The L1 cache is always embedded into the processor die and operates at the processor's full-core speed internally. By full-core speed, I mean that this cache operates at the internal processor's higher clock multiplied speed rather than the external motherboard's speed. This cache is a quick memory section built into the CPU that stores some of the processor's current working set of code and data. Because cache memory runs at the same speed as the CPU core, it can be reached with no wait states.

Because system RAM is nearly always much slower than the CPU, using cache memory relieves a conventional system bottleneck; the performance gap between memory and CPU speed has grown especially big in recent systems. Cache memory eliminates the need for the CPU to wait for code and data from the much slower main memory, resulting in improved speed. A CPU would frequently be forced to wait for system memory to catch up if it didn't have the L1 cache.

In current processors, the cache is even more important since it is sometimes the only memory in the system that can keep up with the chip. The majority of modern processors are clock multiplied, which means they operate at a rate that is a multiple of the motherboard into which they are connected. The L1, L2, and L3 caches integrated into the CPU core are the only forms of memory that match the processor's maximum speed.

The CPU does not have to wait if the data it requires is already in the L1 cache. If the data isn't in the cache, the CPU must get it from the Level 2 or Level 3 caches, or (in less complex system architectures) directly from the system bus—that is, main memory.

Download