You might want to give a try to the
RTL Utility from Oblique Audio. Although the DX1 does not have an input, you might want to aproximate its ASIO performance by using ASIO4ALL which allows to record the input using a different soundcard (e.g. the internal notebook soundcard) and calculate the total latency from sending the sound signal to the DX1 until it is recorded again from the internal sound card. Whatever you get using ASIO4ALL will be higher from what you will get using the DX1 native ASIO drivers (sometimes it is just a 4 ms difference). Since the RTL utility actually shows the total latency (ouput + input) and you are only interested in the output latency, halving that value might be another aproximation of the ouput latency. That assumes the output latency is similar to the input latency, which is not always the case.
20 ms seems to be the maximum most people playing the keyboard will accept, being less than 10ms a comfortable value. So if you manage to get , let's say, 14 ms of RTL using ASIO4ALL with a 128 samples buffer, it is very likely that you output latency using the native ASIO driver will be less than 10ms with that sample size.
GIG Performer has also a Round Trip latency measurements function plus an estimate of the output vs input latency (although I believe this estimated value just shows what the driver reports).