I suppose we all know by now that the 2018 MBP has problems with heat dissipation. Since I use my MBP mostly on my desk in clamshell mode running to my external monitor, I started feeling slightly reduced performance compared to when I was using it by itself so I decided to measure the difference using DXO Photolab and exporting five 42 megapixel images with PRIME noise reduction (a heavy test on the CPU, requiring roughly a full minute per file to export) to peg the CPU at full load until the export completed
Summary of results in the attached image but for my use case:
Slightly lower initial boost clock
Lower sustained clock (dropping even below base clock after a while)
Reduced wattage drawn from the CPU and consequently lower sustained temperatures in clamshell mode (maybe because the dGPU is active with an external monitor?)
An 8% increase in export times using clamshell mode
Performance above base clock is therefore not guaranteed even with the i7 2.6 with maxed CPU-only loads when in clamshell mode
I did not expect that high of a performance hit, so I thought I'd share it here. In real world use there is a small but noticable difference in snappiness and overall speed using the laptop as-is compared to in clamshell mode with an external display, particularly in high workloads
For the sake of completeness, I decided to run the same test with the screen open while connected to my external monitor. Giving it room to breathe does next to nothing: the same export finishes in 4m25s, a whole second faster than when in clamshell mode. Clocks also drop below base clock after a while, just like it did in clamshell mode.
Summary of results in the attached image but for my use case:
Slightly lower initial boost clock
Lower sustained clock (dropping even below base clock after a while)
Reduced wattage drawn from the CPU and consequently lower sustained temperatures in clamshell mode (maybe because the dGPU is active with an external monitor?)
An 8% increase in export times using clamshell mode
Performance above base clock is therefore not guaranteed even with the i7 2.6 with maxed CPU-only loads when in clamshell mode

I did not expect that high of a performance hit, so I thought I'd share it here. In real world use there is a small but noticable difference in snappiness and overall speed using the laptop as-is compared to in clamshell mode with an external display, particularly in high workloads
For the sake of completeness, I decided to run the same test with the screen open while connected to my external monitor. Giving it room to breathe does next to nothing: the same export finishes in 4m25s, a whole second faster than when in clamshell mode. Clocks also drop below base clock after a while, just like it did in clamshell mode.
Last edited: