mirrors/micropython
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

windows/Makefile: Fix float exact int formatting on 32-bit mingw.

Browse Source 
When compiler optimizations are enabled on the mingw version of gcc, we are
getting failing tests because of rounding issues, for example:

    print(float("1e24"))

would print

    9.999999999999999e+23

instead of

    1e+24

It turns out special compiler options are needed to get GCC to use the SSE
instruction set instead of the 387 coprocessor (which uses 80-bit precision
internall).

Signed-off-by: David Lechner <david@pybricks.com>
This commit is contained in:
David Lechner 2022-12-18 18:26:32 -06:00 committed by Damien George
parent ac8e7f7b67
commit 23342eff90
1 changed files with 9 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
ports/windows/Makefile
View File

@ -86,6 +86,15 @@ ifneq ($(FROZEN_MANIFEST),)
CFLAGS += -DMPZ_DIG_SIZE=16 CFLAGS += -DMPZ_DIG_SIZE=16
endif endif
ifeq ($(shell $(CC) -dumpmachine),i686-w64-mingw32)
# GCC disables the SSE instruction set by default on i366 targets and we have
# to specify all three of these options to enable it.
# https://gcc.gnu.org/onlinedocs/gcc/x86-Options.html (see -mfpmath=unit section)
# Enabling the SSE instruction set is necessary to get correct rounding of floating points.
# https://lemire.me/blog/2020/06/26/gcc-not-nearest
CFLAGS += -msse -mfpmath=sse -march=pentium4
endif
CXXFLAGS += $(filter-out -std=gnu99,$(CFLAGS)) CXXFLAGS += $(filter-out -std=gnu99,$(CFLAGS))
include $(TOP)/py/mkrules.mk include $(TOP)/py/mkrules.mk