N card to eat U, A card to steal electricity, this statement in the DIY circle no one I do not know is known to everyone, but here to eat U, is to occupy the CPU.
In fact, there have been countless people who tried to prove this. If we come to an experiment once again, I believe Honorable Judges will feel no meaning. So today we will verify another case of 'eating U' A card and N card, will one party have higher requirements on the CPU frequency it?
The purpose of this experiment is also to give you DIY players a host of reference with the formation of the host, we all know that since the strong launch of Ryzen AMD, the pattern of the CPU consumer market has been very different at the same budget, we can purchase frequency Relatively higher Intel CPU or more AMD CPUs with more core threads.
Consumers have more choices, we can buy products on demand, is a good thing, but consumers will have more confusion, what kind of product is suitable for me?
If we find that one of the A and N cards requires a higher CPU frequency, then we naturally recommend that you work with higher-frequency CPUs. If we find that one of them has a higher CPU usage, then obviously with the core thread More CPU usage will be better Without further ado, let's start testing below.
experimental method
We will use it Similar performance A card and N card, respectively, with different frequency of CPU, and test their performance at different frequencies of the game and then we will respectively Calculate their CPU frequency in the low to low frequency game performance decline, and finally compare the extent of their game performance decline, loss of performance on the larger side, we can think it will be more dependent on high-frequency CPU.
In order to make the game performance differences as large as possible, the test is as far as possible 720p resolution, low full effects Under.
Because of the low resolution, low effects, on the one hand to eliminate the interference of the graphics bottleneck, on the other hand due to low resolution low effects game frame is very high, 1% of the gap can reflect the difference between the number of frames, if the use of high resolution High rate of effects, this 1% gap will be rounded to the decimal point after the number, which seems no difference.
Graphics card
To ensure the objective experiment, the graphics card should use similar performance A card and N card, but also because of the need to avoid the graphics bottleneck, so after careful thinking, calm analysis, Finally chose the RX Vega 56 and GTX 1070 Ti this pair of friends.
And these two cards are AMD and Nvidia Latest release Two game graphics, very representative.
Dylan RX VEGA 56 8GB HBM2
RX Vega 56 is a new generation of AMD next-generation flagship graphics card, using Vega 10 core, with 3584 stream processors, single-precision 10.5T, the specifications are very high. Diran Demon RX Vega 56DeLand Devil RX Vega 56 has the Dylan demon family tradition of color and appearance, with exaggerated cooling scale, frequency and performance are much higher than the public version, and significantly reduce the temperature and noise of the copy machine.
ASUS CERBERUS-GTX 1070 TI-A8G
GTX1070Ti is a ten-series graphics high-end products, using 16nm process and Pascal architecture GP104 core build, energy consumption is excellent, the performance is very close to the GTX1080. ASUS STRIX GTX 107 0Ti-A8G , The product is equipped with a new generation of DirectCU Ⅲ radiator, temperature of copy machine temperature below 70 °, giving consumers excellent user experience, and has ROG descent and belief lamp, excellent quality.
RX Vega 56 and GTX1070Ti tough performance, run 3DMARK FSE scenarios can be close to 10,000 points, even to meet the needs of players 2K games, to deal with 720p resolution is to kill the chicken chopper.
CPU
In the CPU selection, in order to ensure the same performance processor is the same frequency, we use a from Intel i 3-8100 and i3-8350K , Two also have a physical quad-core, in addition to no difference in frequency of the CPU, for the 720p resolution of the game naturally does not occur CPU usage is too high.
I3-8100 belongs to the eighth generation of Core family, using Kaby Lake Refresh architecture and 14nm ++ process to create quad-core four threads, the frequency can be 3.6GHz Is a new generation of Core Duo entry-level CPU.
Intel Core i3-8350K
I3-8350K also belongs to the eighth generation of Core family, the basic frequency to achieve amazing 4.0GHz , And is not locked design, we can overclock it.
testing platform
Test Platform Description:
Test platform using the mainstream Gigabyte (GIGABYTE) Z370 AROUS Ultra gaming motherboard with Chihi memory, Zhenhua power supply, but also more commonly used test room test platform.
Performance Testing
Tested in the game include the popular "Jedi survival" as well as representatives of the DX12 game "Tomb Raider 10" in the CPU frequency we will choose to use I3-8100, I3-8350K and I3-8350K@4.7 GHz for comparison.
Test description and summary:
due to "Jedi survival" and "NBA 2K18"There are limits on the number of frames, in order to avoid test results are full frame, you can not pull open the gap between the performance of different frequency CPU performance, both games are opened in the high-effects, and other game effects are all low.
in "Jedi survival"The game test we can see the CPU frequency change is very small, If you assemble a computer to eat chicken, CPU frequency may not look too important, but in addition to this game, on the whole in the 720p resolution, CPU frequency differences brought about is still very obvious.
To this end, we made a summary to i 3-8350K@4.7GHz as a benchmark , To see how much of their performance loss at low frequencies.
In fact, the final result A card and N card in low-frequency CPU performance loss is about the same, hard to say, RX Vega 56 CPU frequency requirements will be so a little bit higher , Can we say that the A-card has a little bit of U? And we can also see that when the CPU is raised from 3.6GHz to 4.0GHz, the performance of the game is greatly improved. However, from 4.0GHz to 4.7GHz, overclocking After the benefits have become smaller.
So, for most consumers, when we are CPU clocked at more than 4.0GHz, the ultimate overclocking may bring performance improvements compared to cooling costs and power consumption increase , It's a bit overkill.
to sum up
This test we are at 720p resolution, and most of the games all the low special effects, Test results will be our actual use of amplification.
In fact, we use the daily use, 1080p or even 2K resolution using the high-effect game to play the case, This 1% difference is negligible So we are here Buy a lower frequency of the CPU, you can rest assured that the favorite video card to buy, without having to worry about with A card or N card will result in greater performance loss , Both in the low-frequency CPU performance loss is basically the same.
I did not test the original thought before N card performance loss in the low-frequency CPU will be more, if this is the case, such as Ryzen relatively low-frequency CPU with A card will perform better than with the N card, but after the test, The legendary 3A platform mysterious addition, from this perspective, does not exist.
