Everything You Need To Know About DDR, DDR2 and DDR3 Memories
By Gabriel Torres on August 27, 2009
In this tutorial, we will explore the main technical differences between DDR, DDR2 and DDR3 memories. Enjoy!
Before we start going into the specifics, you need to know that DDR, DDR2, and DDR3 are based on SDRAM (Synchronous Dynamic Random Access Memory) design, meaning that they use a clock signal to synchronize things. DDR stands for Double Data Rate. Memories from this category transfer two data chunks per clock cycle. Translation: They achieve double the performance of memories without this feature running at the same clock rate (namely SDRAM memories, which are not available for PCs anymore).
Because of this feature, these memories are labeled with double the real maximum clock rate they can operate. For example, DDR2-800 memories work at 400 MHz, DDR2-1066 and DDR3-1066 memories work at 533 MHz, DDR3-1333 memories work at 666.6 MHz, and so on.
It is very important to understand that these clock rates are the maximum the memory can officially use. This does not, by any means, guarantee that the memory will work at those “speeds” automatically. For example, if you install DDR2-1066 memories on a computer that can only (or it is wrongly configured to) access the memory subsystem at 400 MHz (800 MHz DDR), the memories will be accessed at 400 MHz (800 MHz DDR) and not at 533 MHz (1,066 MHz DDR). This happens because the clock signal is provided by the memory controller, a circuit that is located outside the memory (in the north bridge chip from the motherboard or embedded inside the CPU, depending on the system).
This naming system DDRx-yyyy (where x is the technology generation and yyyy is the DDR clock rate), in theory, is used only for the memory chips. The memory modules – the little printed circuit boards to where the memory chips are soldered – use a different naming system: PCx-zzzz, where x is the technology generation and zzzz is the maximum theoretical transfer rate (a.k.a. maximum bandwidth). This number tells us how many bytes can be transferred per second between the memory controller and the memory module, assuming that data will be transferred on every single clock pulse. This math is easily done by multiplying the DDR clock in MHz by eight. (Actually, the real math is done by multiplying by 64 and then dividing by eight; since 64 / 8 = 8, we can simply multiply by eight to achieve the same result.) This will give us the maximum theoretical transfer rate in MB/s (megabytes per second). For example, DDR2-800 memories have a maximum theoretical transfer rate of 6,400 MB/s (800 x 8), and memory modules using this kind of memory are called PC2-6400. In some cases, the number is rounded off. For example, DDR3-1333 memories have a maximum theoretical transfer rate of 10,666 MB/s. Memory modules using this kind of memory are called PC3-10666 or PC3-10600, depending on the manufacturer.
It is really important to understand that these numbers are maximum theoretical numbers, and they are never reached. This occurs because for the math, we are assuming that the memory will be sending data to the memory controller every single clock cycle, which simply doesn’t happen. The memory controller and the memory need to exchange commands (for example, a command instructing the memory to deliver data stored at a given position), and during such time the memory won’t be transferring data.
Now that you know the basics about DDR memories, let’s talk about the specifics.
One of the main differences between DDR, DDR2 and DDR3 is the highest transfer rate each generation can reach. Below we list the most common speeds for each generation. Some manufacturers can deliver memory chips capable of achieving other speeds than those listed; for example, specialty memories targeted to overclockers. Clocks ending in 33 and 66 MHz are, in fact, periodic decimal expansions (33.3333 and 66.6666, respectively).
Maximum Theoretical Transfer Rate
DDR3 memories operate at lower voltages compared to DDR2 memories, which in turn operate at lower voltages compared to DDR memories. This means that DDR3 memories consume less power than DDR2 memories, which in turn consume less power than DDR memories.
Typically, DDR memories are fed with 2.5 V, DDR2 memories are fed with 1.8 V, and DDR3 memories are fed with 1.5 V (although modules requiring 1.6 V or 1.65 V are common, and chips requiring only 1.35 V may become common in the future). Some memory modules may require higher voltages than those listed. This happens especially with memories supporting the operation at clock rates higher than the official ones (i.e., memories targeted to overclocking).
Latency is the time the memory controller must wait between requesting data and the actual delivery of them. It is also known as CAS (Column Address Strobe) Latency or simply CL. This number is expressed in terms of clock cycles. For example, a memory with CL3 means that the memory controller must wait three clock cycles until data is delivered after a request is made. For a memory with CL5, the memory controller will have to wait longer: five clock cycles. So you should always look for the memory modules with the lowest latency possible.
DDR3 memories have higher latencies than DDR2 memories, which in turn have higher latencies than DDR memories. DDR2 and DDR3 memories have an additional parameter called AL (Additional Latency) or simply A. With DDR2 and DDR3 memories the total latency will be CL+AL. Fortunately, almost all DDR2 and DDR3 memories are AL 0, meaning that no additional latency is necessary. Below we summarize the most common latency values.
Other Common Latencies Available
6, 8, 9
This means that DDR3 memories delay more clock cycles to start delivering data compared to DDR2 memories (just like DDR2 memories delay more clock cycles to start delivering data compared to DDR memories), but this does not necessarily mean a longer wait time (this will be true only when comparing memories working at the exact same clock rate).
For example, a DDR2-800 CL5 memory will delay less time (i.e., is faster) to start delivering data than a DDR3-800 CL7 memory. However, since both are “800 MHz” memories, both provide the exact same maximum theoretical transfer rate (6,400 MB/s). Also, it is important to remember that the DDR3 memory will consume less power than the DDR2.
When comparing modules with different clock rates, you need to do some math to be able to compare the latencies. Notice that we are talking about “clock cycles.” When the clock is higher, each clock cycle is shorter (i.e., shorter period). For example, on a DDR2-800 memory, each clock cycle takes 2.5 ns (1 ns = 0.000000001 second). The math is simple: period = 1 / frequency. (Note that you need to use the real clock, not the DDR clock on this formula. To make things easier, we compiled the reference table below.) The initial wait time of a DDR2-800 memory with CL 5 corresponds to 12.5 ns (2.5 ns x 5). Now suppose we have a DDR3-1333 memory with CL 7. With this memory, each clock cycle will have a period of 1.5 ns (see table below), so the total wait time (latency) will be 10.5 ns (1.5 ns x 7). Even though the latency of this DDR3 memory appears to be higher (7 vs. 5), the wait time is actually lower. So don’t go around thinking that DDR3 memories have worse latencies than DDR2 memories; it will depend on the clock rate you are talking about.
Usually manufacturers announce the memory timings as a series of several numbers separated by a dash (e.g., 5-5-5-5, 7-10-10-10, etc.). The CAS latency is always the first number from these series. See the examples on Figures 3 and 4. If you want to know what the other numbers mean, please read our tutorial Understanding RAM Timings.
Dynamic memories store data inside an array of tiny capacitors. DDR memories transfer two bits of data per clock cycle from the memory array to the memory internal I/O buffer. This is called 2-bit prefetch. On DDR2 this internal datapath was increased to four bits, and on DDR3 it was raised again to eight bits. This is actually the trick that allows DDR3 to work at higher clock rates than DDR2, and DDR2 at higher clock rates than DDR.
The clocks to which we have been referring so far are the clock rates on the “external world,” i.e., on the I/O interface from the memory, where the communication between the memory and the memory controller takes place. Internally, however, the memory works a little differently.
To better understand this idea, let’s compare a DDR-400, a DDR2-400 and a DDR3-400 memory chip. (We know that DDR3-400 memories don’t exist, but pretend they do.) These three chips work externally at 200 MHz transferring two data per clock cycle, achieving an external performance as if they were working at 400 MHz. Internally, however, the DDR chip transfers two bits between the memory array and the I/O buffer, so to match the I/O interface speed this datapath has to work at 200 MHz (200 MHz x 2 = 400 MHz). Since on DDR2 this datapath was increased from two bits to four bits, it can work at half the clock rate in order to achieve the same performance (100 MHz x 4 = 400 MHz). The same thing happens with DDR3. The datapath was doubled again to eight bits, so it can work at half the clock rate as DDR2 or only ¼ of the clock rate of DDR in order to achieve the same performance (50 MHz x 8 = 400 MHz).
Doubling the internal datapath at each generation means that each new memory generation can predictably have chip models with double the maximum clock rate achieved on the previous one. For example, on DDR-400, DDR2-800 and DDR3-1600 memories the memory works internally at the same clock rate (200 MHz).
On DDR memories, the necessary resistive termination is located on the motherboard, while on DDR2 and DDR3 memories this termination is located inside the memory chips – a technique called ODT (On-Die Termination).
This is done in order to make the signals “cleaner.” In Figure 5, you can see the signal that reaches the memory chip. On the left hand side you see the signals on a system that uses motherboard termination (DDR memories), while on the right hand side you see the signals on a system that uses on-die termination (DDR2 and DDR3 memories). Even a layman can easily see that the signals on the right-hand side are cleaner and are more stable than the signals on the left-hand side. On the yellow square you can compare the time frame difference; this time frame is the time the memory has to read or write a piece of data. With the use of on-die termination, this time frame got wider, allowing higher clocks to be achieved since the memory has more time to read or write a data chunk.
Finally, we have the differences at the physical level. You buy memory chips already soldered on a printed circuit board called a “memory module.” Memory modules for each DDR generation are physically different, so for example you won’t be able to install a DDR2 module in a DDR3 socket. Unless your motherboard supports both DDR2 and DDR3 sockets (only a few do), you cannot upgrade from DDR2 to DDR3 without replacing the motherboard and eventually the CPU (if in your system the memory controller is embedded in the CPU, as with all processors from AMD and Core i7 from Intel). The same thing is valid with DDR and DDR2. Other than for a few are rare exceptions, you cannot replace DDR memories with DDR2. DDR2 and DDR3 modules have the same number of pins, however, the key notch is placed in a different position.
Number of Pins
All DDR2 and DDR3 chips use BGA (Ball Grid Array) packaging, while DDR chips almost always use TSOP (Thin Small-Outline Package) packaging. There are a few DDR chips with BGA packaging on the market (like the ones from Kingmax), but they are uncommon. In Figure 9, you can see what a TSOP chip on a DDR module looks like, while in Figure 10 you can see what a BGA chip on a DDR2 looks like.