P4 calculation force

1. This is just one aspect. Originally, people blindly believed that the higher the main frequency, the better the performance because they did not understand the structure of CPU. This was the case at the beginning of the development of CPU. Intel was the leader in this aspect and was fully ahead of AMD. Later, with the development of CPU, the technology of CPU core has been improved continuously, and the main frequency is no longer the only yardstick of performance. AMD's K8 series, such as 3000 +, 3200 +, and so on, only have a core frequency of about 1.8g and 2.0, but its core is more efficient, equivalent to the performance of Intel's 3.0 and 3.2 main frequency CPU< Take the current mainstream procts as an example:
compared with the previous generation p4630 CPU with a dominant frequency of 3.0g and e2160 with a dominant frequency of 1.6g. 630 is a single core, 2160 is a al core, using the latest core architecture of core 2. The pipeline number of core micro architecture (the number of processing times of an instruction from acquisition to execution) is much shorter than that of P4 era micro architecture, only 14 levels. However, P4 has a bloated pipeline series up to 30 levels. The rection of the number of pipeline stages can make the scheling work more efficient, thus achieving lower power and higher performance. This is beyond the reach of P4. In addition, the new CPU also supports more new instruction sets, such as ssse3, which can provide higher efficiency
so in this era, CPU frequency is no longer the only yardstick of performance..

2. When we use different computers to calculate pi, we will find that one computer's calculation is more accurate than the other. Or when we are playing a gunfight game, when a bullet hits the wall, a wall will peel off. The performance of the same scene on a computer may be very rigid and artificial; And on another computer, it will be very vivid, even close to what we see in reality
everything we have seen above comes from the "floating point operation function" added inside the CPU. Floating point operation ability is an important index of multimedia and 3D graphics processing related to CPU. There are only two floating-point execution units in P4, and one of them has to process FADD at the same time

3. When we use different computers to calculate pi, we will find that one computer's calculation is more accurate than the other. Or when we are playing a gunfight game, when a bullet hits the wall, a wall will peel off. The performance of the same scene on a computer may be very rigid and artificial; And on another computer, it will be very vivid, even close to what we see in reality
everything we have seen above comes from the "floating point operation function" added inside the CPU. Floating point operation ability is an important index of multimedia and 3D graphics processing related to CPU. There are only two floating-point execution units in P4, and one of them has to process FADD at the same time

4. Are you kidding downstairs? You say that the old p43.0 floating-point computing power than the al core Celeron e3500! The same as L1! The frequency of p43.0 is higher. Finally, let's look at the architecture, because the main two p43.0 and e3500 are higher than each other. Even considering the architecture, e3500 is still slightly inferior to p43.0! Sb do not understand, do not pretend to understand
the floating-point computing capability of the server is related to the architecture of the server itself, the number of cores, performance (including the number of CPUs, performance, number of graphics cards, performance), memory size, performance, type, and the degree of system optimization
I don't know how much floating-point computing power e5-4640 can be configured! Because there is no solution to this problem! It's said that it's related to many factors, but it's meaningless if you just give a CPU! You know, it's all heterogeneous computing servers now! Thank you
floating point computing power of high performance server is not fixed! There are many reasons, such as architecture design, power supply, system optimization, even the connection quality between moles, core dynamic acceleration technology and so on! Generally, they are divided into theoretical value, maximum value, minimum value and even general value! Look at the HPC ranking
as for the calculation, the theoretical value is obtained by multiplying all cores by a coefficient (determined by the architecture and other factors)! There are many factors in the actual value, which can only be measured after assembly
I really don't know the floating-point operation value of e5-4640, I don't care!

5. Operations can't be counted as 1 + 1.. Because the result will exceed the output bandwidth
generally, complex operations are used for testing.. For example, in Fritz's chess, the results are output according to the number of steps per second and the multiple of P4 computing power, which is also inaccurate because the computing efficiency also involves software programming.

6. And long double. In addition to the operation of very large and very small floating-point numbers, which needs special processor support or software to simulate, double is usually enough. I have forgotten which (double or long double) type can represent the number of atoms in the universe. It is possible that a lot of numerical calculation will be used in 3D drawing and drawing operations. In addition to the special floating-point arithmetic unit, how to design the cache, how to arrange the lines, and what pipeline structure is used may have different effects. For those who are not specialized in technology, we can only listen to them for the time being. Just like the car review program, we say what the performance of a car is. Unless you continue to understand the relevant technology for many years, it is difficult for us to really understand, especially the significance of the business words.

7.

take the flagship SUV Mocha, which represents the brand new development direction of wey, as an example. At the end of 2020, great wall just released lemon hybrid DHT technology and intelligent travel solution coffee intelligent drive, which will be the first to be carried on this vehicle

especially automatic driving technology. Some people compare it with the new car et7 released by Weilai on NiO day. For example, on et7, which will be officially launched a year later, Weilai will carry lidar for the first time. Wey is already equipped with German ibeo next 4D solid-state lidar on Mocha, which will be launched in a few months

for another example, the high-tech 8155 smart chip used in mocha is also known by netizens as "the most powerful computing power on the surface". It is said that this chip is the first digital cockpit chip made by 7Nm technology. It has a new learning neural network processor NPU, and has AI Artificial Intelligence deep learning ability. Compared with the previous generation, the performance of CPU and GPU has been improved by 8.5 times and 20 times, respectively. Compared with the car chip of ideal one and Xiaopeng P7, the performance has been improved by 3 times

advanced hardware is only one aspect. Previously, at the coffee smart drive press conference, great wall also showed its own research and development of Noh (navigation on highway pilot, Chinese name "high-speed automatic pilot assisted driving") for full scene coverage of high-speed and urban expressways. This system will also be carried on mocha

this also makes wey the first traditional car company to launch similar functions after Tesla's NOA, Weilai's NOP and Xiaopeng's NGP

2

seek the optimal solution between mileage anxiety and environmental protection problems

in the past, in the impression of most people, intelligent vehicles are often equated with pure electric vehicles. However, at wey's brand new launch, lemon hybrid DHT was used as another major technical support to create a new label

according to song Dongxian, vice president of Great Wall Motor Technology, although the external acceptance of electric vehicles is getting higher and higher, there are still some inherent defects in the use of electric vehicles e to the imperfect infrastructure. Through the lemon hybrid DHT technology, Great Wall hopes to solve the mileage anxiety of electric vehicles and optimize the shortcomings of the original HEV and PHEV technologies criticized by consumers, so as to be closer to the actual needs of consumers

for example, in view of the problems of slow charging and short driving range of PHEV models in the past, great wall can achieve the maximum pure electric driving range of 200km through fast charging technology and increasing battery pack, which fully meets the daily commuting use

as for HEV models, song Dongxian said that compared with similar Japanese procts, they also have some advantages“ Wey brand with lemon hybrid DHT model, the biggest advantage is fuel saving, energetic, fast response

for example, unlike some Japanese hybrid models which are suitable for urban driving but lack of power advantage on high speed, it is said that Mocha's 1.5T model, equipped with P4 motor and DHT, has a total power of 320 kW and strong power

generally speaking, through the application of lemon hybrid DHT technology, wey hopes to find the optimal solution between mileage anxiety and environmental protection. It weakens the trade-offs made by consumers e to power system problems when buying a car

3

traditional car companies also began to pay attention to "people"

"the arrival of mocha will promote traditional car manufacturing to enter the forefront of the intelligent era, and there will be no" new "or" old "forces from now on."

wey has high expectations for the brand new flagship model launched after the brand renewal

However, apart from the label of intelligent technology, how to treat users is also a big difference between the "new" and "old" forces

In this regard, Qiao Xinyu, deputy general manager of wey brand marketing, frankly said, "the biggest difference between the new force of car building and traditional car building enterprises is that they focus on users and make friends with users.". However, after recognizing the difference, "wey brand has been practicing this concept in recent years," he said

In wey's opinion, the "new" concept of "new generation intelligent vehicle" proposed this time is mainly reflected in "people-oriented"

according to Chen Xianling, digital executive officer of Great Wall Motors, in 2020, Great Wall established a proct digital center, under which a user experience center was set up to integrate user experience into the proct development stage

take some user scenarios of mocha as an example. In addition to the functions of VIMS face recognition, Hwa + AVP vehicle and environment perception, L3 level high-speed automatic driving, gesture moving and so on, mocha's intelligent service system can also monitor the vehicle status at any time, remind and help the vehicle owner to make an appointment for maintenance. If you have any questions in the process of using the car, you can always connect with experts who understand the car to answer them

in addition to the vehicle itself, through the owner's app, the person in charge of each segment of wey brand will also communicate directly with users every day. Users can also participate in the co creation of car body color, car configuration and other issues

in Qiao Xinyu's words, the operation idea of wey brand is "changing from managing procts to managing users". In their planning, the concept of "people-centered" will continue to expand in the future

4

written at the end

"in the past 30 years, Great Wall Motor has built a successful national shenche haver. In the next 30 years, Great Wall Motors will build a successful wey. "

It is said that this is the expectation of Wei Jianjun, chairman of Great Wall Motors, for the future development of the brand

nowadays, great wall has not disclosed too much about the specific confidence of mocha models. However, from hard core intelligent technology to humanized proct building concept, wey after brand renewal has a relatively clear direction compared with before

According to wey's plan, mocha will be available at the end of March. At that time, the new effect of wey brand will be fully tested by the market< br />

8. There are many computing power
among them, floating-point computing is one of them
floating-point computing power has a lot to do with the first level cache
for example,
amd
CPU's floating-point computing power is much higher than inter
dominant frequency is a very important point, but not all
if dominant frequency decides everything, AMD will declare bankruptcy
but Inter is afraid of AMD
amd 2600 + uses super -- pi to measure The result of the test is comparable to the speed of p43.0, but the main frequency of 2600 + is only 1.8g
but the CPU of the same architecture and cache is determined by the main frequency
it can tell you an undeniable fact that the CPU of
p3-866
is faster than that of P4 Celeron 1.8g

9. Is single ratio floating point too much? The maximum floating-point computing power of p4cpu is less than 20 gflops
even the core is only more than 100. You should know that a single "cell" processor has 8 cores

and the "cell" processor is 256gflops at 4GHz, while the real PS3
processor is only 2.8GhZ. At this time, the floating-point computing capacity is only 115.87gflops