The most important three performance metrics for the graphics card
Capacity
Video memory processing system and graphics between the data exchange and display chip operation 3D graphics data cache, so the memory capacity naturally determines the display chip can handle the amount of data. Theoretically, the larger the memory, the better the performance of the graphics card. But these are only theoretical calculations, the actual graphics card performance to love a lot of factors constraints, such as: Display chip speed, video memory bit wide, video memory speed and so on.
Memory bit width
Memory bit width is also an important parameter of the graphics card. Can be understood as the data access to the size of the channel, obviously, in the video memory speed (operating frequency) in the same situation, the greater the bandwidth, the greater the data throughput, the better performance. Now the video card is more common 256BIT and 128BIT, it is obvious that in the same memory frequency, 256BIT memory data throughput 128BIT twice times (in fact, is not in use, this is also affected by other aspects, such as memory, PCI-E slot, Hard drive transmission rate and so on, so the performance will also enhance a lot.
Clock cycle and working frequency
This is a very important performance metric, which refers to the time it takes to process a data memory. The faster the video memory speed, the unit time to Exchange data volume is also greater, in the same situation, the performance of the graphics card will be significantly improved, memory clock cycle in general NS (nanosecond) as the unit, operating frequency at MHz. The memory clock cycle corresponds to the working frequency one by one, and the relationship between them is:
Operating frequency (MHZ) =1÷ clock cycle (NS) x1000
The common memory clock cycle has 5ns,4ns,3.8ns,3.6ns,3.3ns,2.8ns. For DDR SDRAM memory, the equivalent operating frequency is used when describing its working frequency. Because in the clock cycle of the rise and fall along the data can be transmitted, so the operating frequency and data bit width of the same situation, memory bandwidth is twice times the SDRAM. In other words, when the memory clock cycle is the same, the DDR SDRAM video memory is twice times the actual operating frequency of SDRAM memory. For example, 5NS SDRAM memory is working at 200MHZ, while 5NS DDR SDRAM memory is the equivalent operating frequency of 400MHZ.
Let's take a look at the formula for calculating the four main parameters:
Memory capacity = Single memory capacity x number of memory particles
Memory bit width = Single memory bit width x video memory particle quantity
Video memory Working frequency = Single memory working frequency
Bandwidth = working frequency x video Memory bit width ÷8
Knowing the bit width and speed of the video memory, you can know the bandwidth of the video memory:
Bandwidth = working frequency x video Memory bit width ÷8
The reason to divide by 8 is because every 8 bit (bit) equals a byte (byte). Bandwidth is the final measure of video memory speed, the size of the data throughput is the speed of video memory to see bandwidth.
Single Memory speed
1/nanosecond speed X 2 X1000 = frequency
For example, a 1.4NS video memory x1600xt DDR3 memory card, then the graphics card memory frequency calculation method is:
1/1.4 x 2 x 1000 =1428mhz 1/nanosecond speed X 2 X1000 = frequency
The bandwidth calculation method for this video card is:
1428Mhz x 128BIT/8 = 22.857gb/s bandwidth = Working frequency X video memory bit width ÷8
The bandwidth of the video card basically measures a same grade, under the different video memory specification, two graphics card performance is poor.