bandwidth - What drawbacks are there when using memory that is faster than what is rated by the motherboard?

With the Core i7 platform, memory speeds of up to 1600MHz are officially supported, but kits are sold with speeds of up to 2133MHz and beyond.

Are the caveats/drawbacks of using faster-than-rated memory the same as overclocking, or is there a specific technical reason why the faster memory is unsupported or could cause problems?

Edit: I am referring to running the memory at the advertised module speed (2133MHz, for example) on a platform that doesn't "officially" support it. I know that the module would downclock by default and run at the normal supported bus speed without any intervention.

In servers, I realize that keeping the bandwidth in check is a critical issue, especially with dozens of DIMMs and hundreds of GBs flowing around. Perhaps these are related?

Answer

The 1600 MHz limit is generally imposed by the CPU, not the motherboard.¹ Most motherboards support higher clocks.

If your motherboard supports clocks higher than 1600 MHz, you'll be able to overclock the CPU's memory bus. This usually goes well. I'm using 1866 MHz RAM myself with no instability whatsever.

If your motherboard doesn't support clocks higher than 1600 MHz, any memory you install will simply not be able to run at higher speeds.

---

¹ Traditionally, the opposite was true, as the memory controller was part of the northbridge (motherboard). However, Core i7 processors have the memory controller inside the CPU.