Memory Learning 2

2023-01-21   ES  

RAM -Working Principles

Each memory unit is composed of a capacitor that can be stored for a short storage charge. This charge represents what the information stored in the memory unit means. The amount of charge it stored is less than half or no storage charge, which means that the value of this capacitor is 0. Unfortunately, the speed of the memory unit (ie, a capacitor) is very fast (that is, the data stored in it). Therefore, DRAM must include a new circuit, which can continuously check each memory unit (ie capacitor), and then the charges are newly added when needed to keep its value unchanged.

For SRAM, each memory unit has a 2-4 crystal tube. They contain some values (0 or 1) and work in uncertainty in the state of switch (1) or cut off (0). There are 2-4 transistors in the SRAM unit, which makes the physical volume much much compared to DRAM.

When RAM processing information, it actually handles some bites, with only two states in one bit, namely 0 or 1. These consisting of digital string is the so -called binary. RAM stores these digits of data in many rows and columns like grids. These lines and columns include tens of thousands of memory units.

When the CPU (central processor) processing information, it may need to store some information in RAM for later use. If this action is required, the processor will send a “write” signal signal To the CPU, reach the RAM unit through the system bus. These RAM units are then stored in those “grids” according to special addresses. When the CPU needs to read the data in RAM, he will send a request signal to the RAM, which contains address information in these signals to determine the position of the data in the tens of thousands of grids.

System portfolio

Another very important factor of

DRAM is the system bus. The system bus is a channel for transmitting data between CPU and memory. If your processor works at 600MHz, the frequency of processing the data in the CPU is 600MHz, but if the data needs to be transmitted from the CPU to the other peripheral (such as hard disk, etc.), the data must pass through the system bus. Due to the limitation of the system bus bandwidth (at present, the bandwidth of the system bus is generally slower than the clock frequency of the CPU, and the system bus in the computer system is generally 100MHz). Therefore, when the data passes through the system bus, its speed will be Limited to the maximum speed that can be processed by the system bus. As for DRAM, if DRAM can be fast enough to keep up with the system bus, it will not drag the system’s processing speed. Of course, the performance of the system will be relatively improved.

Seven: DRAM -Different types

RAM In addition to the above two basic types, there are many different methods and characteristics in the specific operation operation. The characteristics include data transmission rates, access time, incubation period, quality and operation procedures. The differences in these characteristics have subdivided more memory types, including SDRAM, DDR SDRAM and DRDRAM. The following will introduce the characteristics of the type of memory after the specific subdivision. At present, SDRAM is still the main type of memory. Although there are new DRDRAM and DDR SDRAM, they are not enough for SDRAM to occupy the market. The higher the coming, so you can believe that the use of the latter two types of memory will gradually increase.

1: SDRAM, Synchronous DRAM (synchronous dynamic random memory), which is currently the most widely used and ordinary memory type. As it shows its name, it is synchronized, that is, its working speed is synchronized with the system bus speed. You may have heard or read products such as PC100, PC133 and other words. The 100 and 133 shown here refer to the system bus frequency, that is, the PC100 memory is the most suitable system bus with 100MHz, while PC133 is most suitable for the system system The bus is a 133MHz computer. The current 100MHz system bus is still standard, so the most common on the market is the PC100 SDRAM memory; however It will occupy the PC100 market and its mainstream.

The difference between the

SDRAM unit is its speed (PC100 or PC133), access time, CAS speed, and packaging mode. Memory is determined to be generated by PC100 or PC133 from 100MHz or 133MHz. Theoretically PC100 memory has 800Mbps memory subsystem bandwidth. If the memory is PC133, the memory bandwidth will have 1.1Gbps or 1100Mbps.

The access time of the memory unit is usually only one billionth of one -second. One billionth is actually the minimum time spent by a memory unit in one clock cycle. Most of the PC100 memory deposit time is 8NS, which theoretically allows operation of 125MHz bus bandwidth. If your memory deposit time is 10NS, then it can theoretically run under 100MHz; but this will definitely have a certain program error deviation, so its access time requirement when the existence of PC100 is reached as the request to reach it as the existence of the PC100 as the existence. 8NS; in this way, the support bus is 125MHz. After the deviation part is deducted, the actual support bus frequency is about 100MHz. Similarly, the PC133 memory is required to access 7.5ms during manufacturing to cooperate with the 133MHz system bus. Just as you guessed, smaller memory deposit time means faster.

CAS lurking of a memory unit is the delay in time, that is, the time interval between the first cluster from registering a reading command to the first cluster from the same period in the same period; the typical CAS includes CAS2 and CAS3, CAS2 takes two clocks at the same time when reading information to generate first -digit information results. Smaller CAS means that the memory is faster.

As for the package mode of SDRAM, the use is the same as the DIMM configuration introduced above, which can provide 64 -bit data connection for memory.

2: DDR SDRAM or Double Data Rate SDRAM (dual data speed SDRAM), which is a product launched with the latest memory technology. Compared to the SDRAM used now, theoretically DDR memory can provide double the speed of SDRAM, which will also bring double performance. Like SDRAM, DDR is also synchronized with the system bus clock. The difference is that DDR performs data processing and transmission when the signal rises and falls. It is double that of SDRAM. Therefore, the performance of the 133MHz DDR memory is equivalent to the 266MHz SDRAM memory.

According to what we discussed above, a PC133 memory can provide 1.1Gbps memory bandwidth; based on the reasons mentioned above, it has a 2.2Gbps memory subsystem bandwidth for DDR memory of the same frequency. This can be regarded as a small leap for system performance! At the same time, if the 200MHz DDR memory is used, his memory subsystem bandwidth will reach 3.2Gbps.

is similar to SDRAM. DDR memory also uses the DIMM module, as we described above, DIMM design can provide 64 -bit data connection to RAM.

3: DRDRAM or Direct Rambus DRAM. This is a type of memory that Intel is mainly pushed to cooperate with Intel 820 chipset. Now it seems to be a failed memory. The reason for its failure can be attributed to: the cost is too high. Although the performance of this memory can exceed SDRAM, because its cost is too expensive, the cost of the computer has increased significantly. It can be regarded as a loss, so it is generally believed that Rambus DRAM memory is a failed product.

Generally speaking, the price of DRDRAM is 2-3 times the price of SDRAM, because RAM manufacturers need to re-establish their production lines in order to create DRDRAM memory, and these new equipment and development plans require a lot of money, and this is this, and this is the same, and this is the same. Some funds need to be included in the production cost of DRDRAM, which makes the cost of Rambus DRAM rising sharply. At the same time, the chip set of chip sets supporting this Rambus DRAM memory is relatively limited. It can be seen that this is not the mainstream of the market after all.

The negative impact on DRDRAM memory above, but he also has a good aspect. First of all, the design of DDR has a revolutionary leap in design compared to SDRAM. And the fastest of this bus is 400MHz. Similar to the working principle of DDR SDRAM, DRDRAM also performs data processing and transmission at the same time at the rise and decrease edge of the signal. Therefore, the 400MHz bus speed is equivalent to 800MHz working rate at theoretically.

The above we said that SDRAM uses 16 -bit high -speed data connection, and the SDRAM introduced above is a 64 -bit data connection. Do Rambus DRAM want to use a smaller memory subsystem bandwidth? The reason for this is that smaller memory bandwidth can get higher speeds. In order to achieve the minimum speed shown above, Rambus DRAM uses 16 -bit high -speed data connection. This also makes Rambus theoretically reach the highest bandwidth of 1.6Gbps.

According to the above, you can guess that DRDRAM is not using a standard DIMM memory module. It replaces a module or RIMM called Rabmus Inline Memory Module. This RIMM design is very imagined with DIMM in size and shape.



Random Posts

Detailed interpretation: MIT classic semantic segmentation data set ADE20K, attached to the load link

serialization tool — Introduction to Object Serialization Library MSGPACK Introduction

Ubuntu Docker installation and deployment Java, MySQL, Nginx mirror

python crawling CSDN blog expert all blog content

Apply shared object ApplicationWang Once the