| Message ID | c9b263c6182555a3241c48f368e477e5@autistici.org |
|---|---|
| State | New |
| Headers | show |
| Series | Optimization for std::to_string() | expand |
On 20/07/18 12:44 +0300, niXman wrote: > >Hi, > > > >Patch in attachments. Thanks. How did you verify it's an optimization? The to_string functions always pass at least __n=16 to __to_xstring, which is larger than the SSO buffer in std::__cxx11::string, and so forces a memory allocation. The current code uses alloca and if the result fits in the SSO buffer there is no allocation. e.g. to_string(999999) doesn't allocate from the heap at all for the new std::__cxx11::string. For the old COW string this doesn't make much difference, because we always allocate. Your patch will mean we return a string with excess capacity, rather than one of exactly capacity()==length(). That isn't a problem. >Tested on x86_64-linux-gnu. > > >-- >Regards, niXman >___________________________________________________ >C++ for Bitcoins: github.com/niXman >Index: libstdc++-v3/include/ext/string_conversions.h >=================================================================== >--- libstdc++-v3/include/ext/string_conversions.h (revision 262879) >+++ libstdc++-v3/include/ext/string_conversions.h (working copy) >@@ -100,19 +100,18 @@ > __builtin_va_list), std::size_t __n, > const _CharT* __fmt, ...) > { >- // XXX Eventually the result should be constructed in-place in >- // the __cxx11 string, likely with the help of internal hooks. >- _CharT* __s = static_cast<_CharT*>(__builtin_alloca(sizeof(_CharT) >- * __n)); >+ _String __str(__n, 0); > > __builtin_va_list __args; > __builtin_va_start(__args, __fmt); > >- const int __len = __convf(__s, __n, __fmt, __args); >+ const std::size_t __len = __convf(&__str[0], __n, __fmt, __args); > > __builtin_va_end(__args); > >- return _String(__s, __s + __len); >+ __str.resize(__len); >+ >+ return __str; > } > > _GLIBCXX_END_NAMESPACE_VERSION
Jonathan Wakely 2018-07-20 13:05: > On 20/07/18 12:44 +0300, niXman wrote: > Thanks. How did you verify it's an optimization? Optimization is that there is no superfluous copying into string. > The to_string functions always pass at least __n=16 to __to_xstring, > which is larger than the SSO buffer in std::__cxx11::string, and so > forces a memory allocation. I didn't think about this...
On 20/07/18 13:51 +0300, niXman wrote: >Jonathan Wakely 2018-07-20 13:05: >>On 20/07/18 12:44 +0300, niXman wrote: > >>Thanks. How did you verify it's an optimization? >Optimization is that there is no superfluous copying into string. > >>The to_string functions always pass at least __n=16 to __to_xstring, >>which is larger than the SSO buffer in std::__cxx11::string, and so >>forces a memory allocation. >I didn't think about this... Avoiding an unconditional dynamic allocation is the main advantage of using alloca.
Index: libstdc++-v3/include/ext/string_conversions.h =================================================================== --- libstdc++-v3/include/ext/string_conversions.h (revision 262879) +++ libstdc++-v3/include/ext/string_conversions.h (working copy) @@ -100,19 +100,18 @@ __builtin_va_list), std::size_t __n, const _CharT* __fmt, ...) { - // XXX Eventually the result should be constructed in-place in - // the __cxx11 string, likely with the help of internal hooks. - _CharT* __s = static_cast<_CharT*>(__builtin_alloca(sizeof(_CharT) - * __n)); + _String __str(__n, 0); __builtin_va_list __args; __builtin_va_start(__args, __fmt); - const int __len = __convf(__s, __n, __fmt, __args); + const std::size_t __len = __convf(&__str[0], __n, __fmt, __args); __builtin_va_end(__args); - return _String(__s, __s + __len); + __str.resize(__len); + + return __str; } _GLIBCXX_END_NAMESPACE_VERSION