There really doesn't seem to be one. I've ran 200-300 simultaneous connections with no issue. You can set the number as high as you like, and it will try to run that many instances of your script. The main limiter then is the hardware running the machine, and any limits on your network regarding how many simultaneous connections you can spin off (or have open) at a time. When I get to 500, I know my computer starts having noticeable issues. FYI, I belive that XP SP2 has a cap, preventing you from spawning more than 10 socket connections at a given time. I use 2k in the office.
The stress on the client (not server) increases as the number of browser connections increase.
U can easily check out the processor and memory utilisation of ur client comp when u hav a high number of simaltaneous browser conn. This could lead to a lot of socket errors after a particular point. So the limit on number of simaltaneus brosers depends on the superiority of the computer from which u r running ur tests.
Generally the thing that i have observed is it depends upon the request that you are sending if the request takes more amount of time and the response that is received is also more then even with few number of simultaneous browser connections the complete local resources would be utilized and it would not in a great way effect the results ( BUT THERE WOULD BE A SLIGHTLY VARIATION IN THE RESULTS) that are given by the ACT. The point that i would like to make here is it is not fixed that for 80 simultaneous connections the results would be correct or the hardware supports only 80 simultaneous connection it even depends upon the environment and the operation systems ,the newtwork everything needs to be taken into consideration.