[libc-commits] [PATCH] D136495: [libc] add performance options for string to float

[Michael Jones via Phabricator via libc-commits]

michaelrj marked an inline comment as done.
michaelrj added inline comments.


================
Comment at: libc/src/__support/str_to_float.h:214
 
-  uint32_t exp2 = static_cast<uint32_t>(exp10_to_exp2(exp10)) + BITS_IN_MANTISSA +
+  uint32_t exp2 = static_cast<uint32_t>(exp10_to_exp2(exp10)) +
+                  BITS_IN_MANTISSA +
----------------
sivachandra wrote:
> Why did this and the above changes, which look like just formatting changes, happen?
Previous patches must've forgotten to format the file. I didn't touch these manually, they came from running `clang-format`.


================
Comment at: libc/src/__support/str_to_float.h:610
 
+#ifndef LLVM_LIBC_DISABLE_SIMPLE_DECIMAL_CONVERSION
   simple_decimal_conversion<T>(numStart, outputMantissa, outputExp2);
----------------
sivachandra wrote:
> Why should this be a conditional also?
It's mostly there for my own performance testing, but could also be used for cases where speed is preferred over accuracy.


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D136495/new/

https://reviews.llvm.org/D136495