Graphics performance depends on the display core, so to distinguish the graphics performance, you must know some of the graphics card parameters!
To facilitate the viewing of parameters, a tool designed to view the parameters of the graphics card is gpu-z.
Through gpu-z, we can compare the graphics card parameters to identify the performance of the graphics card, or even distinguish between true and false graphics card.
The following is a schematic diagram of the GPU (explaining only the important parts):
"Name": Can directly show which chip company produced the model
Note: Nvidia can directly display a specific model, such as the figure directly marked 7600GS
ATI only displays the same series name for the video card, for example, X1650GT only displays 1650 Series
"GPU": Display core, core code, refers to the development of the company's internal number, can also be used to distinguish between performance and height.
"Process": the core of the production process, Unit Nano, the smaller the more advanced technology, the lower power consumption.
Raster: The higher the number, the better the performance of the graphics card.
The grating belongs to the output unit, is responsible for the later rendering, the pixel point raster, mainly affects the anti-aliasing, the dynamic blur and so on special effects, but does not have the big influence to the light.
"Bus Interface": Provide data traffic bandwidth, the current mainstream interface is PCI-E 16x, can provide 8g/s data flow (dual channel, up and down 4g/s).
Note: 16x@16x, representing the highest support flow, as well as the current flow of work, if the motherboard or power effect, the working interface may be reduced!
Shader: The old schema is "render pipeline + shaded vertex", and the new schema is unified as "Unified rendering unit", i.e. "flow processor", the higher the number of high-performance the better.
Legacy architecture distinguishes performance by how much of the "Render pipeline" and "shaded vertices" you can learn about graphics performance.
The new architecture is easier to understand because it has only one cardinality, and the higher the number, the better.
Note: N Card A stream processor works, so the number of stream processors appears to be small.
A card for the "Unified render unit" positioning is not the same, to 5 flow processor unit can work, so the number looks very high.
"Directx support": referred to as DX, is the program written by Microsoft, for multimedia instruction, in graphics, is for the screen effects, the highest level is DX11 (Windows 7).
"Image number filling rate": Raster work data processing flow, formula GPU frequency x Grating = image number filling rate.
"Texture fill rate": Renders the data processing flow of the pipeline/stream processor, the formula GPU frequency x pipeline (processor unit) = The texture fill rate.
"Memory Type": Video memory, provide storage data and exchange data, the higher the memory algebra, the higher the ram frequency, the greater the transmission of data, the current highest level of GDDR5, can be as high as 4600mhz/s above the speed.
"Memory bit width": Memory bit width is a clock cycle can be transmitted data in the number of digits, the larger the number of bits can be transmitted in a moment of data volume, which is an important parameter of memory
Note: Due to the current GDDR5 high frequency video memory, even if the bit width is very low, can also provide a very high video memory bandwidth, even if 128bit, can also provide more than dozens of GB/s bandwidth, so the bit width does not necessarily demand high.
There is also a special example, HD2900XT video memory bit width reached 512bit, but due to the core structure of the reasons, rendering performance can not play, even if there is a 512bit bit width can not achieve the desired effect.
"Memory capacity" memory is used as data storage and exchange of data, but not large capacity means that the graphics card has a high performance.
High-end graphics cards require high-capacity memory support because of the huge core processing data.
Low-end graphics cards do not improve the performance of the video card, even with high-capacity memory, because the core itself is handled at a lower rate.
Note: Due to the relatively cheap memory particles, some manufacturers will put some low-end card configuration of high capacity memory, profiteering, such as GF8500GT 1G, GF9400GT 2G version, but the actual game performance of these two models is equal to GF7300GT 256M high-frequency version.
"Memory Bandwidth": Video memory bandwidth refers to the display of the chip and memory data transfer rate, it in bytes/sec units. Memory bandwidth is one of the most important factors that determine the performance and speed of graphics cards.
Memory Bandwidth Calculation formula: Video Memory bandwidth = working frequency x video Memory bit width/8
Current GDDR5 memory, even if the bit width of 128bit, can also provide more than dozens of g/s bandwidth.
"GPU, Video Memory Frequency": can be used to directly distinguish the same series of high and low, such as HD4850, HD4830 (800SP), the difference between them is only in the GPU, video memory frequency, if the same value, performance is the same.
"Actual frequency": Refers to the GPU, video memory current frequency of work, this value can be changed
"Default frequency" refers to the frequency that is set by the manufacturer, but the frequency with which each vendor is developed varies from one to the other.
"Multi-card interconnection": the use of motherboard support, you can allow 2 of pieces, or more than 2 graphics cards in parallel to the screen rendering processing.
Nvidia:sli
Ati:crossfire, Cross fire.
Graphic:
1, 7600GS VS X1650GT
2, 9800GT VS HD4770
3, graphics performance contrast, ranked (from high to low)
Graphics Card Model Distinction
Amd-ati:ati ye Tian
ATI was founded in 1985, is a focus on graphics processing and motherboard chipset development company, November 2006 by AMD Super Granville acquisition. Video card to deal with the advantages of the strengths, but in the driver installation is somewhat cumbersome, there have been DIY players jokingly: "If you want to become a computer DIY master, be sure to use the ATI graphics card." ”
Partner (Domestic): Sapphire, Dylan Heng Jin
Video card performance suffix: xtx>xt>pro>gt>le
XTX: Represents top level performance, also known as flagship model
XT: High-level, high-frequency version, which means high performance.
Pro: Enhanced version, the model from the establishment of ATI all through, has a very important representative.
GT: Down frequency version (unlike Nvidia appellation), in the same series will be reduced frequency.
SE: Memory bit width reduction version, the general positioning of high-end, will be the same series of video memory bit width reduction.
LE: Reduced architecture, core internal, video memory bit width is reduced.
Radeon X1650GT (RV560):
RV560 Display Core: Internal research and development model title, the general integer r*00 represents the top, the higher the number of performance.
Radeon represents the game level graphics card, X is the Roman numeral "10" meaning, x1*** meaning for the 11th generation graphics card
The digital x*600 represents the performance, the 8~9 represents the high-end, 6 represents the middle end, 5 represents the low-end, the 3~4 represents the low-end.
The digital x**50 represents the generic product, which belongs to the improved type.
The letter GT in the ATI appellation is reduced frequency reduction meaning, in the same series belongs to the frequency reduction version.
New title, starting from the HD3000 series, ATI discarded the previous use of suffix letters to represent the performance of the practice, will be a digital representation of performance, the higher the number of high-performance the better.
HD****X2: Represents a dual-core, one-card integration of two independent display cores.
HD**70: A rather previous XT class
HD**50: A fairly old pro rating
HD4870 (R770):
R770: Internal research and development model title, R*70 represents high-end.
HD represents support for hardware HD decoding and supports direct X10 hardware effects above.
hd4***: Represents the 14th generation graphics card.
HD*800: Represents the high-end of the series.
HD**70: Equivalent to the previous XT level, is a high-frequency version.
X1650GT and X1650pro: These two are the few Gt>pro to rank
x1650gt:rv560 display core, 1:3 gold architecture, 8 pixel rendering pipeline, 24 pixel rendering unit, 8 pixel coloring engine.
x1650pro:rv530 display core, 1:3 gold architecture, 4 pixel rendering pipeline, 12 pixel rendering unit, 5 pixel coloring engine.
Nvidia Hongda
Nvidia was founded in 1993 to focus on graphics processing and motherboard chipset development. Game Card Other graphics advantage is the speed to win, driving the installation and maintenance is easier to understand, is the entry level and fever level common orientation.
Partners (Domestic): News Xfx, Shadow, Li Tai, Ying, Elsa, Sotec
Video card performance suffix: Ultra >gtx>gts>gt>gs>le
Ultra: It means "radical, extreme," and therefore represents top performance, also known as the flagship level.
GTX: Represents high-end performance.
GTS: High performance, also can be called high-frequency product line.
GT: Enhanced version, generally at the midrange level of the use of this title, but also represents the public level.
GS: Frequency reduction version, in the same series, the frequency of the resulting model, the performance is roughly the same.
LE: Low-end, also known as reduced version, not only frequency reduction, or even render pipelines or flow processors are reduced, memory bit width will be reduced to varying degrees.
GeForce 9800GT (G92)
G92: Internal research and development shows the core model, Integer g*0 represents the top model, the smaller the number of performance higher.
GeForce represents the game-level graphics card, 9*** represents the Nineth generation graphics card.
Digital GF *800 representative can differentiate, 8~9 represents high-end, 6 represents the middle end, 5 represents the low-end, 3~4 represents the low-end.
The letter GT represents the enhanced level, which is a high-frequency version in the same series.
Note:
Since the title is likely to be modified with the release of the next generation display core, the name is for reference only
However, the number representing performance does not change and will continue.