If you are letting your TV upscale the PS3 to 1080p, that would be causing input lag, hence the difference. The PS3 will not scale the image, it will only display in 720p, then your TV takes over, hence the lag. Whereas the 360 itself can scale to 1080p, which is highly reccomended over letting the TV do it. But for purposes of this test, you want to set it up like you would PS3.
HOWEVER, I could be wrong and it could be this..... What is your TV's native resolution? To find this out, hook your 360 through HDMI and look for the "optimal resolution" setting. It will tell you what your TV is.
Now if your saying the PS3 input is actually faster, then the latency on 360 could be explained by *720p render* *360 scale to 1080p* *TV scale to actual resolution*. Vs. your PS3 which would be something like *720p render, 720p output* *TV scale to TV res*. If this is the case, changing the 360 to *optimal resolution* will take out one of the scaling jobs out of the equation. Would explain everything.
Though if I'm right, then 360 @ optimal resolution might swing the pendulum the other way, resulting in faster input to PS3. So if you want them to match, I would say have 360 @720p, PS3 @720p.
In a nutshell, 360 @ optimal resolution = 1 scale, by the 360 itself, no lag. 360@720p = 1 scale, but by the TV itself (same as PS3), whereas 360 @ 1080p is 2 scales, 360 internally to 1080p, then your TV again to it's native resolution, resulting in the most latency of the 3.