Malit860 mining

1.

GPU uses a large number of computing units and ultra long pipeline, but it only has very simple control logic and saves cache. The CPU not only occupies a lot of space by cache, but also has complex control logic and many optimization circuits

The

GPU is designed based on large throughput

GPU has many ALUs and few caches. The purpose of caching is not to save the data that needs to be accessed later, which is different from CPU, but to improve the service for thread. If many threads need to access the same data, the cache will merge these accesses, and then access DRAM (because the data to be accessed is stored in DRAM instead of cache). After obtaining the data, the cache will forward the data to the corresponding thread. This time is the role of data forwarding. However, e to the need to access DRAM, it will naturally bring the problem of delay

The control unit of GPU (left yellow area block) can combine multiple accesses into less accesses

Although GPU has DRAM latency, it has a lot of ALUs and threads. In order to balance memory latency, we can make full use of the characteristics of multiple ALUs to achieve a very large throughput. Allocate as many threads as possible. Generally speaking, GPU Alu has very heavy pipelines because of this

So it is good at logic control and serial operation with CPU. Different from general data operations, GPU is good at large-scale concurrent computing, which is also needed for password cracking. So in addition to image processing, GPU is more and more involved in computing

Most of the work of

GPU is like this. It has a large amount of calculation, but it has no technical content, and it has to be repeated many times. Just like you need to calculate hundreds of millions of times in a job, the best way is to hire dozens of primary school students to calculate together, and one person can calculate part of it. Anyway, these calculations are not technical, they are just physical work. CPU is like an old professor, who can calculate integral and differential calculus, but the salary is high. An old professor is worth 20 primary school students. If you were Foxconn, which one would you hire? GPU is like this, with a lot of simple computing units to complete a large number of computing tasks, pure sea of people tactics. This strategy is based on the premise that the work of pupils a and B is independent of each other. Many problems involving a lot of computing have this feature, such as password cracking, mining and many graphics computing. These calculations can be divided into several simple tasks, each task can be assigned to a primary school student. But there are still some tasks related to "flow". For example, if you go on a blind date, both parties can continue to develop only when they look good. I can't. You haven't seen each other yet. I'll find someone over there to get all the certificates. This kind of more complex problem is done by CPU

in a word, CPU and GPU are different in design because they are used to deal with different tasks at first. Some tasks are similar to the problems GPU was originally used to solve, so GPU is used to solve them. The computing speed of GPU depends on how many pupils are employed, and the computing speed of CPU depends on how many powerful professors are employed. The professor's ability to deal with complex tasks is very good, but for less complex tasks, it still can't hold many people. Of course, the current GPU can also do some slightly complicated work, which is equivalent to upgrading to the level of junior high school students and senior high school students. But it still needs CPU to feed the data to the mouth before it can start to work. After all, it still depends on CPU to manage

2. Although it can't be said that CPU performance is saturated, it's enough if you don't play games. At present, there are still a few mobile game players pursuing the ultimate experience, most of them are used for office work, so P10 will be more popular, and the CPU of MediaTek is very cost-effective

3. The difference between E7200 and 8400 lies in FSB front-end system bus. The former is 1066MHz, while the latter is 1333mhz, that is to say, the speed of e8400 is higher than that of 7200,
here, the secondary cache of 7200 is only 3M, while 8400 has 6m. The cache will affect the CPU efficiency, that is, the hit rate.
from the price point of view, 7200 is 7808400 in bulk, 1200 in bulk, and the cost performance is 7200, If you just play ordinary games, I suggest using E7200, such as the island crisis is enough. At present, the impact of playing games on graphics card is much higher than CPU, and the most integrated computer is GPU, which is the display chip.
a good graphics card plus E7200 will make you get a better game experience.
for example, gf9800gt
ATI
hd4800
they are all the best of thousands of yuan, 600 yuan with gf9600gt
HD3850 is a good choice for you

4. Adreno510 is on the same level as adreno330 and adreno330. The mali-t860 MP2 belongs to Mali, not to MediaTek. There are several frequency versions, which are located at the middle and low end. The highest frequency version is 700MHz. MP2 is also the al core version. The T860 is not the flagship model, so it is doomed that no matter how high the frequency is, it will not reach the high-end basic level. The performance is certainly not as good as that of the adreno510. Mali's latest and strongest GPU is the t880 MP12, and the exynos 8890 is equipped with the GPU with a frequency of 650MHz. I hope that some slag will not be installed with B. the T860 is only a model. The 8-series GPU is the latest, but it is not the latest or the best. It also has many specifications and parameter versions. Do you want to see only the model? Funny! T760 high frequency or one card one card? Is it T760 MP2 or T760 mp8!

5. The mali-t880 GPU is enhanced on the basis of the mali-t860, with the same architecture and specifications (even the same as the mali-t760), and at most 16 shader cores
the biggest difference also lies in the manufacturing process. The design of mali-t860 is mainly combined with 28nm HPM (which is somewhat out of date). The recommended frequency is 650MHz, while the design of mali-t880 is optimized for 16nm FinFET. The frequency is increased to 850mhz, so the triangle output rate is increased to 1.7 billion per second, and the pixel output rate is up to 13.6 billion per second, which is a full 30% increase
the only new thing worth mentioning is that it can support up to 4K and 120fps output when combined with video codec mali-v550 and display controller mali-dp550
arm claims that the performance of the mali-t880 is about 1.8 times that of the mali-t760, and the energy efficiency is improved by 40%.

6. GPU gflops (floating point operations per second):
mali-t860 MP2@700Mhz ：47.6
Adreno 330@578Mhz : 166.5

the number of running points of A330 is about 3.5 times that of t860mp2, and there is not such a huge gap in actual use, but there is no doubt that the performance of Adreno 330 is stronger, and the performance of Adreno 330 is still at a relatively high level at present, The t860mp2 is a low-end GPU with different market positioning

if you don't understand, please ask. Please take it in time after the problem is solved. Thank you.

7. Theoretically speaking, the performance of Mali T860 is higher than that of Mali t450, but Mali t450 is enough to deal with all the current scenarios, while 860 is somewhat surplus, and 450 is more cost-effective.

8. There are several versions of mali-t860, such as T860 MP ~ T860 MP4, and each version has various frequency versions. You can't compare them if you don't make it clear.

9. It depends on 880 or 860

10. The mali-t800 is still based on "Midgard", which is no essential difference from the mali-t600 / T700. The basic "shader core" (XX core refers to it) is the same. It belongs to an optimized and upgraded version. At the same time, it focuses on optimizing the area, energy efficiency and adding some new technical features
adreno405 is a midrange GPU from Qualcomm, and its performance is better than that of adreno320