RAM latency

RAM Latency is literally the amount of wait time that a computer experiences when trying to access data in RAM in any given computer. RAM latency is measured in memory bus clock cycles. Due to the fact that a modern CPU is much faster than the system's RAM access time, and that the CPU often has to wait for memory accesses to complete before it can make further progress RAM latency is a significant bottleneck for system performance.

RAM access

RAM is notionally organized into a grid like pattern, with "rows", and "columns". The data stored on RAM comes in blocks, defined by the coordinates of the row, and column of the specific information. RAM access has 4 main measurements (quantified in memory bus clock cycles) important in defining the RAM latency in a given computer:

• RAS: Row Address Select = (time needed to access certain row of RAM data / memory bus clock speed) = RAS, in clock cycles
• tRCD : (Row address to Column address Delay) Clock cycles between RAS and CAS (time between the computer defining the first and second coordinate of the given memory block)
• CAS: Column Address Select = (time needed to access certain column of RAM data / memory bus clock speed) = CAS, in clock cycles
• tRP: Row Precharge time = Time given to the RAM to close an open memory block, and put it in the next clock cycle for the CPU to have access to.

For the CPU to access any data in RAM, it goes through these steps:

1. defines which row the requested data is in (RAS)
2. gives the RAM time to access requested row (tRCD)
3. defines the column of requested data (CAS)
4. waits for the RAM to send requested data to the CPU
5. precharges the row to recycle it for subsequent use (tRP)

Measurements

As with almost all latency issues, the lower, the better. When picking out RAM to buy, these four measurements are only important in high performance computers. RAM speeds are given by the four numbers above, for example, "2-2-2-5" is a fast speed as of 2005. The numbers go accordingly to the list above: RAS=2, tRCD=2, CAS=2, tRP=5.

Most computer users don't need to worry about RAM latency, because the computer can handle the auto-adjustment to RAM timing based on a Read Only Memory (the Serial Presence Detect , SPD) inside the RAM packaging that defines the four timing values, decided by the RAM manufacturer. Although the RAM latency timing can be adjusted manually, the values given by the RAM manufacturer should not be exceeded, as it can cause your computer to crash, overheat, or not start up at all.

See also

• Dynamic random access memory

External links

• http://www.ocfaq.com/article.php/overclocking/63
• http://www.sysopt.com/articles/latency/