由於要練習FireFox OS跟 Ubuntu OS於是還是做了點小投資買了一檯二手的Android Galaxy Nexus來刷
(由於在FFOS 與 Ubuntu OS 的device list找到比較便宜與容易購買的就是這個)

使用心得:
由於我自己習慣使用iPhone跟設計iPhone的App，其實很不習慣Android的“上一頁”實體按鍵．不過多使用幾次之後其實也還好了． 也發現Android沒有iOS的一些方便的功能如下:

• Tab bar: 也就是下方的Tab按鍵，這是Android App RD要花心思去搞定的．

• Storyboard: Android 使用Activity設計概念並且用Intent來溝通，但是沒有StoryBoard的階層觀念．

其他細節等到我其它文章再寫..

關於Root 手機

• 打開Development mode. [開發人員選項]

  • 打開[設定] -> [關於手機] -> 選到[版本號碼] ( 大概點個七下) 啓動  (這是什麼怪東西？)

• 進入 [設定]->[開發人員選項]-> [USB 偵錯]  啓動

• 要Root 前，必須要先解開Boot loader lock的限制才能夠Root

  • 下載 http://downloadandroidfiles.org/download-unlock-bootloader-gsm-verizon-sprint-zip/

  • 進入 [FASTBOOT Mode] 

    • 先關機

    • 左手按下兩個音量鍵 （兩個都要按)不要放

    • 再按下 電源鍵

  • 這時候手機會進入FastBoot Mode (也就是有機器人躺在那邊)

  • 執行剛剛下載的檔案

  • 這時候會進入警告標語

  • 利用 音量鍵選擇 ，電源鍵確認

• 解開後，手機的OS會重灌

• 一樣的步驟

  • 進入 [設定]->[開發人員選項]-> [USB 偵錯]  啓動

  • 進入 [FASTBOOT Mode] 

• 下載 [Root kit] http://downloadandroidfiles.org/download-updated-galaxynexusroot-gsm-verizon-sprint-zip/

  • 執行先解開 [Boot Loader Lock]

  • 開機 -> 設定 -> USB 偵錯

  • 按下enter繼續跑

  • 選擇(1)  GSM ClockworkMod Recovery (Version 6.0.2.3)

  • 利用 [音量鍵] 改到 [Recovery mode] -> 電源鍵執行

  • 選擇 install from SD card 

  • 安裝位於 目錄 0/UPDATE-SuperSU-v1.45.zip

  • 安裝跑完最後選擇No

  • 重開機

• 開機完 看到應用程式裡面有 [SuperSU] 代表你Root 成功．

參考資料：

• http://dottech.org/87439/how-to-unlock-usb-debugging-mode-on-android-4-2-jelly-bean-and-higher-guide/

• http://www.androidrootz.com/2012/12/galaxy-nexus-one-click-mac-toolkit.html

• http://forum.xda-developers.com/showthread.php?t=1790399

Slide: http://goo.gl/RPbGVn

About Evernote SDK API trick or experience discussion: (for iOS/Android.)

• findNote — http://dev.evernote.com/doc/reference/NoteStore.html#Fn_NoteStore_findNotes

• • It must 

  • It is depressed, because it has performance issue to get all notes entity (include all attachment) 

  • But it still could use for now. 

  • It alway need to assign number of notes, but you could recessive to get all notes.

  • Currently Evernote official will suggest to use “findNotesMetadata” and “getNote”

  • if you need to ordering your result, using filter.

• findNotesMetadata — http://dev.evernote.com/doc/reference/NoteStore.html#Fn_NoteStore_findNotesMetadata

• • Please note: findNote performance is still better  than findNotesMetadata + getNote 

• getNote — (http://dev.evernote.com/doc/reference/NoteStore.html#Fn_NoteStore_getNote)

• getNote(string authenticationToken, Types.Guid guid, bool withContent, bool withResourcesData, bool withResourcesRecognition,

  bool withResourcesAlternateData)

  • withResourcesRecognition — 

  • It help to get back image recognition(evernote specific), but it will depends on priority (pay account first)

  • withContent — 

  • content is HTML data using HTML data , you could easily to put  HTML data to browser.

  • withResourcesData — 

  • For attachment or image it will contain a hash code for your to reserve file .

• createNoteBook — http://dev.evernote.com/doc/reference/NoteStore.html#Fn_NoteStore_createNotebook

• • NoteBook “setName” could not duplicate.

  • It could be easily to getNoteBook first before you try to createNoteBook.

• createNote —  http://dev.evernote.com/doc/reference/NoteStore.html#Fn_NoteStore_createNote

• • createNote using “setTitle” not setName. :)

  • Need to add for your HTML before you create note, using EvernoteUtil.createEnMediaTag 

  • EML (Evernote Metadata Language)

  • Need take care if you need using browser to display it.

  • might broken, because any WYSIWYG editor has the same issue.

• Image handle in Android

• • Need using InputStream to handle it locally before you add to note.

Google Cloud Platform

• Slide:

• • http://www.slideshare.net/littleq0903/introduction-to-google-cloud-platform/1

• What  is google cloud platform

• • Using google storage 

  • Goole bandwidth 

  • Cloud encapsulation 

• Cloud Platform:

• • Cloud SQL

  • it is MySQL 5.5

  • Cloud Datastore

  • NoSQL (Non-Relationshop SQL) storage. Much faster than normally SQL.

  • Cloud Storage:

  • Protection, similar with Google Drive

  • GAE

  • It much better performance for now.

• Full Functional Service

• • feature:

  • Scalable 

  • Normally

  • LAMP

  • Apache2 (three apache )with one  DB.

  • Master/Slave DB mapping with Apache2.

• PaaS (Platform as a Service)

• GAE (Google App Engine)

• • Language:

  • PHP,JAVA, Python

  • GO: (New language), refer https://developers.google.com/appengine/docs/go/gettingstarted/helloworld

  • Communication:

  • Channel:

  • • Using Web socket

  • Mail:

  • • Using specific mail alias to pass some email to GAE to parsing.

  • XMPP:

  • • message like Facebook and google talk.

  • Outbound socket

  • • Only outbound could pass socket to another service from GAE.

  • Process manager:

  • Cron Job: Similar with Crontab.

  • Computation:

  • Image API: For image resizing or scaling.

  • Map-reduce: Like OpenCL or multiple threading.

  • Application:

  • Big Query:

  • • Terabyte data analysis 

    • SQL-like

Google Compute Engine:

• IaaS (Infrastructure as a Service) — all related could be plug as component to add/remove.

• • Instance—  You could create VM via compute engine for this. [Speed]

  • Similar with MSFT Azure setting but faster and support command line.

  • Persistent Disc— PD [Storage]

  • only 1 for R/W. multiple only support read only.

  • Bound by Zone.

  • Security base on AES-128.

  • Networking— It about communication speed.

  • Block port SMTP(port 25) and block SMTP on SSL.

  • No matter your instance is in US or Euro it also could 

  • API

  • JSON, OAuth2, RESTful.

• Why using Google Compute Engine:

• • Speed

  • Scale

  • The computing power could be change or scale it up according to your request.

  • • Using an agent which create by GAE and it could monitor current status

    • And run CLI if working flow is archive the boundary. (Add VM or Add PD).

  • Global footprint:

  • you could easily to expend your business to whole world.s

• Google has world record about transfer 1TB data and compute it done within 55 sec.

• Out side is open API outbound, but it could has its own storage, service, internet. All working on API.

• Every charge is by request:

• • Every request has its own storage and internet transfer.

• Zone different management:

• • It could be separate to different zone to manage your service.

記錄一下，本來是寫英文後來還是決定改成中文的

• 我的環境:

  • MBA (Mac Book Air 2012 Ivy-Bridge) with 8G Ram, 128G SSD.

• 環境設定與準備:

  • 主要有兩種方向可以給Android開發者，我兩種都有試試看來決定最容易與方便使用的IDE.

  • Eclipse with ADT (Android Development Tool)

    • 最常使用的開發環境，設定算簡單，不過個人覺得界面使用起來總是哪裡怪怪的，可能因為Eclipse本身屬於開放式架構的關係．

    • 下載 http://developer.android.com/sdk/index.html

  • Android Studio

    • 2013年發佈的Android Studio，我覺得開發新的應該會很棒．不過似乎在打開一些舊的檔案會因為檔案架構的原因出現一些問題．

    • 這個部分對於剛開始學習開發得人是比較麻煩的，可能要等後面版本在修改得時候看看能不能解決得好．

• Android 模擬器 – AVD (Android Virtual Device)

  • AVD 就是一個Android的模擬器可以來執行你的程式．

  • Android Development Tool所提供的模擬器，其實相當不好用．我發生以下的一些問題：.

    • 啟動這個ADT的模擬器(AVD)時間相當的久．在我的環境裡面大概需要3~5分鐘才能夠開完Android的模擬器．更別說把App部署(deploy)到模擬器上面，當然如果要debug那就更殘忍了．

    • 可能是因為需要太多的系統資源，所以常常在編輯過後要重新部署的時候都容易發生錯誤．

  • 利用虛擬機器 Virtual Box所執行的模擬器  Genymotion 

    • 這個是我參加Evernote Developer Meetup 所得到的知識，相當快速（極快）的模擬器．大概在我的環境裡面Android跑起來到App部署完全不需要一分鐘。 真是太令人感動了. 

    • 這裡有詳細的教學:

      • http://blog.mosil.biz/2013/12/android-emulator/

      • http://jdev.tw/blog/3663/using-fastest-android-genymotion-emulator

• 關於Evernote SDK in Android:

  • 這篇文章是官方的文章，其實可以很快地建立出Evernote的程式 http://www.evernoteapi.com/tagged/android

  • 除了裡面的範例程式之外，其實只剩下一些需要注意的地方

  • Evernote API key 申請

    • 可以去這裡申請 (http://dev.evernote.com/doc/) 不過要注意的是裡面的API 是對應另外一台伺服器的帳號 (Sandbox.evernote.com)．跟你所平時在用的並不相同，除非你有寄出Production Server的申請

    • 由於是Sandbox的伺服器上面，所以你可能需要另外在這個伺服器上面去註冊你的帳號(Evernote的帳號)才能用你在Sandbox的API Key去存取他.

  • 申請好之後其實很快速就可以寫出一個login與logou的範例程式．

  • 關於Login/Logout的部分:

    • 不過這邊還有一個可以註記一下的是由於登入Login 是透過OAuth的方式來登入，所以App會離開後然後再度回來，在這裡要注意到的是會由 OnResume回來

    • 但是Logout卻不會遇到這樣的狀況．這是可以注意的．

接下來應該要努力嘗試 Android Development Tool with Eclipse 的程式 porting 到 Android Studio了…

Basic requirement:

• VS2012 runtime

• DX11 runtime

• Use test program “VerifyAmpDevices” to verify it

Advantage:

• Compatible with any GPU which could run DX11 with one EXE.

  • NVIDIA GPUs

  • AMD GPUs (and APUs)

  • Intel GPUs (Ivy Bridge and later)

  • ARM GPUs from various IHVs (soon, e.g. see Mali design)

• Could use Visual Studio IDE for GPU threading debugging

• Could use Visual Studio IDE for GPU profiling

• Base on C++ and STL

Performance Comparison: (refer here)

**
** Video Clip:

• http://channel9.msdn.com/Events/AMD-Fusion-Developer-Summit/AMD-Fusion-Developer-Summit-11/KEYNOTE AFDS keynote which also introduce C++ AMP concept.

Code Resource:

• Hello world with C++ AMP http://blogs.msdn.com/b/nativeconcurrency/archive/2012/04/30/hello-world-using-textures-in-c-amp.aspx

• All C++ AMP samples: http://blogs.msdn.com/b/nativeconcurrency/archive/2012/01/30/c-amp-sample-projects-for-download.aspx

• Very simple code sample slide: http://ecn.channel9.msdn.com/content/DanielMoth_CppAMP_Intro.pdf

• Detail analysis article http://kheresy.wordpress.com/2011/06/16/c-accelerated-massive-parallelism/

• Limitation on C++ AMP http://blogs.msdn.com/b/nativeconcurrency/archive/2012/02/07/double-precision-support-in-c-amp.aspx

• • WDDM 1.1, there will be double precision problem, only resolve on WDDM 1.2 (Win8)

• Related code survey found:

  • Debugging in Intel is easy refer Using the Intel® OpenCL SDK Debugger

    • Add file path when you build cl program

      • Note: Must using original path not copy path.

        • EX: If you copy your CL file in post build process, need add original path not debugging CL code path.

      • Note: Path need full path

    • Enable Intel OpenCL SDK debugging in toolIntel SDK

      • Note: Work item set 0,0,0 as default is enough.

• OpenCL kernel need warm up

  • Run any other kernel code first (even not the same application), it will speed up your major CL kernel code

  • AMD’s magic number is to run “twice” on dump kernel

    • Testing result: (SW/CPU 160ms)

      • Intel:

        • 1st time setup 700ms, effect 160

      • AMD:

        • 1st time

          • setup 6000ms

          • effect 16ms

        • 2nd time

          • setup 160ms

          • effect 16ms  (it might goes to 0ms some time)

    • It could be reduce time to pre-load *.cl file but no way to not reduce clBuildProgram. (Program will cache result as previous one, even you reset twice the argument.

    • Run two times the same kernel code:

      • After second time, (no matter the same program or not) the CL kernel will cache it and very fast.

      • Run second time, the CL setup code will faster the first time (Found on AMD GPU)

      • Note: IVB don’t have such issue.

• Refer: 

  • Basic concept of OpenCL and sample program OpenCL 教學（一） - Hotball’s Hive

  • Advanced concept and clarify OpenCL comment Richard’s blog: OpenCL 介紹