[PConline Miscellaneous] If the exhibition is a "must fight" for various manufacturers, then this year's CES exhibition, for AMD and Intel, may be the pinnacle of the leaders of the martial arts genre.
Now let's take a look at how wonderful the so-called "fairy fight" is.
7nm breaks into mobile terminal, AMD completes short board and declares war on I / N
The layout of AMD's desktop products this year has been very perfect. Consumer-grade third-generation Ryzen, enthusiast-level three-generation thread tearers, and 7nm Navi 10 graphics cards, whether it is facing the gun with Intel or Taiji with NVIDIA, are all very easy to go. It was all a cry of "AMD Yes!".
But Su Ma is obviously not only satisfied with this, this time CES, has become AMD's big show.
At CES2020, AMD and Intel developed the conference in the same place on the same day. The so-called is not that the enemies don't get together. The two factories A and I simply face each other and take out their new products.
▼ Ryzen Threadripper 3990X: The Dominance Benchmark of Audiophile Processors
When starting the third-generation thread tearer, it is the most exciting time. The only thing you want to see is the moment when the toothpaste factory is surpassed: the 3960X, which is just a thread tearer, can press the i9-10980XE to the ground to rub.
The 3990X issued this time is just a brand product. Let's take a look at the parameters first.
Threadripper 3990X has a horror specification of up to 64 cores and 128 threads. The total cache size has reached 288MB, three buffers of 256MB, a base frequency of 2.9GHz, and a BOOST acceleration frequency of up to 4.3GHz.
CineBench R20 runs, the score of 25000+ is approximately equal to twice that of the 24-core 48-thread Threadripper 3960X, and approximately five times that of the 8-core 16-thread i9-9900KS.
There is no need to question the performance. It is currently the strongest enthusiast processor. People are still staring at the rendering progress bar. 3990X said that it has already finished work.
However, so many cores reflect a fatal problem: the number of cores is a bottleneck.
The number of theoretical cores is doubled, and the performance is almost doubled, but it is obvious here that the 3990X score of 64 cores is only about 1.5 times the 3970X score of 32 cores. AMD also claims that in the later Zen 3 and Zen 4 architectures, the maximum number of cores will not be greatly increased, so it can be proven on the side that the number of cores has now been added to a critical value. The issue to be considered after AMD is how to strengthen the unit like a toothpaste factory. Nuclear performance.
▼ RX 5600 XT: Before Navi 21 is released, it will be enough for NV to drink
I thought that AMD would take out the Navi 21 killer to ignite the audience, but in the end it just came up with a new mid-range RDNA card RX 5600 XT, with some disappointments. The specifications are the same as the RX 5700, which castrates the bandwidth of the video memory. There is a sense of old yellow knife in the shadows.
RX 5600 XT is the fourth 7nm graphics card with RDNA architecture. The core has 36 computing units, 2304 stream processors, a game frequency of 1375MHz, an acceleration frequency of up to 1560MHz, a total performance of up to 7.2TFLOPS, and a TDP of 150W. Power optimization is excellent, reducing power consumption by 33% compared to the previous generation RX 590, but improving performance by 42%.
To achieve your goal of defeating yourself, of course, you have to compare it with friends. Compared with the GTX 1660 Ti, the RX 5600 XT has achieved a comprehensive crushing of performance whether it is a 3A masterpiece or an online game.
Combined with the above price, the price of 279 US dollars is equivalent to about 2,000 yuan. Currently, we have received news that the domestic RX 5600 XT e-commerce initial price is 2099 yuan. Now most of the GTX 1660 Ti non-public models sell for 2099 yuan, with stronger performance and the same price. I want to know who to buy with my fingers.
▼ Ryzen 4000 Mobile Processor: Take off the shame of friends
Playing multi-core on mobile can only be Intel's one? AMD: I have 8 cores and 16 threads.
The 7nm Zen 2 mobile-end processor will bring dramatic changes to the current notebook market.
Although the code name of the processor was not mentioned at the meeting, it is the mobile-end processor codenamed Renior that has been exposed before, 7nm CPU + 12nm GPU (Vega).
Ryzen 7 4800U, low-voltage flagship, 8-core 16-thread design, 12MB cache, base frequency 1.8GHz, can boost BOOST to 4.2GHz, and such a strong specification requires only 15W TDP. The Vega 8-core display integrates 8 computing units, 512 stream processors, and a frequency of 1750MHz.
Compared with the previous generation, the Ryzen 4000U series has reduced power consumption by up to 20%, energy efficiency up to 2 times, energy-saving state entry and recovery delays up to 5 times, and additional support for LPDDR4X memory.
AMD compared the Ryzen 7 4800U with Intel's 10th-generation mobile flagship i7-1065G7 (4C8T). The former's CPU single-thread performance is 4% ahead, and the CPU multi-thread performance is 90% ahead.
Ryzen 7 4800H, the standard flagship, also has 8 cores and 16 threads, but the frequency is significantly increased: the base frequency of 2.9GHz, the BOOST frequency reaches 4.2GHz, and the TDP is 45W.
On this processor, AMD brings everyone a "live" operation: compared with the desktop i7-9700K with the same 8-core performance, the content creation performance Ryzen 7 4800H is 7% stronger than the i7-9700K, and the game performance 10% stronger.
With 8-core and 16-thread specifications plus high frequency, the Ryzen 7 4800H is indeed very suitable for gaming notebooks, and AMD has also opened a "green channel" for Asus: the new Zephyrus G15 notebook will get the exclusive right of 35W TDP Ryzen 7 4800HS for half a year .
Asus is a real figure in the field of high-end games, and AMD is also very wise to open its market to its own processors with such a big sign.
To be honest, I feel that AMD is now more and more knowledgeable in marketing. Intel ’s long-term mobile cake has also had to give AMD away. Imagine how powerful AMD will become.
▼ The absence of them is more worthy of our expectations
Although there are quite a lot of things released this time, some of the things predicted before the press conference have not yet appeared. Here is a brief summary of the new AMD products that can make us hot.
1.PS5, Xbox Series X specifications
This should actually be categorized into Sony and Microsoft for release, but how to say it is related to AMD, there must be something new when it comes out.
There was related news before, saying that both the PS5 and Xbox Series X used AMD's Zen 2 architecture processor and RDNA v2 architecture graphics card . This time Sony released an icon, which is a bit boring.
2.Zen 3 architecture fourth generation Ruilong, EPYC
The release of Zen 3 was completely unexpected. After all, Zen 2 had only been out for half a year, and it was obviously inappropriate to get into the ground so soon.
According to the previous news, the third generation EPYC will be launched in 2020, codenamed Milan, the CPU architecture will be upgraded to Zen 3, and the process will be upgraded to TSMC's 7nm +, which is the second generation 7nm process. The enthusiast market is also expected to launch the Ryzen Threadripper 4000 series, code-named Genesis Peak, which is also a 7nm + process and Zen3 architecture. The three-generation Zen 2 7nm Ryzen series on the desktop will also be upgraded to the fourth-generation Zen 3 7nm + Ryzen. Matisse will be upgraded to the code name Vermeer, Ryzen 4000,
Navi 21 is also a new code card series that is very interesting this year. Now AMD does not have a graphics card that can threaten NVIDIA flagship models. Navi 21 flagship is the most promising model that can compete with RTX 2080 Ti. .
The Navi 21 flagship model can double the core area of the RX 5700 XT. The RDNA2 architecture will add a light-tracking unit. The video memory is equipped with a large capacity GDDR6. Professional graphics cards are also expected to match SK Hynix's latest HBM2E, which has a bandwidth increase of about 50%.
Intel embarks on mobile market defense battle, AI becomes high-frequency word
The 10th-generation Core Commet Lake-S on the desktop is expected to be released. There are no other DIY players disappointed. However, some are happy and sad. Consumers who like laptops on the mobile end are very enjoyable at this CES. Whether it is AMD or Intel, both regard the mobile market as the main battlefield. After talking about AMD earlier, let's talk about the trend of the semiconductor giant Intel, which accounts for a large share of the mobile processor market, at CES.
▼ Next-generation mobile processor "Tiger Lake"-to make PCs "smart"
What the average consumer sees is what happened in an industry in recent months. Participants in related industries may predict and think about the development trend of the industry in the next year. As a company, their thinking dimensions and plans tend to be 3 years. Or 5 years or even 10 years.
When AMD released the first generation of Ryzen, when Ryzen did not have a mobile model, Intel might have thought of the impact Ryzen would have on mobile in the future. Then in front of Intel is a very simple multiple choice question. On the one hand, although there are many hard-core users, the DIY market has a small total market, and on the other hand, the mobile terminal market, which has a market share of more than 10 times, fools know how to choose.
When the AMD 7nm processor achieved great success on the desktop, it unexpectedly began to attack the mobile market with more money. The performance and energy consumption ratio of Ryzen 7 4800u and Ryzen 7 4800H released by CES is very amazing, but In recent years, Intel, which has been focusing on mobile processors, is not in the stage of passive battering as it is on the desktop. The backhand officially announced the next-generation mobile processor "Tiger Lake" .
Tiger Lake is the successor to the current first-generation 10nm processor Ice Lake. It uses a 10nm + process . The new processor has ambitious goals. Intel claims to rely on it to "redefine the mobile platform."
Tiger Lake is a series of products aimed at thin and light notebooks. It comes with a core display and a new Willow Cove microarchitecture for the CPU . It is reported that the new architecture IPC will increase by 10% compared to Sunny Cove on Ice Lake. Intel ’s nuclear display, which has been ridiculed, also uses the new Xe architecture , which shares the same origin as Intel ’s future independent display DG1. We will talk about this DG1 later.
The new processor is worthy of praise is the integrated Thunderbolt 4 control chip, which means that notebooks equipped with this series of products can use Thunderbolt 4 interface. This is also the first time Thunderbolt 4 has been announced to the world, but it is confusing that the new interface has the same bandwidth as Thunderbolt 3, and it is still unknown where it will improve.
AI is the current buzzword, and it is also a new performance dimension that Intel often mentions when new processors are released recently, this time is no exception. Tiger Lake can use the DL Boost deep learning acceleration technology unique to previous server platforms. Under the acceleration of Xe nuclear display and Low Power Accelerators, the new processor AI performance is doubled compared to the old processor . It can be used for Picture noise reduction, audio and video processing, and more.
The improvement of AI performance may make our PCs more intelligent and smarter. The scene where the computer becomes your personal steward in science fiction may be being realized. From here we can also see that when the traditional CPU performance improvement encounters a bottleneck As an industry leader, Intel is working to re-plan the direction of future CPU development.
It is reported that Tiger Lake specific products will be released later this year, let us wait and see.
▼ Graphics card market enters "RGB" era-DG1 alone reveals the veil
Intel's independent DG1, which has been brewing for more than half a year, finally lifted the mystery, and Intel performed for the first time at this CES.
The graphics card is the same as Xiaobian expected before. It is positioned as an entry-level graphics card and focuses on mobile platforms. This demonstration also uses a notebook for demonstration.
Let's take a look at the official introduction of the product first. Intel DG1 is its first discrete graphics product for consumer platforms. It is based on the new Xe GPU architecture and integrates a powerful media and display engine. It is designed for power-optimized platforms ( (I.e. focus on energy efficiency), and optimize for games and content creation.
If you want to find a competitor for DG1, then it should be the MX250 that basically dominates all the thin and light display positions . In terms of performance, DG1 may not be as good as MX250. Consumption ratio will be higher.
In addition, another major reason for Intel to develop GPUs is the deep learning and supercomputing markets. It is well known that GPUs that are good at parallel computing in deep learning computing are tens of times more efficient than CPUs that are good at serial computing. If there is no GPU supporting, Intel has no complete solution for AMD compared to AMD, which is at a disadvantage. status.
Therefore, Intel ’s GPU series Ponte Vecchio prepared for supercomputing has also surfaced at this CES. Ponte Vecchio is manufactured using 7nm process and integrates EMIB and 3D Foveros packaging technology. It supports HBM high-bandwidth memory and CCIX high-speed interconnection protocol. It is used in the next generation of American supercomputer "Aurora" (Aurora), six-way parallel, and the future can be extended to Xeon with codename Sapphire Rapids.
▼ Intel pushes dual-screen notebooks-is this the direction of future mobile platforms?
The timeline goes down, I believe many friends still remember that there was news that Intel will work with Microsoft to develop standards for dual-screen notebook products.
Looking back, looking back at the whole of 2019, it is not difficult to find that in 2019, there are many dual-screen books appearing in our field of vision, and there are also many large-scale manufacturers participating, and each manufacturer is like a single Being independent, each company has a different definition of dual screens.
Now, Intel will lead various manufacturers to work together for the continuous development of the dual-screen concept in 2020.
ThinkPad X1 fold:
Dell dual-screen notebook:
It's not difficult to see that after the laptop becomes a dual screen, there are folding screens and dual screens, but no matter what, they look a bit like a foldable tablet. The dual-screen notebook can not only present more content, but also have higher usability and portability compared to tablet computers.
The dual-screen version with a folding screen design is more exciting, but I don't know how the on-screen keyboard experience is?
▼ 56 nuclear beast disclosure-Intel is also moving towards the nuclear road
In the face of the aggressive nuclear war offensive of friends and merchants, Intel is not willing to be outdone. Finally, it disclosed its own next-generation server Zhiqi processor Cooper Lake, which is still built with 14nm, but the AI performance has increased by 60%, and the highest specification is 56 cores. 112 Thread will be available in the first half of this year.
AI performance is an important part of the new processor. It can also be seen that Intel has started to look at the improvement of AI inferred performance when the traditional CPU performance cannot be significantly improved. This is actually correct. Intel predicts that the market size of AI chips in 2024 will reach 25 billion US dollars, which is the next outlet of the semiconductor industry. It has only a short relationship with our ordinary consumers.
Summary: In addition to these products mentioned above, foreign media have revealed that Intel will also announce the information of the 10th generation of Core compression processor Comet Lake H at CES. The maximum 8 cores and 16 threads will reach the maximum core frequency of 5.0GHz, but the deadline At the time of writing, Intel has not officially announced information about this series of processors.
In general, from this Intel ’s big move at CES, we can see that the blue giant is determined to stay on the mobile terminal market, and the notebook market defense war has begun. I do n’t know if you are optimistic that AMD can recover. Carve the miracle of the desktop market, or do you believe in the strength of the industry giant Intel?
For 20 years, can AMD reproduce the glory of x64?
AMD's actions in 2020 will be very large. Except for the APU that is behind the default generation architecture by default, other product lines will be gradually upgraded to 7nm + Zen 3 architecture processors to increase frequency and speed.
From this point of view, it is not difficult for AMD processors to reach 5GHz. There are many reports that Zen 3 can increase the frequency by 100-300MHz. The acceleration frequency of Ryzen 9 3950X has reached 4.7GHz, and 300MHz can easily reach 5GHz.
Relying on the advantages of Skylake's high-frequency and strong single-core performance, Intel has ate so much tea for a long time, and it's time to wake up with twelve minutes to deal with it. This time, it has not released a desktop processor, which makes people have to doubt the 10th generation Core Whether there is a problem of dystocia.
However, Intel continues to rely on high-frequency meals, and can be restless in a short period of time. The previous news broke that the 10th generation Comet Lake Core series will increase the frequency to 5.3GHz and increase the number of cores to 10 cores. Intel does not know how many + signs by 14nm, and should be able to achieve 10nm production.
At that time, AMD relied on the AMD64 instruction set to burn Intel's Itanium IA64 backyard. At that time, AMD had the confidence and the right to speak. The x64 processor that ruled for a while was later overtaken by the Core series, so that the toothpaste factory has been at ease.
This year, 20 years later, will this be a turning point for AMD? Back to Sohu, see more