capacity unit
The computer's judgment of the data is mainly based on whether there is electricity to record, so theoretically, for each recording unit, it only recognizes 0 and 1. The binary unit of 0/1 is called bit (bit), but the bit is too small, so each data will be recorded with a size of 8 bits, so the unit of Byte (byte) is defined, Their relationship is 1 Bytes = 8 bits.
But Byte is still too small. In order to facilitate reading, a simplified unit representation is derived, as shown in the following table:
Generally speaking, the file capacity is in binary, so the file size of 1GBytes is actually: 1024*1024*1024Bytes; the speed unit often uses decimal, for example, 1GHz means 1000*1000*1000Hz.
Tips: To distinguish between binary and decimal capacity, GB and GiB are often used to distinguish the capacity. GB is in decimal, and if i becomes GiB, it is in binary. All capacities can be handled this way, including KiB, TiB, etc.
speed unit
The operating speed of the CPU is often measured in units such as MHz or GHz, and this Hz is one-second. In terms of network transmission, since the network uses bit as the unit, the unit often used for network transmission is Mbps, that is, Mbits per second, that is, how many Mbits per second.
For example: 20M/5M optical fiber transmission speed, if converted into Bytes of file capacity, the theoretical maximum transmission speed is: 2.5MBytes per second / 625KBytes per second download/upload speed.
clock
After a clock pulse, the CPU's signal line needs time to settle to its new state. If the signal of the previous pulse has not been processed and the next clock pulse comes too fast (before all signal lines have completed the transition from 0 to 1 or from 1 to 0), it will produce wrong results.
Chip makers have specifications for "maximum clock frequencies," and chips are tested to ensure they meet the specifications for "maximum clock frequencies" before they are sold. The test will execute the most complex instructions, process the most complex data model, determine the maximum processing time used (tested at the most suitable voltage and stability to ensure that the CPU runs at the lowest performance), and ensure that there will be no conflicts at the highest clock frequency.
CPU clock
The clock is the number of times the CPU can work per second, which is the clock frequency. So the higher the clock, the more the CPU can do per unit time.
For example, Intel's i7-4790 CPU clock is 3.6GHz, which means that this CPU can perform 3.6*10^9 jobs in one second, and each job can perform a few instruction operations. Tips: Different CPUs cannot simply use the clock to judge the computing performance, because each CPU's micro-instruction set, architecture, available second-layer cache and its computing mechanism, and the work that can be performed by each clock. The number of instructions may vary, so currently it can only be used to compare the speed of the same CPU.
CPU working clock: FSB and multiplier
The early CPU architecture mainly connected the most important CPU, memory, and graphics card of the system through the north bridge. Because all devices have to be connected through the north bridge, the operating frequency of each device should be the same, so the front side bus (FSB) was born.
However, because the computing speed of the CPU is faster than that of other devices, and in order to meet the frequency of the FSB, the manufacturer accelerates it inside the CPU, so there are so-called FSB and multiplier.
In general, in the early CPU design, the so-called FSB refers to the speed of data transmission between the CPU and external components, and the multiplier is a multiplier used by the CPU to accelerate work efficiency. is the clock speed of the CPU.
For example, the internal frequency of Intel Core2 E8400 is 3.0GHz, and the external frequency is 333MHz, so the multiplier is 9 times. (3.0G=333M*9, of which 1G=1000M)
CPU working clock: overclocking
Overclocking refers to a way to change the CPU's multiplier or FSB to a higher frequency through the motherboard's setting function. However, because the multiplier of the CPU is usually locked at the factory and cannot be modified, the frequency that is often overclocked is the FSB.
For example, if you want to overclock a CPU like the above 3.0GHz, you can adjust its FSB from 333MHz to 400MHz, but in this way, the operating frequency of each component of the entire motherboard may be increased to 1.2 times the original, although the CPU may be able to Up to 3.6GHz, but because it is not a normal speed, it may cause problems such as downtime.
But in this way, all the data are stuck by the north bridge, and the north bridge cannot be faster than the CPU, so it is often the bottleneck of performance.
In the new CPU design, the memory controller has been integrated into the CPU, and the design of the controller connecting the CPU, memory and graphics card uses QPI (Quick Path Interconnect) and DMI technology in the Intel part; and Hyper is used in the AMD part. Transport. These technologies allow the CPU to communicate directly with devices such as memory and graphics cards, without the need for external connection chips.
Because there is no north bridge now, the clock design of the CPU does not need to consider the FSB to be synchronized, but only needs to consider the overall frequency. And most of the current CPUs have automatic overclocking technology.