I've never actually measured input lag myself, so take the following with a grain or three of salt. From what I understand it unfortunately seems rather difficult to get precise measurements. I'd imagine others on this board would know much more about getting good measurements than I do. With that out of the way...
It would be best to compare with a display that's known to have little or no lag, like many (all?) CRTs.
The least precise way would just be to compare how both feel. Obviously this is totally subjective! But it's trivial to do, and large differences in latency will be pretty apparent to anyone at all sensitive to lag. (This was actually all I was expecting when I asked you by name, fagin... on rereading, I guess it looks a bit like I was asking you for measurements, when I instead intended to ask the separate questions "does anyone have precise numbers" and "fagin could you eyeball it against your nearly lag-free scaler, since you have both". Sorry if I sounded demanding... But if you could be bothered to try for something more precise that would be great!
For a bit more accuracy, you could try the 240p test suite
. This is another thing I haven't tried myself, but according to the wiki page there's a lag test that beeps in time with a visual cue. If you try this, don't route the sound through the Edge, as I've read that it delays sound to match its current picture delay. You could try recording the test with a video camera, and see how far apart the frames with the visual and aural cues are... though this presumes your camera's video and audio are perfectly in sync, and even then the results would have error bounds based on your camera's framerate (lower = less accurate). If you're really accurate at timing button presses you can try to press it in time with the cues, and the test will display an average, but this is again subjective (but less so if you have good timing).
Some rhythm games have lag testers. Rock Band 2 can measure lag using the guitar and microphone with no user input (source
), though I don't know how accurate it is.
Many consoles can display the same picture on several outputs simultaneously, though often this would mean modding a video cable so that the pins for multiple outputs are connected. You could try (for instance) running svideo to the Edge, and RGB to the CGA to VGA, from the same console at once, to two displays, and see if one lags more than the other. This assumes that both outputs on the console are equally laggy (probably true), and that both displays are equally laggy (less likely). Again, you could record this and try to count how many frames one is behind the other, but the same concern about camera framerate applies (though you wouldn't need sound in this case).
Benjamin Heckendorn has created a few modified controllers
that light up LEDs when various buttons are pressed. The idea is that you press a button and use a high speed camera to count how many frames pass between the LED light-up and an on-screen reaction. The same disclaimer about camera framerate applies. You likely wouldn't want to buy one of Ben's for this purpose as they're quite expensive and only produced for consoles that (I believe) don't support 240p. So I guess you'd have to mod a controller for this, though you'd only really need one button for testing scaler lag...
Any tests you're willing to do would be appreciated. I'm trying to decide on a scaler; I don't think I'd mind the video quality of the CGA to VGA as long as I pair it up with an SLG3000, but if it has high latency it would be a deal breaker for me.