t1, t2; for (int i = 31; i >= 0; i--) { t1 = f1 * (f2 * 2 - f1); t2 = f2 * f2 + f1 * f1; f1 = t1; f2 = t2; if ((n & (1 << i))) { t1 = f1 + f2; f1 = f2; f2 = t1; } } return f1; } ``` Fast doubling 有 `__builtin_clz()` 版本: ```c unsigned long long double_fib2(int n) { unsigned long long f1 = 0, f2 = 1; int i = 31 - __builtin_clz(n); for (; i >= 0; i--) { unsigned long long t1, t2; t1 = f1 * (f2 * 2 - f1); t2 = f2 * f2 + f1 * f1; f1 = t1; f2 = t2; if ((n & (1 << i))) { t1 = f1 + f2; f1 = f2; f2 = t1; } } return f1; } ``` 在加入 `__builtin_clz()` 前,不論欲求的 n 值為何,迴圈必定會執行 32 ,加入後,可以減少迭代的次數,避免在 n 值很小的時候執行過多的計算 再來,參考並修改了 [Hsuhaoxiang](https://hackmd.io/@Hsuhaoxiang/fibdrv) 的作法,他先計算了傳入的 n 值 (這裡為 k) 在 bitwise 中有多少個 1 並紀錄,以此切換 a 及 b 的位置 ```c unsigned long long double_fib3(int k) { int bin_k[8]; int count = 0; while (k) { bin_k[count] = k % 2; k = k >> 1; count++; } unsigned long long int a = 0, b = 1, t1, t2, temp; for (int i = count - 1; i >= 0; i--) { t1 = a * ((2 * b) - a); t2 = b * b + a * a; a = t1; b = t2; if (bin_k[i] == 1) { temp = a + b; a = b; b = temp; } } return a; } ``` 這裡直接計算 k 的 1 位元個數,然後由 fast doubling 可減少一半迭代數的特性算出 `count` ,在效能上已和使用 `__builtin_clz()` 類的手法接近,改善的方式可能要從記憶體的使用或是加入大數運算 ## 加入大數運算 因為 c 語言沒有合適的資料型態可以存儲超過 $2^{64}-1$ 或 $-2^{64}$ 的數值,故我們需要使用其他手段來作大數運算,參考 [eecheng87](https://hackmd.io/@oscarshiang/linux_fibdrv#%E8%A8%88%E7%AE%97%E7%AC%AC-92-%E9%A0%85%E4%B9%8B%E5%BE%8C%E7%9A%84-Fibonacci-%E6%95%B8) 的作業中所述,他使用以下結構來存取一個動態長度的陣列 ```c typedef struct { char digits[MAXDIGITS]; /* represent the number */ int signbit; /* 1 if positive, -1 if negative */ int lastdigit; /* index of high-order digit */ } bignum; ``` 加法的實作為 ```c int add_bignum(bignum *a, bignum *b, bignum *c) { .. if (a->lastdigit < b->lastdigit) return add_bignum(b, a, c); int k = c->lastdigit = a->lastdigit + 1; c->digits[k--] = '\0'; carry = 0; n_carry = 0; for (i = b->lastdigit - 1, j = a->lastdigit - 1; i >= 0; i--, j--) { carry = b->digits[i] - '0' + a->digits[j] - '0' + carry; c->digits[k--] = (carry % 10) + '0'; carry = carry / 10; if (carry) n_carry++; } for (; j >= 0; j--) { carry = a->digits[j] - '0' + carry; c->digits[k--] = (carry % 10) + '0'; carry = carry / 10; if (carry) n_carry++; } if (carry) c->digits[k] = carry + '0'; else { char string[MAXDIGITS]; strlcpy(string, &c->digits[1], MAXDIGITS); strlcpy(c->digits, string, MAXDIGITS); c->lastdigit = c->lastdigit - k - 1; } return n_carry; } ``` 上面省去了判斷加負數的部份,加負數可以替代成減法的形式。我們需要以字串來紀錄大數,使用時再將其一個一個輸出成大數的樣子 再來是減法的部份 ```c int subtract_bignum(bignum *a, bignum *b, bignum *c) { // int borrow; /* has anything been borrowed? */ // int v; /* placeholder digit */ register int i, j, op = 0; /* counter */ int n_borrow; int temp; c->signbit = PLUS; if ((a->signbit == MINUS) || (b->signbit == MINUS)) { b->signbit = -1 * b->signbit; n_borrow = add_bignum(a, b, c); b->signbit = -1 * b->signbit; return n_borrow; } .. .. int k = c->lastdigit = MAX(a->lastdigit, b->lastdigit); n_borrow = 0; c->digits[k--] = '\0'; for (i = a->lastdigit - 1, j = b->lastdigit - 1; j >= 0; i--, j--) { temp = a->digits[i] - '0' - (b->digits[j] - '0' + op); if (temp < 0) { temp += 10; op = 1; n_borrow++; } else op = 0; c->digits[k--] = temp + '0'; } while (op) { temp = a->digits[i--] - op - '0'; if (temp < 0) { temp += 10; op = 1; n_borrow++; } else op = 0; c->digits[k--] = temp + '0'; } for (; i >= 0; i--) c->digits[k--] = a->digits[i]; for (i = 0; !(c->digits[i] - '0'); i++) ; c->lastdigit = c->lastdigit - i; if (i == a->lastdigit) strlcpy(c->digits, "0", MAXDIGITS); else { char string[MAXDIGITS]; strlcpy(string, &c->digits[i], MAXDIGITS); strlcpy(c->digits, string, MAXDIGITS); } return n_borrow; } ``` 減法中有很多借位的算法,目前還在嘗試理解 :::info 這邊有個疑問:加法、減法輸出 `n_carry` 、`n_borrow` 進位或是借位目的是什麼,用在哪裡? ::: 最後是乘法 ```c void multiply_bignum(bignum *a, bignum *b, bignum *c) { // long int n_d; register long int i, j, k = 0; short int num1[MAXDIGITS], num2[MAXDIGITS], of = 0, res[MAXDIGITS] = {0}; // n_d = (a->lastdigit < b->lastdigit) ? b->lastdigit : a->lastdigit; // n_d++; for (i = 0, j = a->lastdigit - 1; i < a->lastdigit; i++, j--) num1[i] = a->digits[j] - 48; for (i = 0, j = b->lastdigit - 1; i < b->lastdigit; j--, i++) num2[i] = b->digits[j] - 48; res[0] = 0; for (j = 0; j < b->lastdigit; j++) { for (i = 0, k = j; i < a->lastdigit || of; k++, i++) { if (i < a->lastdigit) res[k] += num1[i] * num2[j] + of; else res[k] += of; of = res[k] / 10; res[k] = res[k] % 10; } } for (i = k - 1, j = 0; i >= 0; i--, j++) c->digits[j] = res[i] + 48; c->digits[j] = '\0'; c->lastdigit = k; c->signbit = a->signbit * b->signbit; } ``` 因為字串沒有所謂的相乘,所以只能把數字擷取出來再做乘法,最後把結果存到 `c.->digits` 有了以上大數運算,就可以運用到 Fibonacci 的數列上了 Fast doubling 演算法支援大數運算 ```c bignum a, b; bignum big_two; int_to_bignum(0, &a); int_to_bignum(1, &b); int_to_bignum(2, &big_two); for (int i = 31 - __builtin_clz(k); i >= 0; i--) { bignum t1, t2; bignum tmp1, tmp2; multiply_bignum(&b, &big_two, &tmp1); (void) subtract_bignum(&tmp1, &a, &tmp2); multiply_bignum(&a, &tmp2, &t1); multiply_bignum(&a, &a, &tmp1); multiply_bignum(&b, &b, &tmp2); (void) add_bignum(&tmp1, &tmp2, &t2); copy(&a, &t1); copy(&b, &t2); if ((k & (1 << i)) > 0) { (void) add_bignum(&a, &b, &t1); copy(&a, &b); copy(&b, &t1); } } return a; ``` 輸入第 100 、200 項時,看看輸出結果為何 ```c 354224848179261915075 \\ n = 100 280571172992510140037611932413038677189525 \\ n = 200 ``` 比對 [The Fibonacci numbers](https://r-knott.surrey.ac.uk/Fibonacci/fibtable.html) 發現相同,大數運算成功執行。 ## 研讀上述費氏數列相關材料(包含論文),摘錄關鍵手法,並思考 [clz / ctz](https://en.wikipedia.org/wiki/Find_first_set) 一類的指令對 Fibonacci 數運算的幫助。請列出關鍵程式碼並解說 * Fibonacci演算法實作程式碼: [seasonwang0905](https://github.com/seasonwang0905/fibdrv) James L. Holloway 發表於 1988 年的碩士論文 〈[Algorithms for Computing Fibonacci Numbers Quickly](https://ir.library.oregonstate.edu/downloads/t435gg51w)〉針對多種 Fibonacci 數列之演算法進行模擬及分析,包括遞迴、重複相加、[Binet's Formula](https://proofwiki.org/wiki/Euler-Binet_Formula) (公式解)、[Vorobev等式](https://hackmd.io/SNFjz6Y5SVeYsvooDu-z5A?both)、二項式係數法等等逐項討論,以實際成果來展示各演算法在 Fibonacci 數列之項數 n 改變的情況下,其執行時間及效率究竟如何,最後驗證其論文結果。 * 遞迴 (natural recursive) $fib(n)=\begin{cases} 0 &if \quad n=0\\ 1 &if \quad n=1\\ fib(n-1)+fib(n-2) \quad &if \quad n \ge 2 \end{cases}$ 在介紹函式遞迴的時,常見的例子即計算 Fibonacci 數,因重複呼叫函式的關係,演算法時間複雜度來到 $\mathcal{O}(\lambda_1^{n})$,其中 $\lambda_1>1.5$ ,隨著 n 值增加,執行時間會以指數倍成長,故在 n 值較大時,幾乎不會選用此演算法。 * 重複相加 (repeated addition) 原理與前述相同,只不過使用了 for 迴圈來取代函式遞迴的動作,效率卻有顯著的提昇,隨著 n 值增加,執行時間則呈線性增長。另外,論文中提到的 [K-Fibonacci](https://www.geeksforgeeks.org/k-fibonacci-series/) 數列也有相關的演算法,不過這裡暫不討論此部份。 透過 gnuplot 製圖: ![](https://hackmd.io/_uploads/Sy1jhX0D3.png) 橫軸代表 Fibonacci 數列第 n 項,即 $fib(n)$,縱軸則是其執行時間,不難看出僅僅是 n = 10 ,遞迴演算法的執行時間已來到最初的 6 倍了,其效率可說是驚人的低落。重複相加的演算法可以作為 gnuplot 的練習及效能的參考。 接著,列出改善計算 Fibonacci 數列的演算法 * [Fast Doubling](https://www.nayuki.io/page/fast-fibonacci-algorithms) $F(2n)=2F(n) \times F(n+1)-F(n)^2 \\ F(2n+1)=F(n)^2+F(n+1)^2$ 根據上述關係,著手撰寫程式碼 ```c unsigned long long fast_doubling(int n) { if (n <= 2) return !!n; unsigned long long f1 = 1,f2 = 1, t1, t2, temp; int count = n / 2; for (int i = 0; i < count; i++) { t1 = 2 * f1 * f2 - f1 * f1; t2 = f1 * f1 + f2 * f2; temp = f2; f2 = f1 + temp; f1 = temp; } if (n & 1) return t2; return t1; } ``` 寫完後,馬上來測試效能如何,結果如下圖所示 ![](https://hackmd.io/_uploads/B1X8XNmO2.png) 藍色的線是參考上述提及的 eecheng87 的 fast doubling 無硬體加速版本的演算法,綠色則是我實作的部份,可以發現綠色的執行時間與 n 呈線性成長,效能只有比重複相加的演算法好一些 提問:在 fast doubling 演算法中,最多只需要做 $n/2$ 次的迭代即可算完每項 $F(n)$,即使我設定 n 不會超過 32 次,輸出結果是正確的,但執行時間仍就無法降低呢?是否是因為多了 `f2 = f1 + temp` 這一項呢?又或者應該要支援大數運算後再重新比較效能看看? :::warning 你用 perf 一類的工具去觀察執行的表現,會發現裡頭存在耗時的指令 :notes: jserv :::

Import from clipboard

Paste your webpage below. It will be converted to Markdown.

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 is not available.
Upgrade
All
  • All
  • Team
No template found.

Create custom 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

How to use Slide mode

API Docs

Edit in VSCode

Install browser extension

Get in Touch

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

No updates to save
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