You think five minutes is a long time to wait for tests to complete!  Try several hours.  I had motivation for the following.
Using Boost threads, CppUnit threading is pretty easy.  CppUnit already has some hooks for synchronization so the following should make it thread safe:
class Mutex : public CPPUNIT_NS::SynchronizedObject::SynchronizationObject
{
public:
    void lock() { this->mutex->lock(); }
    void unlock() { this->mutex->unlock(); }
private:
    boost::mutex mutex; 
};
With this, you can modify your test runner to make your TestResult thread safe.  Just write something like CPPUNIT_NS::TestResult testResult(new Mutex);.  Now here's a threaded test suite:
class TestSuiteThreaded : public CPPUNIT_NS::TestSuite
{
public:
    TestSuiteThreaded(std::string name = "", int nThreads = 0)
        : TestSuite(name)
        , nThreads(nThreads ? nThreads : boost::thread::hardware_concurrency())
    {
    }
    void doRunChildTests(CPPUNIT_NS::TestResult *controller)
    {
        ThreadPool pool(this->nThreads);
        for (int i=0; i < getChildTestCount(); ++i)
        {
            pool.add(
                boost::bind(threadFunction, getChildTestAt(i)
                , controller));
        }
    }
private:
    static void threadFunction(
        CPPUNIT_NS::Test *test, 
        CPPUNIT_NS::TestResult *controller)
    {
        test->run(controller);
    }
    const int nThreads;
};
You may well need a macro for easy use of the threaded test suite.  You should be able to use TestSuiteThreaded suite either as a top level suite or a suite comprising multiple methods of the same text fixture.  Here's how you do the latter - put this in place of CPPUNIT_TEST_SUITE_END.  Some of this is pasted from CppUnit so please respect the license:
#define CPPUNIT_TEST_SUITE_END_THREADED(n)                                     \
    }                                                                          \
    static CPPUNIT_NS::TestSuite *suite()                                      \
    {                                                                          \
      const CPPUNIT_NS::TestNamer &namer = getTestNamer__();                   \
      std::auto_ptr<CPPUNIT_NS::TestSuite> suite(                              \
         new CPPUNIT_NS::TestSuiteThreaded( namer.getFixtureName(), n));       \
      CPPUNIT_NS::ConcretTestFixtureFactory<TestFixtureType> factory;          \
      CPPUNIT_NS::TestSuiteBuilderContextBase context( *suite.get(),           \
                               namer,                                          \
                               factory );                                      \
      TestFixtureType::addTestsToSuite( context );                             \
      return suite.release();                                                  \
    }                                                                          \
  private: /* dummy typedef so that the macro can still end with ';'*/         \
    typedef int CppUnitDummyTypedefForSemiColonEnding__
Now there is the small matter of a ThreadPool.  I tried using various publicly available ones with no success.  My company has one but I'm unable to publish it here.  So roll your own - thread pools are pretty easy and fun to make, with help from Boost.  Here is the interface expected by TestSuiteThreaded:
class ThreadPool
{
public:
    // Create thread pool, launching n worker threads
    ThreadPool(unsigned n); 
    // Join all worker threads and clean up
    ~ThreadPool();
    // You can have add() do one of two things.  Both will work:
    // Either: push a new task to the back of the threadpool's work queue
    // Or: block until a worker is free then assign task to that thread
    void add(boost::function0<void> task);
};
I leave this as an exercise for the reader.  Have fun!