"tools_def.template" currently suggests, in the documentation of the LzmaF86Compress utility, that said tool is generally unhelpful on binaries built with the GCC44 toolchain, relative to LzmaCompress.
This statement doesn't apply to the GCC48 toolchain. I compressed 126 NOOPT_GCC48/IA32 unique EFI modules (built with gcc-4.8.5, as part of OVMF) with both LzmaCompress and LzmaF86Compress. I repeated the same for 117 NOOPT_GCC48/X64 unique EFI modules. On average, the LzmaF86Compress output size was 92.4% of the LzmaCompress output size in the IA32 case (best relative compression: 86.01%, poorest relative compression: 97.47% -- still a win). In the X64 case, the LzmaF86Compress output size was 92.95% of the LzmaCompress output size, on avarege (best relative compression: 87.69%, poorest relative compression: 97.65% -- again, still a win). Given the consistent improvement from LzmaCompress to LzmaF86Compress, remove the statement (rather than updating it to GCC48). Cc: Bob Feng <[email protected]> Cc: Liming Gao <[email protected]> Cc: Yonghong Zhu <[email protected]> Ref: https://bugzilla.tianocore.org/show_bug.cgi?id=1377 Contributed-under: TianoCore Contribution Agreement 1.1 Signed-off-by: Laszlo Ersek <[email protected]> --- BaseTools/Conf/tools_def.template | 1 - 1 file changed, 1 deletion(-) diff --git a/BaseTools/Conf/tools_def.template b/BaseTools/Conf/tools_def.template index d1324d56a1a4..5afbb4ff2743 100755 --- a/BaseTools/Conf/tools_def.template +++ b/BaseTools/Conf/tools_def.template @@ -6588,7 +6588,6 @@ RELEASE_RVCTCYGWIN_ARM_CC_FLAGS = "$(CCPATH_FLAG)" $(ARCHCC_FLAGS) $(PLATFORM_F ################## # LzmaF86Compress tool definitions with converter for x86 code. # It can improve the compression ratio if the input file is IA32 or X64 PE image. -# Notes: If X64 PE image is built based on GCC44, it may not get the better compression. ################## *_*_*_LZMAF86_PATH = LzmaF86Compress *_*_*_LZMAF86_GUID = D42AE6BD-1352-4bfb-909A-CA72A6EAE889 -- 2.19.1.3.g30247aa5d201 _______________________________________________ edk2-devel mailing list [email protected] https://lists.01.org/mailman/listinfo/edk2-devel
