For C++? No.
Too hard of a language to waste phd students maintaining :)
Over the years there have been a few more (openC++ is a good example, ROOT used to be one too), but they are all dead now AFAIK.
One reason is that research in compilers is pretty much not in the frontends anymore. For any research still being done that uses C++ as a base, gcc/clang/llvm pretty much work fine. LLVM in particular has an IR that works for most researchers, is not hard to understand, etc.
For people who want to try to build larger solutions, they usually start with C (see, for example, libfirm and friends).
Commercial folks use EDG or, actually, a lot are also starting to use clang now as a frontend.
(IBM was the other major company that had their own C++ parser for a longer time)
Bit of a side-note, but I would disagree about ROOT/CINT ever being a C++ compiler. It was an interpreter for a language that looked somewhat related to C++, if you squint quite a bit.
* All variables were hoisted up to function scope. This meant that you couldn't use "int i" in one loop, and "unsigned int i" in the next. This also caused destructors to be incorrectly delayed until the end of the function call.
* Use of templates required pre-compiled dictionaries for each type the template might be instantiated for. Any templates occurring in interpreted code would be silently ignored.
* Incomplete standard library implementation. std::abs(long) is missing, for example.
* const is silently ignored.
Not entirely relevant to the current discussion, but good heavens, I am glad that monstrosity is gone.
For C++? No. Too hard of a language to waste phd students maintaining :)
Over the years there have been a few more (openC++ is a good example, ROOT used to be one too), but they are all dead now AFAIK. One reason is that research in compilers is pretty much not in the frontends anymore. For any research still being done that uses C++ as a base, gcc/clang/llvm pretty much work fine. LLVM in particular has an IR that works for most researchers, is not hard to understand, etc. For people who want to try to build larger solutions, they usually start with C (see, for example, libfirm and friends).
Commercial folks use EDG or, actually, a lot are also starting to use clang now as a frontend.
(IBM was the other major company that had their own C++ parser for a longer time)