Synchronous DRAM, commonly known as SyncDRAM, which is the latest type of PC memory architecture. This provides a significant performance improvement over conventional Fast Page Mode (FPM) or Extended Data Output (EDO). A key difference found in SyncDRAM is that it is synchronized to the system clock signal, which means that all operations run at the same speed as the processor bus, i.e. 60 or 66MHz. This is compared to conventional memory architectures which run asynchronously to the processor bus. Due to this synchronization, data bits are retrieved faster and thereby increase system performance.

PC applications commonly access DRAM data in 4-quad word (Qword) burst lengths. Optimum system performance is achieved when data is fed to the processor on par with the system clock. Data access from an L2 cache comes close to meeting this requirement as typical only one or two wait states are inserted while accessing the first data. However, the remaining three Qwords are delivered on a par with the processor speed, denoted as a 2/1/1/1 burst rate. When, however, the requested data must come directly from DRAM, the burst rate is degraded. Assuming the burst is from a DRAM page that is already open (page hit), the achievable burst rate at 66MHz using 60ns FPM DRAMs is 5/3/3/3 cycles; for EDO DRAMs it is 6/2/2/2. SyncDRAMs can reduce burst rates to 7/1/1/1 for L2 cache misses. While SyncDRAM takes more time to locate the first data bit, it then is faster at retrieving the next three Qwords. Furthermore, it really excels at back to back reads as noted in the following comparison which assumes a page hit:

Tests employing the Winstone 32 system level benchmark show SyncDRAM-based systems outperforming comparable EDO-based systems by about 15% for configurations without an L2 cache. Even in systems equipped ith a 512K cache, SyncDRAM provides a performance benefit of 5%. These results confirm that consumers will benefit from a performance increase in any application when using a SyncDRAM-based system.