DDR and DDR2 details 2004-11-23CPLUSE article directory 1. Foreword 2.ddr and DDR2 details 3 .ddr and DDR2 detailed introduction (continued) 4. Introduction to the product 5. Test platform and set 6. Test results - SUPER PI7. Test Results - QUAKE3 ARENA8. Test Results - Pcmark049. Test Results - Sisoftware Sandra 200410. Test Results - SCIENCEMARK 2.011. Conclusions DDRDDR (Double Data Rate) SDRAM, as the nameship is two times in SDRAM at the same operating frequency Speed to transfer data, that is, transmit two data per hour cycle - clock signal rising edge and falling edge transmission once. So what is the doubled data transmission rate? The DDR memory stored array operating frequency and the same frequency SDRAM (such as the SDRAM and DDR200 of the PC 100), but the internal bus - the bus width between the internal storage arrays to the I / O cache is an external bus (buffer to memory) Two times the controller), that is, the frequency doubled from the cache to the memory controller. That is, the storage array uses a wide but slower bus, but a narrower but fast bus is used when the data is transmitted to the controller. In fact, the internal DRAM storage array in DDR266 memory is 133MHz, in DDR333, the operating frequency of the storage array is 166MHz, the memory array operating frequency in the DDR400 is 200MHz, the current fastest DDR SDRAM frequency (here Does not include those overclocking memory) reaches 600 MHz, and its internal array operating frequency reaches 300MHz, which is difficult to continue to increase. However, the development of the CPU continues, and the requirements for memory bandwidth are not stopped, so DDR2 techniques have occurred. DDR2 is easy to understand by the above-faced DDR SDRAM, like DDR, and its internal storage arrays to I / O cache pass a spacious 64-bit, 100MHz bus, but data from cache transmission To the external controller passes a fast and narrow bus (16-bit, 200 MHz), the external bus still uses double-time strategies, and we get the data transfer rate of 400MHz. Therefore, the 64-bit module requires simultaneous use of 4 BANK to reach a 64-bit bus width. This memory module is called DDR2 400, and its tag method and DDR memory are identified by memory data transfer rate. Therefore, the DRAM storage array operated in the same 100MHz frequency, we use different memory module widths, get different memory bandwidth, SDRAM is 800MB / s, DDR SDRAM is 1600MB / s, DDR2 SDRAM reaches 3200MB / s data Transmission rate! In this way, DDR2's performance seems to be much easier than DDR. In fact, in addition to bandwidth, there is still an important parameter to be delayed. It is because of the influence of delays, DDR2 performance is not as general ideal. Let us see if there are those delays, such as the memory combination of memory arrays work 2-2-2, if the memory array works in all schemes, all modules have the same delay (here It is PC100, DDR200, DDR2 400). They are only the difference between bandwidth. By the way, the meaning of 2-2-2 combination is: CAS delay, RAS to CAS delay and RAS precharge time.
The first number is the delay time of the column address, the second number is the delay between the rows and column addresses, and the third number is a storage array charging time, and the precharge is actually a read operation of the data. However, in fact, the storage array does not work on the same frequency. For example, the PC133 is a very common SDRAM, and its DRAM unit operates on 133MHz. Although the DDR200 has a higher bandwidth than the PC133, its corresponding delay is slower (the internal array is only 100MHz), and the frequency of the storage array of PC 133 is 33% higher than the frequency of the DDR200 storage array. As a result, DDR266 has the advantage of the same delay like PC133. Now we also see similar situations, DDR200 and DDR2 400 have the same delay, while the latter has a double bandwidth. In fact, DDR2 400 and DDR400 have the same bandwidth, which are 3.2gb / s, but the DDR400's storage array operating frequency is 200MHz, and the memory array of DDR2 400 is 100MHz, that is, DDR2 400 delay It is higher than DDR400. The above figure shows the comparison of DDR2 533 and DDR266, which can see the delay of the two is at a level. Let us compare the number, take DDR2 533, which is typically 3.75 ns (min) / 8ns (max) when calculating the CL (CAS Latency) parameter, while DDR400 is 5ns. Therefore, for the DDR2 533 memory for the CAS delay to 4 cycles, its delay time is 15 ns (min) / 32ns (max), and the DDR400 of the CAS delay 2 cycle is 10 ns, even if the DDR400 of CL is 2.5 can also Reaching 12.5 ns, lower than DDR2 533. As for why DDR2's CAS delay parameters are minus, since the external I / O bus frequency of DDR2 is caused by inconsistencies in the operating frequency of the internal chip, the external I / O bus frequency is 266MHz. (Cycle is about 3.75 ns), and the internal chip operating frequency is 133MHz (cycle 7.5ns), so the fastest 3.75Ns can be achieved in a smooth situation, and the slowest is only 8ns. The situation seems to be absurious, although DDR2 can provide greater bandwidth, with potential advantages, but DDR2 initial products in the performance of DDR. So in addition to the future, it can develop to a higher level (such as DDR2 800), what is DDR2? DDR2 advantages
DDR memory usually uses TSOP chip packages. This package can work well on 200MHz. When the frequency is higher, it will produce high impedance and parasitic capacitance, which will affect it. The difficulty of stability and frequency improvement. Here is another package -BGA or Ball Grid Array, which can adapt to higher operating frequencies, such as high-speed memory with a large number of BGA packages in the graphics card. This package has a very low impedance and parasitic capacitance, and has a smaller geometric size, allowing more cooling. DDR2 memory particles will be encapsulated in BGA. However, the cost of the BGA package is higher than TSOP. Of course, the BGA package is not unique to DDR2 memory. In addition to a large number of available on the graphics card, the Kingmax has already used the DDR memory in the BGA package.
Another difference between the DDR and DDR2 is the signal termination resistance. Since the high frequency signal reflects the end of the signal line, this affects the signal quality, confuses useful signals and reverse signals. We need to prevent this happening, so that the signal reflection is reduced by using the end of the circuit endpoint. There are hundreds of signal terminal resistors on the motherboard using DDR SDRAM, you can take a closer look at those under the DIMM slot, which is the so-called end resistance, a large number of end resistance increasing the manufacturing cost of the motherboard. DDR2 is built directly into the core in the core, so there is no longer need to provide end circuits on the motherboard. This is the so-called ODT (ON-DIE TERMINATION). In fact, different memory modules do not have the requirements of the end circuit. The size of the end resistance determines the signal-to-noise ratio and reflectivity of the data line. The end-tension resistance is low, but the signal-to-noise ratio is also low. The end is high, and the signal-to-noise ratio of the data line is high, but the signal reflection will increase. Therefore, the termination resistance on the motherboard does not match the memory module, which will affect the signal quality to a certain extent. DDR2 can build a suitable end resistance in accordance with their own characteristics, which ensures optimal signal waveform. Using DDR2 can not only reduce the cost of the motherboard, but also the best signal quality, which is DDR that cannot be comparable.
There is also a big advantage of DDR2 relative to DDR to heat generation. The core operating voltage of DDR2 is only 1.8V, while the core working voltage of DDR is 2.5-2.6V (high voltage means high fever), and therefore, the heat generated by DDR2 is less than DDR, and the general estimate can be reduced by 30%. In fact, we also noticed that there are many DDRs work in 1.8V, but since the DDR module specification requires 2.5V operating voltage, it needs to convert 2.5V input voltage, this part will become heat distributed, so Use low voltage DDR to reduce the amount of heat of the module.
In general, the current DDR2 does not fully defeat DDR in performance, of course, there is no advantage in production costs. However, due to its low fever and low frequency storage array frequency, the future development potential is very large. I believe that the emergence of higher-specific DDR2 will make their performance more on the floor.
See the table below for DDR and DDR2.
DDR
DDR2
Data transmission frequency
200/266/333/400 MHz
400/533 / (667) MHz
I / O
Bus frequency
100/133/166/200 MHz
200/266 / (333) MHz
Storage array frequency
100/133/166/200 MHz
100/133 / (166) MHz
Batch Reading Size
2/4/8
4/8
Cas latency
1.5, 2, 2.5
3 , 4, 5
Write latency
1T
Read Latency-1
Pin
Quantity
184PIN
240pin
DDR2 advantages
DDR memory usually uses TSOP chip packages. This package can work well on 200MHz. When the frequency is higher, it will produce high impedance and parasitic capacitance, which will affect it. The difficulty of stability and frequency improvement. Here is another package -BGA or Ball Grid Array, which can adapt to higher operating frequencies, such as high-speed memory with a large number of BGA packages in the graphics card. This package has a very low impedance and parasitic capacitance, and has a smaller geometric size, allowing more cooling. DDR2 memory particles will be encapsulated in BGA. However, the cost of the BGA package is higher than TSOP. Of course, the BGA package is not unique to DDR2 memory. In addition to a large number of available on the graphics card, the Kingmax has already used the DDR memory in the BGA package.
Another difference between the DDR and DDR2 is the signal termination resistance. Since the high frequency signal reflects the end of the signal line, this affects the signal quality, confuses useful signals and reverse signals. We need to prevent this happening, so that the signal reflection is reduced by using the end of the circuit endpoint. There are hundreds of signal terminal resistors on the motherboard using DDR SDRAM, you can take a closer look at those under the DIMM slot, which is the so-called end resistance, a large number of end resistance increasing the manufacturing cost of the motherboard. DDR2 is built directly into the core in the core, so there is no longer need to provide end circuits on the motherboard. This is the so-called ODT (ON-DIE TERMINATION). In fact, different memory modules do not have the requirements of the end circuit. The size of the end resistance determines the signal-to-noise ratio and reflectivity of the data line. The end-tension resistance is low, but the signal-to-noise ratio is also low. The end is high, and the signal-to-noise ratio of the data line is high, but the signal reflection will increase. Therefore, the termination resistance on the motherboard does not match the memory module, which will affect the signal quality to a certain extent. DDR2 can build a suitable end resistance in accordance with their own characteristics, which ensures optimal signal waveform. Using DDR2 can not only reduce the cost of the motherboard, but also the best signal quality, which is DDR that cannot be comparable.
There is also a big advantage of DDR2 relative to DDR to heat generation. The core operating voltage of DDR2 is only 1.8V, while the core working voltage of DDR is 2.5-2.6V (high voltage means high fever), and therefore, the heat generated by DDR2 is less than DDR, and the general estimate can be reduced by 30%. In fact, we also noticed that there are many DDRs work in 1.8V, but since the DDR module specification requires 2.5V operating voltage, it needs to convert 2.5V input voltage, this part will become heat distributed, so Use low voltage DDR to reduce the amount of heat of the module.
In general, the current DDR2 does not fully defeat DDR in performance, of course, there is no advantage in production costs. However, due to its low fever and low frequency storage array frequency, the future development potential is very large. I believe that the emergence of higher-specific DDR2 will make their performance more on the floor.
See the table below for DDR and DDR2.
DDR
DDR2
Data transmission frequency
200/266/333/400 MHz
400/533 / (667) MHz
I / O
Bus frequency
100/133/166/200 MHz
200/266 / (333) MHz
Storage array frequency
100/133/166/200 MHz
100/133 / (166) MHz
Batch Reading Size
2/4/8
4/8
Cas latency
1.5, 2, 2.5
3 , 4, 5
Write latency
1T
Read Latency-1
Pin
Quantity
184PIN
240pin
DDR2 advantages
DDR memory usually uses TSOP chip packages. This package can work well on 200MHz. When the frequency is higher, it will produce high impedance and parasitic capacitance, which will affect it. The difficulty of stability and frequency improvement. Here is another package -BGA or Ball Grid Array, which can adapt to higher operating frequencies, such as high-speed memory with a large number of BGA packages in the graphics card. This package has a very low impedance and parasitic capacitance, and has a smaller geometric size, allowing more cooling. DDR2 memory particles will be encapsulated in BGA. However, the cost of the BGA package is higher than TSOP. Of course, the BGA package is not unique to DDR2 memory. In addition to a large number of available on the graphics card, the Kingmax has already used the DDR memory in the BGA package.
Another difference between the DDR and DDR2 is the signal termination resistance. Since the high frequency signal reflects the end of the signal line, this affects the signal quality, confuses useful signals and reverse signals. We need to prevent this happening, so that the signal reflection is reduced by using the end of the circuit endpoint. There are hundreds of signal terminal resistors on the motherboard using DDR SDRAM, you can take a closer look at those under the DIMM slot, which is the so-called end resistance, a large number of end resistance increasing the manufacturing cost of the motherboard. DDR2 is built directly into the core in the core, so there is no longer need to provide end circuits on the motherboard. This is the so-called ODT (ON-DIE TERMINATION). In fact, different memory modules do not have the requirements of the end circuit. The size of the end resistance determines the signal-to-noise ratio and reflectivity of the data line. The end-tension resistance is low, but the signal-to-noise ratio is also low. The end is high, and the signal-to-noise ratio of the data line is high, but the signal reflection will increase. Therefore, the termination resistance on the motherboard does not match the memory module, which will affect the signal quality to a certain extent. DDR2 can build a suitable end resistance in accordance with their own characteristics, which ensures optimal signal waveform. Using DDR2 can not only reduce the cost of the motherboard, but also the best signal quality, which is DDR that cannot be comparable.
There is also a big advantage of DDR2 relative to DDR to heat generation. The core operating voltage of DDR2 is only 1.8V, while the core working voltage of DDR is 2.5-2.6V (high voltage means high fever), and therefore, the heat generated by DDR2 is less than DDR, and the general estimate can be reduced by 30%. In fact, we also noticed that there are many DDRs work in 1.8V, but since the DDR module specification requires 2.5V operating voltage, it needs to convert 2.5V input voltage, this part will become heat distributed, so Use low voltage DDR to reduce the amount of heat of the module.
In general, the current DDR2 does not fully defeat DDR in performance, of course, there is no advantage in production costs. However, due to its low fever and low frequency storage array frequency, the future development potential is very large. I believe that the emergence of higher-specific DDR2 will make their performance more on the floor.
See the table below for DDR and DDR2.
DDR
DDR2
Data transmission frequency
200/266/333/400 MHz
400/533 / (667) MHz
I / O
Bus frequency
100/133/166/200 MHz
200/266 / (333) MHz
Storage array frequency
100/133/166/200 MHz
100/133 / (166) MHz
Batch Reading Size
2/4/8
4/8
Cas latency
1.5, 2, 2.5
3 , 4, 5
Write latency
1T
Read Latency-1
Pin
Quantity
184PIN
240pin
DDR2 advantages
DDR memory usually uses TSOP chip packages. This package can work well on 200MHz. When the frequency is higher, it will produce high impedance and parasitic capacitance, which will affect it. The difficulty of stability and frequency improvement. Here is another package -BGA or Ball Grid Array, which can adapt to higher operating frequencies, such as high-speed memory with a large number of BGA packages in the graphics card. This package has a very low impedance and parasitic capacitance, and has a smaller geometric size, allowing more cooling. DDR2 memory particles will be encapsulated in BGA. However, the cost of the BGA package is higher than TSOP. Of course, the BGA package is not unique to DDR2 memory. In addition to a large number of available on the graphics card, the Kingmax has already used the DDR memory in the BGA package.
Another difference between the DDR and DDR2 is the signal termination resistance. Since the high frequency signal reflects the end of the signal line, this affects the signal quality, confuses useful signals and reverse signals. We need to prevent this happening, so that the signal reflection is reduced by using the end of the circuit endpoint. There are hundreds of signal terminal resistors on the motherboard using DDR SDRAM, you can take a closer look at those under the DIMM slot, which is the so-called end resistance, a large number of end resistance increasing the manufacturing cost of the motherboard. DDR2 is built directly into the core in the core, so there is no longer need to provide end circuits on the motherboard. This is the so-called ODT (ON-DIE TERMINATION). In fact, different memory modules do not have the requirements of the end circuit. The size of the end resistance determines the signal-to-noise ratio and reflectivity of the data line. The end-tension resistance is low, but the signal-to-noise ratio is also low. The end is high, and the signal-to-noise ratio of the data line is high, but the signal reflection will increase. Therefore, the termination resistance on the motherboard does not match the memory module, which will affect the signal quality to a certain extent. DDR2 can build a suitable end resistance in accordance with their own characteristics, which ensures optimal signal waveform. Using DDR2 can not only reduce the cost of the motherboard, but also the best signal quality, which is DDR that cannot be comparable.
There is also a big advantage of DDR2 relative to DDR to heat generation. The core operating voltage of DDR2 is only 1.8V, while the core working voltage of DDR is 2.5-2.6V (high voltage means high fever), and therefore, the heat generated by DDR2 is less than DDR, and the general estimate can be reduced by 30%. In fact, we also noticed that there are many DDRs work in 1.8V, but since the DDR module specification requires 2.5V operating voltage, it needs to convert 2.5V input voltage, this part will become heat distributed, so Use low voltage DDR to reduce the amount of heat of the module.
In general, the current DDR2 does not fully defeat DDR in performance, of course, there is no advantage in production costs. However, due to its low fever and low frequency storage array frequency, the future development potential is very large. I believe that the emergence of higher-specific DDR2 will make their performance more on the floor.
See the table below for DDR and DDR2.
DDR
DDR2
Data transmission frequency
200/266/333/400 MHz
400/533 / (667) MHz
I / O
Bus frequency
100/133/166/200 MHz
200/266 / (333) MHz
Storage array frequency
100/133/166/200 MHz
100/133 / (166) MHz
Batch Reading Size
2/4/8
4/8
Cas latency
1.5, 2, 2.5
3 , 4, 5
Write latency
1T
Read Latency-1
Pin
Quantity
184PIN
240pin
