Gpuz computing power
Publish: 2021-04-26 04:09:01
1.
The calculation formula of pixel fill rate, texture fill rate and video memory bandwidth in gpu-z:
(1) pixel fill rate refers to the number of pixels that can be rendered by the graphics card in each unit time, the unit is gppixel / S (billion pixels per second)
pixel fill rate = GPU clock × The pixel fill rate of Fermi frame is SM array × two × Core frequency
(2) texture fill rate refers to the number of texture maps that the graphics card can process per unit time, in gtexel / S (billion textures per second)
texture fill rate = core frequency (GPU clock) × Number of texture units / 1000
(3) the bit width of video memory refers to the number of bytes that the video memory can transmit in a clock cycle, and the unit is GB / S (gigabytes per second)
the band width of video memory = bus width × Equivalent frequency of video memory / 8
take geforce gtx670 as an example, with core frequency of 1059mhz, video memory frequency of 1502mhz, 32 raster units and 112 texture units,
pixel fill rate = 16 × two × 608 = 19.456gpixel/s
texture fill rate = 608 × 64 / 1000 = 38.912 gtexel / s
band width = 384 ×( eight hundred and fifty-four × 4) / 8 = 163.968 GB / s
note: when calculating the bandwidth of video memory, attention should be paid to the memory type. For gddr1 / 2 / 3 / 4 video memory, its data bus adopts DDR Technology (data transmission is carried out at both rising and falling edges through differential clock, and data is transmitted twice in one cycle, which is equivalent to 2 times of SDRAM frequency), so its equivalent frequency of video memory = video memory frequency × 2 Gddr5 is different. It has two data buses, which is equivalent to Rambus's QDR technology. Its transmission capacity is equivalent to four times of SDRAM frequency, so the equivalent frequency of video memory = video memory frequency × 4.
2. When it comes to processor identification tool CPU-Z, its popularity and necessity need not be mentioned. The hardware website techpowerup.com now provides us with a similar tool, "gpu-z" for graphics card recognition. Gpu-z takes both a card and N card, and it can also recognize fire and mixed fire. Before you buy or upgrade your graphics card, you'd better take a look at the screenshot of gpu-z in the online evaluation to see what's better than your current one, so as to decide whether to sell or not
0.42: the free application for Asus graphics card has been added. You can get the chance to fill in a lucky draw registration in one minute. Although it's a gimmick, you can try your luck. Note: it is said that application for integrated graphics card is not supported.
0.42: the free application for Asus graphics card has been added. You can get the chance to fill in a lucky draw registration in one minute. Although it's a gimmick, you can try your luck. Note: it is said that application for integrated graphics card is not supported.
3. Directcompute: this function is only supported by Dx11 graphics cards and some DX10.1 graphics cards (supported by my HD4830)
OpenCL: relatively new technology, NVIDIA gtx200 series, ATI radon hd5000 series or newer graphics cards
CUDA, PhysX: these two 8600gt should be supported, maybe the driver is not installed, or... Fake card
OpenCL: relatively new technology, NVIDIA gtx200 series, ATI radon hd5000 series or newer graphics cards
CUDA, PhysX: these two 8600gt should be supported, maybe the driver is not installed, or... Fake card
4.
In fact, you can download a Chinese version. Here is the description of each project corresponding to the Chinese Version (this GPU is Intel, so it is the logo of Intel)
5. It's not gpu-z, it's CPU-Z that tests your system. It can detect your system configuration information
6. Gpu-z can't test the performance of the graphics card, it can only display the specification parameters of the graphics card, such as core frequency, real-time frequency, video memory, OpenGL, etc., which are more detailed, and then compare with the parameters published on the official website to see if there are any differences; Test the performance of the graphics card, you can use professional running software and games, the game is more comprehensive.
7. ① . graphics card name part:
name / Name: the name of the graphics card is displayed here, which is the graphics card model used in this notebook
(after the author's test, it is found that the identification of the graphics card name in the latest version 0.1.0 gpu-z is not based on the hardware ID code of the display core, but comes from the hardware name in the graphics card driver information. Although this detection method is relatively simple to implement, the information obtained is not very reliable, which will be explained in detail later.)
2. Display chip model:
core code / GPU: the code of GPU chip is displayed here, such as g84m, g86m, RV630, rv610, etc
revision / revision: the step number of the GPU chip is displayed here
manufacturing process / Technology: the manufacturing process of GPU chip is shown here, such as 80nm, 65nm, 55nm, etc
core area / die size: the core size of the GPU chip is shown here This database has not been added to the current version of gpu-z)
③ hardware information of the graphics card:
BIOS version / BIOS version: the version number of the BIOS of the graphics card is displayed here
device ID / device ID: the ID code of the device is displayed here
manufacturer / subcontractor: the name of the manufacturer of the graphics card is displayed here
4. Display chip parameters:
raster engine / ROPS: the number of ROP raster operation processors owned by GPU is displayed here. The more the performance is, the stronger the performance is
bus interface / bus interface: the bus interface type and interface speed between the graphics card and the north bridge chip of the main board are displayed here
shading units / shaders: the number of shaders owned by GPU is displayed here. The more the performance is, the better
(currently, the mainstream NVIDIA geforce 8 and ATI hd2000 Series graphics cards adopt a unified rendering architecture, and the shading units are collectively referred to as stream processor. The previous graphics card's shading unit is divided into two parts: pixel shaders and vertex shaders.)
DirectX version / DirectX support: the version of DirectX supported by GPU is displayed here
pixel fill rate / pixel fill rate: the pixel fill rate of GPU is displayed here. The more the performance is, the better
texture fill rate / texture fill rate: the texture fill rate of GPU is displayed here. The more the performance is, the better< (5) video memory information part:
video memory type / memory type: the video memory type used by the video card is displayed here, such as GDDR3, DDR2, DDR, etc
memory bit width / bus width: the bandwidth between GPU and memory is displayed here. The more the bandwidth, the better the performance
(this parameter of the graphics card is what we should focus on. Many "castrated" cards in the market have been manipulated by the black heart manufacturers. There are many such examples. For example, a high-end notebook of a famous brand (in order to avoid unnecessary trouble, I won't disclose it). The geforce 86000m GS, which was originally 128bit wide, was cut by half in order to rece the cost, leaving only 64bit, which greatly reced the 3D performance. Consumers happily buy the machine and play the game, only to find that the actual performance is very bad. At this time, it's too late to regret. Therefore, we must pay special attention to this parameter when we purchase the unique display notebook. Note: geforce8400 and hd2400 have only 64 bits of native memory, so you don't have to worry about passivity.)
memory capacity / memory size: the physical memory capacity of the graphics card is displayed here
video memory bandwidth: the data transmission speed between gpu-z and video memory is shown here. The more the performance is, the better< (6) driver part:
driver version: here is the version number of the graphics card driver currently used in the system< (7) frequency part of graphics card:
core frequency / GPU clock: displays the current running frequency of GPU
memory / memory: displays the current running frequency of the video memory
shader / shader: displays the current running frequency of the shading unit
original core frequency / default clock: displays the default running frequency of GPU
(original) memory / memory: displays the default running frequency of video memory
(original) shader / shader: displays the default running frequency of the shading unit
Other: SLI or ATI crossfire: displays whether SLI or crossfire is turned on.
name / Name: the name of the graphics card is displayed here, which is the graphics card model used in this notebook
(after the author's test, it is found that the identification of the graphics card name in the latest version 0.1.0 gpu-z is not based on the hardware ID code of the display core, but comes from the hardware name in the graphics card driver information. Although this detection method is relatively simple to implement, the information obtained is not very reliable, which will be explained in detail later.)
2. Display chip model:
core code / GPU: the code of GPU chip is displayed here, such as g84m, g86m, RV630, rv610, etc
revision / revision: the step number of the GPU chip is displayed here
manufacturing process / Technology: the manufacturing process of GPU chip is shown here, such as 80nm, 65nm, 55nm, etc
core area / die size: the core size of the GPU chip is shown here This database has not been added to the current version of gpu-z)
③ hardware information of the graphics card:
BIOS version / BIOS version: the version number of the BIOS of the graphics card is displayed here
device ID / device ID: the ID code of the device is displayed here
manufacturer / subcontractor: the name of the manufacturer of the graphics card is displayed here
4. Display chip parameters:
raster engine / ROPS: the number of ROP raster operation processors owned by GPU is displayed here. The more the performance is, the stronger the performance is
bus interface / bus interface: the bus interface type and interface speed between the graphics card and the north bridge chip of the main board are displayed here
shading units / shaders: the number of shaders owned by GPU is displayed here. The more the performance is, the better
(currently, the mainstream NVIDIA geforce 8 and ATI hd2000 Series graphics cards adopt a unified rendering architecture, and the shading units are collectively referred to as stream processor. The previous graphics card's shading unit is divided into two parts: pixel shaders and vertex shaders.)
DirectX version / DirectX support: the version of DirectX supported by GPU is displayed here
pixel fill rate / pixel fill rate: the pixel fill rate of GPU is displayed here. The more the performance is, the better
texture fill rate / texture fill rate: the texture fill rate of GPU is displayed here. The more the performance is, the better< (5) video memory information part:
video memory type / memory type: the video memory type used by the video card is displayed here, such as GDDR3, DDR2, DDR, etc
memory bit width / bus width: the bandwidth between GPU and memory is displayed here. The more the bandwidth, the better the performance
(this parameter of the graphics card is what we should focus on. Many "castrated" cards in the market have been manipulated by the black heart manufacturers. There are many such examples. For example, a high-end notebook of a famous brand (in order to avoid unnecessary trouble, I won't disclose it). The geforce 86000m GS, which was originally 128bit wide, was cut by half in order to rece the cost, leaving only 64bit, which greatly reced the 3D performance. Consumers happily buy the machine and play the game, only to find that the actual performance is very bad. At this time, it's too late to regret. Therefore, we must pay special attention to this parameter when we purchase the unique display notebook. Note: geforce8400 and hd2400 have only 64 bits of native memory, so you don't have to worry about passivity.)
memory capacity / memory size: the physical memory capacity of the graphics card is displayed here
video memory bandwidth: the data transmission speed between gpu-z and video memory is shown here. The more the performance is, the better< (6) driver part:
driver version: here is the version number of the graphics card driver currently used in the system< (7) frequency part of graphics card:
core frequency / GPU clock: displays the current running frequency of GPU
memory / memory: displays the current running frequency of the video memory
shader / shader: displays the current running frequency of the shading unit
original core frequency / default clock: displays the default running frequency of GPU
(original) memory / memory: displays the default running frequency of video memory
(original) shader / shader: displays the default running frequency of the shading unit
Other: SLI or ATI crossfire: displays whether SLI or crossfire is turned on.
8. Memory controller load: usage rate of video memory controller Hereinafter referred to as MC)
video engine load: utilization rate of video engine (processor) VE)
bus interface load: bus controller utilization BI)
power consumption: power consumption Hereinafter referred to as PC)
VddC: VddC voltage VddC)
in fact, these parameters reflect the load or the state of some control units of the graphics card, and there is no more than a change in the size of the value, which is the literal meaning of the above. But we can combine these parameters to judge the status of some graphics cards
generally speaking, MC is moving, because the normal operation of GPU will call the video memory, which will naturally make MC change, especially when the GPU usage and memory used values are high
and ve may be zero for a long time, unless you turn on hardware acceleration. For example, when you turn on hardware decoding and watch 1080p video, the value of VE will be increased e to calling the video encoding & decoding unit of the graphics card, and the utilization rate of VE will be high when the video specification is very high
the reference value of Bi is not very great. Basically, there is intense GPU activity, and the value of Bi will also increase
PC is a parameter of the graphics card. Generally, the higher the GPU utilization rate is, the higher the value will be. When it reaches 100%, it means that the graphics card has reached the default full load power consumption ratio. For the graphics cards of NVIDIA's Kepler architecture and Maxwell architecture, the actual boost frequency may be affected
VddC generally has several gear values, which is a lower value in standby mode, and this value will maintain a higher value as the graphics card enters high load or full load. When overclocking, you can focus on this value, or try to improve the value of the line, which may be of great help to overclocking.
video engine load: utilization rate of video engine (processor) VE)
bus interface load: bus controller utilization BI)
power consumption: power consumption Hereinafter referred to as PC)
VddC: VddC voltage VddC)
in fact, these parameters reflect the load or the state of some control units of the graphics card, and there is no more than a change in the size of the value, which is the literal meaning of the above. But we can combine these parameters to judge the status of some graphics cards
generally speaking, MC is moving, because the normal operation of GPU will call the video memory, which will naturally make MC change, especially when the GPU usage and memory used values are high
and ve may be zero for a long time, unless you turn on hardware acceleration. For example, when you turn on hardware decoding and watch 1080p video, the value of VE will be increased e to calling the video encoding & decoding unit of the graphics card, and the utilization rate of VE will be high when the video specification is very high
the reference value of Bi is not very great. Basically, there is intense GPU activity, and the value of Bi will also increase
PC is a parameter of the graphics card. Generally, the higher the GPU utilization rate is, the higher the value will be. When it reaches 100%, it means that the graphics card has reached the default full load power consumption ratio. For the graphics cards of NVIDIA's Kepler architecture and Maxwell architecture, the actual boost frequency may be affected
VddC generally has several gear values, which is a lower value in standby mode, and this value will maintain a higher value as the graphics card enters high load or full load. When overclocking, you can focus on this value, or try to improve the value of the line, which may be of great help to overclocking.
9. It is recommended to use gpu-z detection version 0.6.6
your gt630 graphics card should be gddr5, the frequency of video memory is 3200mhz, and the gpu-z detection will show 800MHz.
your gt630 graphics card should be gddr5, the frequency of video memory is 3200mhz, and the gpu-z detection will show 800MHz.
Hot content