\nLooking at the profile for our program so far,\n35% of the CPU time is spent copying strings around.\nLet’s see if we can improve that.\n<\/p>\n
\nThe best way to speed up copying strings is not to copy them in\nthe first place. Using a \nLet’s use plain string pointers rather than \nThe \nSince we aren’t using \nThere isn’t a \nThe last bits of the change are to get rid of the\ntemporary \nThis new program clocks in at 120ms\n(or 290ms if you include the time it takes to destroy the\ndictionary).\nAgain, we’re twice as fast as the previous version, but still\nhaven’t quite crossed the 100ms barrier.\nAnd the time to destroy the dictionary isn’t starting to be a\nmore significant factor.\nBut I have another trick up my sleeve.\n<\/p>\n \n[Raymond is currently on vacation; this message was pre-recorded.]<\/p>\n","protected":false},"excerpt":{"rendered":" Looking at the profile for our program so far, 35% of the CPU time is spent copying strings around. Let’s see if we can improve that. The best way to speed up copying strings is not to copy them in the first place. Using a wstring in our DictionaryEntry structure forces the vector class to […]<\/p>\n","protected":false},"author":1069,"featured_media":111744,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"_acf_changed":false,"footnotes":""},"categories":[1],"tags":[25],"class_list":["post-35613","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-oldnewthing","tag-code"],"acf":[],"blog_post_summary":" Looking at the profile for our program so far, 35% of the CPU time is spent copying strings around. Let’s see if we can improve that. The best way to speed up copying strings is not to copy them in the first place. Using a wstring in our DictionaryEntry structure forces the vector class to […]<\/p>\n","_links":{"self":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts\/35613","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/users\/1069"}],"replies":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/comments?post=35613"}],"version-history":[{"count":0,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/posts\/35613\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/media\/111744"}],"wp:attachment":[{"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/media?parent=35613"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/categories?post=35613"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/devblogs.microsoft.com\/oldnewthing\/wp-json\/wp\/v2\/tags?post=35613"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}wstring<\/code> in our\nDictionaryEntry<\/code> structure forces the vector<\/code>\nclass to copy the string data, when all we really need to copy\nis the pointer and size information. The actual characters can stay\nput. C++ has a copy constructor but not a “move” constructor.\n<\/p>\nwstring<\/code>\nobjects. The “copy constructor” for a string pointer is just to\ncopy the pointer—exactly what we want here.\n<\/p>\n\nstruct DictionaryEntry\n{\n bool Parse(const WCHAR* begin, const WCHAR* end<\/font>);\n void Destruct() {\n delete[] m_pszTrad;\n delete[] m_pszSimp;\n delete[] m_pszPinyin;\n delete[] m_pszEnglish;\n }\n LPWSTR m_pszTrad;\n LPWSTR m_pszSimp;\n LPWSTR m_pszPinyin;\n LPWSTR m_pszEnglish;<\/font>\n};\n<\/pre>\nDictionaryEntry<\/code> is no longer a structure of\nwstring<\/code>s but rather is just a structure of\nLPWSTR<\/code>s, which copy much faster.\nThe cost for this, however, is having to free all the strings\nmanually in the dictionary destructor (which we will see below).\n<\/p>\nwstring<\/code>s any more, we need\nto allocate the memory for the strings and copy them the old fashioned way.\n<\/p>\n\nLPWSTR AllocString(const WCHAR* begin, const WCHAR* end)\n{\n int cch = end - begin + 1;\n LPWSTR psz = new WCHAR[cch];\n lstrcpynW(psz, begin, cch);\n return psz;\n}<\/font>\nbool DictionaryEntry::Parse(\n const WCHAR* begin, const WCHAR* end<\/font>)\n{\n const WCHAR* pch = std::find(begin, end, L' ');\n if (pch >= end) return false;\n m_pszTrad = AllocString(begin, pch);\n begin = std::find(pch, end, L'[') + 1;\n if (begin >= end) return false;\n pch = std::find(begin, end, L']');\n if (pch >= end) return false;\n m_pszPinyin = AllocString(begin, pch);\n begin = std::find(pch, end, L'\/') + 1;\n if (begin >= end) return false;\n for (pch = end; *--pch != L'\/'; ) { }\n if (begin >= pch) return false;\n m_pszEnglish = AllocString(begin, pch);<\/font>\n return true;\n}\n<\/pre>\nstd::rfind<\/code> function, so I coded\nup a backwards-search-for-slash loop inline.\nExercise<\/strong>: Why don’t I have to check that\npch<\/code> hasn’t underflowed beyond the beginning of the\nstring?\n<\/p>\n\nclass Dictionary\n{\npublic:\n Dictionary();\n ~Dictionary();<\/font>\n int Length() { return v.size(); }\n const DictionaryEntry& Item(int i) { return v[i]; }\nprivate:\n vector<DictionaryEntry> v;\n};\nDictionary::Dictionary()\n{\n ...\n if (cchResult){\n \/\/ wstring line(buf, cchResult);<\/strike><\/font>\n DictionaryEntry de;\n if (de.Parse(buf, buf + cchResult<\/font>)) {\n ...\n}\nDictionary::~Dictionary()\n{\n for (vector<DictionaryEntry>::iterator i = v.begin();\n i != v.end(); i++) {\n i->Destruct();\n }\n}<\/font>\n<\/pre>\nwstring<\/code> we parsed from, since we don’t\nneed it any more, and to free all the strings in the\nDictionary<\/code>‘s destructor.\n<\/p>\n