Thanks to Daftmike, Marshallh, and Zeturi.
With an 18°C (64°F) ambient temp, running the 3.3V line at 3.05V, after 45 mins of 8 bot perfect dark multiplayer without heatsinks, in free air, my rcp was at 64°C (147°F), the memory at 60°C (140°F), and the cpu at 49°C (120°F).
I didn't test until the system shut down but the temps had stabilised after 45 mins, in a typical n64p volume case, you are looking at at least 20°C (68°F) above ambient and I would recommend keeping the package temps below 85°C (185°F).
When I did my testing there were no crashes for four hours with a slight under-volt (3.05V) and no heatsinks, the temperatures levelled out after 45 mins.
Theoretically to keep the chips under 80°c (176°F), inside a case, I estimate your heatsinks would need only about 12°C (54°F)/W thermal resistance. This is relatively modest, most pc ramsinks should be more than suitable with a little airflow.
My dev 64 has been run for hours at a time with no heatsink, just the aluminum blocks on each chip. They get pretty hot but the N64 never shut down or started to crash.
I took a record of the heat coming off the n64 for a couple of hours:
3:38 PM started test,
5:13 P.M. Ended test,
Looks like my temps won't climb beyond these last recorded ones. (Though possible +20 more degrees while in the case.)[img]http://i636.photobucket.com/albums/uu90/zeturi/GEDC0015.jpg[/img
I do have another n64, I hooked it up to my TV and let it run without heat sinks. I got bored after 20 minutes, because the temps never got high, really. Here's what I got:
Granted, it only ran for 20 minutes, but the numbers didn't climb any higher than these. Usually they were about 4 degrees lower. The heat is almost the same as when I had heat sinks! Though the heat might climb a bit more over time, with heat sinks like what I used, I doubt temperatures would increase much. (But will increase once inside a case.) The revision of the board I just tested is 3, and the revision that I've already messed with is a Revision 8. With a small fan installed, there would be little to no heat.
Graph of Zeturi's Test: