Well, I just tried an experiment. I exported 1000 RAW photos from Capture One Pro and had a watermark added onto them. I exported them with my 2018 MBP in the following modes:
- clamshell vertical with vents on bottom (minimally obstructed by stand)
- clamshell vertical with vents on top
- clamshell flat on desk
- clamshell flat on desk with lid slightly cracked
I ran the export and once the fans kicked on strong, I started changing positions with my MBP. My expectation would be that if any one mode had a heat penalty, I'd see the package temperature start bumping up, but I basically saw nothing conclusive. If any of the modes were better for heat than others, it was only better by 1 degree Celsius. Once the temperature stabilized, it ran at around 78 degrees +/- 2 degrees Celsius no matter how I oriented my 2018 MBP. It mostly stayed within 1 degree of 78.
Now it could be that the fans were working slightly harder in one mode than another and I couldn't tell the difference. That's totally possible, but that would mean that the fans were working well in all modes and any real difference would have been hard to distinguish... at least for this particular scenario. My job stressed the GPU much harder than the CPU. Maybe results would be different if someone pushed all 8 CPU cores for 10 minutes instead.