Apparently, we set COMPILER(GCC) on clang, so we are at least feeding this information to the clang compiler. COMPILER(GCC) is a confusing name for a macro that also include clang. :(
Adam On Mon, Oct 1, 2012 at 11:11 PM, Ryosuke Niwa <[email protected]> wrote: > I don't think so. I wrote a code that can be 10% slower without UNLIKELY > several weeks ago: > http://trac.webkit.org/browser/trunk/Source/WebCore/html/HTMLCollection.cpp#L289 > > On Mon, Oct 1, 2012 at 10:55 PM, Adam Barth <[email protected]> wrote: >> >> According to >> <http://stackoverflow.com/questions/1851299/is-it-possible-to-tell-the-branch-predictor-how-likely-it-is-to-follow-the-branc>, >> __builtin_expect (what our LIKELY and UNLIKELY macros expand to [1]) >> doesn't do anything on modern CPUs. Apparently, these used to be >> important for PowerPC, but I don't think many folks use WebKit on >> PowerPC anymore. >> >> Should we remove these macros? I wasted some time today experimenting >> with them without realizing that they compile to no-ops in clang. >> >> Adam >> >> [1] http://trac.webkit.org/browser/trunk/Source/WTF/wtf/Compiler.h#L188 >> _______________________________________________ >> webkit-dev mailing list >> [email protected] >> http://lists.webkit.org/mailman/listinfo/webkit-dev > > _______________________________________________ webkit-dev mailing list [email protected] http://lists.webkit.org/mailman/listinfo/webkit-dev
