In LabVIEW RT 6.1 and earlier, there's a shipping example entitled "Array Initialization Benchmark.vi", which uses counters to timestamp sample code, generating an execution histogram. The histogram reveals the level of jitter inherent in the sample code.
I'd recommend measuring the jitter of individual pieces of your application. First, make a copy of the benchmark example and replace the counter VIs with Get Tick Count. Then stick FP Write.vi (by itself) inside the loop with some bogus value wired to it. The execution histogram will be accurate to within milliseconds and will tell you whether FP Write.vi alone is responsible for the 10+ms jitter.
You might have to try a few combinations before you find the "problem" VI. If in fact the problem VI is
something out of vi.lib or one of our examples, your next step should be to e-mail or call an RT applications engineer with your data.