It's several things. Desktops dissipate more heat, so sustained clocks are possible. Max Turbo is theoretical in most laptops. A decent example of this phenomenon is the m5 vs m7. Though the much is theoretical faster by double digits, in practice the clocks that those systems can sustain are identical (so save $100 by getting the m5).
Let's compare a 8550u and 8700k. The desktop has a base clock of 3.7 instead of 1.8. it has 12mb cache instead of 8mb. Twice the bus speed. Faster supported memory and more bandwidth. Desktop also has 16 pcie Lanes vs 4 (both can access more via the chipset, but then there's that bus bandwidth issue). Instruction and feature support is almost identical (except some things like vpro).
Sustained clocks, bandwidth, and more cache make a huge difference.
Very true, trying to evaluate laptop CPU performance is highly deceiving. Unlike desktop parts the actual performance depends completely on the particular laptop's cooling capability, which is impossible to know from looking at the spec. And often that cooling is terrible.
That's why desktop high end CPUs have such a large gap in performance in desktop vs. mobile. The gap seems small if you just look at the specs, but is much larger in practice.
> I'm guessing that the mobile version does have less execution units.
They don't. The architecture of their mobile CPUs is exactly the same as the desktop versions. (They use the same die.) The difference is in the thermal limits, which determine how long the CPU can spend at each power state.
Another huge difference can be whether the machine has a memory configuration that runs in dual-channel mode or not, as well as differences in cache sizes and BIOS settings for things like cache pre-fetching.
TurboBoost on laptops has always been poop. Take the 3720qm and 3940xm - yeah the latter has a higher TB, but under high load it runs very close to base clock, while the 3720qm runs a bit higher than base... resulting in a minimal difference :/
That's disregarding TDP/TPL adjustments/overclocking, of course (which is disabled by BIOS on workstations anyway).
yes I used to have an older i7 4 core desktop and a newer i7 4 core laptop, and by reading the spec sheets you would think the laptop should be as fast or faster, but the desktop was much faster.
Every benchmark i know from intel desktop cpu versus mobile cpu is the same: desktop wins by big margins.
I'm guessing that the mobile version does have less execution units.