I'd second replacing the MS GS synth with TTS-1. As SuperG says, the MS synth is a kind of software fallback for consumer/gaming soundcards without built-in general MIDI chips.  It's intended for media player or web browser playback of general MIDI files and to provide sound in games that expect to find a "built in" synth in a PC and so use it.

High quality it isn't.

Using it in combination with soft-synths will cause problems as it will have a different latency and response time to the synths in Sonar, and Sonar can't compensate automatically for that.

It might be possible to get it to sync by experimenting to find the time difference between the MS synth and Sonar's synths then shifting the MS synth's MIDI notes that far backwards, but to be honest it's not worth it.

I use mostly hardware synths and run into a similar issue when recording the audio output of one or two as the synths have different response times, though in my case this is usually something like a 3-5ms spread so isn't really noticeable at all. If necessary (or I'm feeling very particular) I just shift the recorded audio the required number of milliseconds.

TTS-1 is not only better behaved in a DAW but it sounds better and offers far more controls as well. It's also straightforward to convert the TTS-1 track to audio, while the MS synth's output will, I suspect, have to be recorded as audio by looping the output of the soundcard back to the input one way or another.