How and where does clang use llvm?

When using command like this:

clang -### -O3 -a.c

And then, it will output the followings:

clang version 1.1 (branches/release_27)Target: x86_64-unknown-linux-gnuThread model: posix"/tensorflow/bin/llvm+clang-2.7/bin/clang""-cc1""-triple""x86_64-unknown-linux-gnu""-S""-disable-free""-main-file-name""a.c""-mrelocation-model""static""-mdisable-fp-elim""-mconstructor-aliases""-munwind-tables""-target-cpu""x86-64""-resource-dir""/tensorflow/bin/llvm+clang-2.7/lib/clang/1.1""-O3""-fmessage-length""141""-fgnu-runtime""-fdiagnostics-show-option""-fcolor-diagnostics""-o""/tmp/cc-w9Hpfd.s""-x""c""a.c""/usr/bin/gcc""-O3""-c""-m64""-o""/tmp/cc-8haDzc.o""-x""assembler""/tmp/cc-w9Hpfd.s""/usr/bin/gcc""-O3""-m64""-o""a.out""/tmp/cc-8haDzc.o"

So, we can see that the early version of llvm(llvm-2.7) uses gcc's assembler and linker.

But how, when and where, clang uses llvm?

Does clang use llvm in clang -cc1?

How to prove it? (Either documentation or command line output can be the evidence. But I can not find it.)

I think llvm+clang is a flexible toochain, I can find many ways to generate an executable from a C program.

Thanks in advance.