-
Notifications
You must be signed in to change notification settings - Fork 2.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Yet another simplification of Dragonbox #2984
Conversation
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Nice simplification!
@@ -1337,7 +1337,7 @@ template <typename T> decimal_fp<T> to_decimal(T x) noexcept { | |||
|
|||
if (r < deltai) { | |||
// Exclude the right endpoint if necessary. | |||
if (r == 0 && z_mul.is_integer && !include_right_endpoint) { | |||
if (r == 0 && (z_mul.is_integer & !include_right_endpoint)) { |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Why switch to bitwise AND? Better codegen?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yes. With logical AND, it generates three test
instructions and three conditional branches. With bitwise AND, it generates two test
instructions and two conditional branches. I thought compilers are aware of how to optimize this when all the things are just booleans variables (thus no short-circuit is really involved), but apparently they don't bother doing so. I am not sure why...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I guess, compilers are not sure if all boolean variables involved are really just 0 or 1. Like, if one is 2
and another is 1
, they are both true
so &&
should give true
, but &
will give 0
. It is wrong from the first place to use bool
for things with more than two states, but maybe the current standard allows bool
variables to hold nonsensical values without UB.
static const int divisibility_check_by_5_threshold = 39; | ||
static const int case_fc_pm_half_lower_threshold = -1; |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Should these be removed from float_info<double>
as well?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Ha. Another careless mistake. Sorry about that, it's fixed now.
Merged, thanks! |
- Update from version 9.0.0 to 9.1.0 - Update of rootfile - Changelog 9.1.0 - 2022-08-27 * ``fmt::formatted_size`` now works at compile time `#3026 <https://github.com/fmtlib/fmt/pull/3026>`_ For example (`godbolt <https://godbolt.org/z/1MW5rMdf8>`__): .. code:: c++ #include <fmt/compile.h> int main() { using namespace fmt::literals; constexpr size_t n = fmt::formatted_size("{}"_cf, 42); fmt::print("{}\n", n); // prints 2 } * Fixed handling of invalid UTF-8 `#3038 <https://github.com/fmtlib/fmt/pull/3038>`_, `#3044 <https://github.com/fmtlib/fmt/pull/3044>`_, `#3056 <https://github.com/fmtlib/fmt/pull/3056>`_ * Improved Unicode support in ``ostream`` overloads of ``print`` `#2994 <https://github.com/fmtlib/fmt/pull/2994>`_, `#3001 <https://github.com/fmtlib/fmt/pull/3001>`_, `#3025 <https://github.com/fmtlib/fmt/pull/3025>`_ * Fixed handling of the sign specifier in localized formatting on systems with 32-bit ``wchar_t`` `#3041 <https://github.com/fmtlib/fmt/issues/3041>`_). * Added support for wide streams to ``fmt::streamed`` `#2994 <https://github.com/fmtlib/fmt/pull/2994>`_ * Added the ``n`` specifier that disables the output of delimiters when formatting ranges `#2981 <https://github.com/fmtlib/fmt/pull/2981>`_, `#2983 <https://github.com/fmtlib/fmt/pull/2983>`_ For example (`godbolt <https://godbolt.org/z/roKqGdj8c>`__): .. code:: c++ #include <fmt/ranges.h> #include <vector> int main() { auto v = std::vector{1, 2, 3}; fmt::print("{:n}\n", v); // prints 1, 2, 3 } * Worked around problematic ``std::string_view`` constructors introduced in C++23 `#3030 <https://github.com/fmtlib/fmt/issues/3030>`_, `#3050 <https://github.com/fmtlib/fmt/issues/3050>`_ * Improve handling (exclusion) of recursive ranges `#2968 <https://github.com/fmtlib/fmt/issues/2968>`_, `#2974 <https://github.com/fmtlib/fmt/pull/2974>`_ * Improved error reporting in format string compilation `#3055 <https://github.com/fmtlib/fmt/issues/3055>`_ * Improved the implementation of `Dragonbox <https://github.com/jk-jeon/dragonbox>`_, the algorithm used for the default floating-point formatting `#2984 <https://github.com/fmtlib/fmt/pull/2984>`_ * Fixed issues with floating-point formatting on exotic platforms. * Improved the implementation of chrono formatting `#3010 <https://github.com/fmtlib/fmt/pull/3010>`_ * Improved documentation `#2966 <https://github.com/fmtlib/fmt/pull/2966>`_, `#3009 <https://github.com/fmtlib/fmt/pull/3009>`_, `#3020 <https://github.com/fmtlib/fmt/issues/3020>`_, `#3037 <https://github.com/fmtlib/fmt/pull/3037>`_ * Improved build configuration `#2991 <https://github.com/fmtlib/fmt/pull/2991>`_, `#2995 <https://github.com/fmtlib/fmt/pull/2995>`_, `#3004 <https://github.com/fmtlib/fmt/issues/3004>`_, `#3007 <https://github.com/fmtlib/fmt/pull/3007>`_, `#3040 <https://github.com/fmtlib/fmt/pull/3040>`_ * Fixed various warnings and compilation issues `#2969 <https://github.com/fmtlib/fmt/issues/2969>`_, `#2971 <https://github.com/fmtlib/fmt/pull/2971>`_, `#2975 <https://github.com/fmtlib/fmt/issues/2975>`_, `#2982 <https://github.com/fmtlib/fmt/pull/2982>`_, `#2985 <https://github.com/fmtlib/fmt/pull/2985>`_, `#2988 <https://github.com/fmtlib/fmt/issues/2988>`_, `#3000 <https://github.com/fmtlib/fmt/issues/3000>`_, `#3006 <https://github.com/fmtlib/fmt/issues/3006>`_, `#3014 <https://github.com/fmtlib/fmt/issues/3014>`_, `#3015 <https://github.com/fmtlib/fmt/issues/3015>`_, `#3021 <https://github.com/fmtlib/fmt/pull/3021>`_, `#3023 <https://github.com/fmtlib/fmt/issues/3023>`_, `#3024 <https://github.com/fmtlib/fmt/pull/3024>`_, `#3029 <https://github.com/fmtlib/fmt/pull/3029>`_, `#3043 <https://github.com/fmtlib/fmt/pull/3043>`_, `#3052 <https://github.com/fmtlib/fmt/issues/3052>`_, `#3053 <https://github.com/fmtlib/fmt/pull/3053>`_, `#3054 <https://github.com/fmtlib/fmt/pull/3054>`_ Signed-off-by: Adolf Belka <adolf.belka@ipfire.org> Reviewed-by: Michael Tremer <michael.tremer@ipfire.org>
jk-jeon/dragonbox#31 pointed out that I was doing something silly. I corrected that, which simplifies the source code a little bit (and the generated assembly as well), while the flow of the logic is not affected. Also, I replaced several logical operators into bitwise operators which seems to improve the code generation a little bit.