Four years ago, if you asked about the laptop market and told others that you wanted to buy a laptop with a processor code at the beginning of a, the other side would often earnestly advise you, "don't buy AMD's notebook, balabalabala." finally, you would say, "I3 is silent for two seconds.".
But recently, I found that more and more friends were talking about AMD notebooks. Many of them even said directly, "the new ryzen 4000 is a little fragrant, and I will start when the new machine comes out. "
In the past four years, how amd has gradually reversed the reputation of notebook market? Whether ryzen 4000 series can turn amd around in notebook market? We need to start from the era of APU.
Bulldozer era: APU's sad past
In the strategy of AMD, if notebook computers want to achieve low power consumption, they must integrate traditional CPU and GPU, which amd calls "heterogeneous system architecture". The final implementation of the product is the familiar APU.
Amd has a good idea. Heterogeneous computing can not only make CPU and GPU play their own advantages, but also improve system efficiency. A single chip solves all the problems and is a more cost-effective choice. Users no longer need to buy a video card alone to get a good enough graphics, video and game experience.
▲ group photo of each generation of APU
Although the idea is good, but due to the low efficiency of bulldozer structure and the technology of our own wafer factory is not competing. Amd has been left behind by Intel. When Intel's 5th generation core entered the 14nm era in 2014, AMD's APU just used the 28nm technology, and has been using the 7th generation APU.
▲ in January 2017, the performance gap between FX 8800p and I3 6100h was not large
How miserable is the seventh generation APU? Take the highest end fx-9800p at that time for example, its performance is not even comparable to the I3 low-voltage processor in the next house.
AMD at that time in the notebook has sunk to the bottom of the chain of contempt, as soon as we heard APU, see the beginning of the notebook, words do not say, turn away. This kind of notebook also nicknamed "high achiever special purpose machine"
▲ supreme performance, have you been cheated by this logo?
Ryzen comes, amd Jedi survive
Fortunately, in 2017, amd finally launched ryzen processor with Zen architecture, and grofangde finally used 14nm technology, initially catching up with Intel.
The Zen generation
Compared with the previous bulldozer architecture, Zen architecture has a 40% improvement in IPC. With the CCX core strategy, Intel is forced to squeeze out toothpaste, bringing the notebook into the 4C / 8t era. In fact, the early ryzen is close to the same level of Intel products in terms of multi-core, but it still lags behind in terms of single core performance. GPU architecture has been upgraded to Vega, and the core display is higher than the entry-level single display level.
Time axis of Zen
However, due to the newer architecture, drive, game optimization and other problems, the sharp dragon APU with excellent scores still can not match Intel in actual applications and game scenes. Although Hewlett Packard, Lenovo and other computer manufacturers have launched ryzen APU computers, consumers who have been hurt by bulldozers have not yet slowed down, and they do not regard amd as an opponent of Intel. However, it is also recognized that AMD has a World War I power with Intel in the field of notebook computers.
But people expect more from AMD. They hope that AMD will bring pressure to Intel and promote its rapid progress. Intel has been using 14nm technology for five generations of CPUs in a row, and the new 10nm technology is still too late for mass production. It is necessary to know that the next cell phone SOC is ready to enter the 7Nm era.
Manufacturing process comparison between AMD and Intel
In view of Grosvenor's performance in developing the 14nm process, amd knows that when Grosvenor's 7Nm process is ready, its competitors will use 5nm. With this in mind, amd decided to give up the oil bottle of grofangde and turn its head to TSMC and Samsung.
AMD released the second-generation mobile Ryzen processor in the first half of 2019
▲ entering 4C / 8t Era
The performance of the second generation mobile ryzen is only 10% higher than that of the first generation, and the gap between the second generation mobile ryzen and Intel is further narrowed, even beginning to erode Intel's share in the middle and low-end market.
Under the old script, Intel should do his best to declare the ten nm standard and step AMD under his feet again. but Intel i5/i7 series yield is too low in 10 nm process. Due to shipping pressure, you can only shop 10 processors in nm low-power market. continue to use 14 processes on high-end i5/i7/i9 nm.
Comet lake-h of 14nm process
Though we know that the Intel 10nm process is stronger than TSMC 7 nm, even the density of TSMC 7 nm improved process N7.. but in the market of standard pressure processor, nm 7 process of AMD still have an advantage.
Ryzen 4000, the blade of AMD's counter attack on the mobile market
Amd just grasped the time difference that Intel's manufacturing process has not been fully popularized. At the beginning of this year, CES 2020 released the third generation of ryzen processor on the mobile side, that is, the ryzen 4000 series that published all the details the other day. It is a comprehensive evolution of AMD's mobile processor.
Ryzen 4000 all used nm 7 processes, compared to 14 processes, has obvious technological advantages.
In the previous three generations of desktop ryzen, Zen 2 has made comprehensive improvements in process, packaging, single-core and multi-core.
It includes inheriting SMT multithreading technology and adding new tag branch prediction, which reduces the false hit rate of branch prediction by 30%, and greatly improves the hit accuracy and efficiency. Avx2 instruction is also fully supported, with bit width increased from 128bit to 256bit, and floating-point performance directly doubled.
In the integer execution unit, the number of schedulers has increased from 84 to 92, the number of physical registers has increased from 168 to 180, and the number of physical registers has increased from 6 to 7, which further optimizes the efficiency and execution speed of the execution unit.
CCX module and I / O module are separated and manufactured by different processes to reduce production pressure. The if bus connecting the CCX module has also evolved to the second generation, improving parallelism, latency and processor performance.
After the whole operation, the single-core IPC of Zen 2 architecture is greatly improved compared with the previous generation, which changes the disadvantage of poor performance of ryzen single-core. Due to the modular design of CCX, multi-core is better than Intel
Of course, considering that the mobile terminal has more stringent requirements on chip area, the chips used in ryzen desktop before can not be used in notebook computers.
So amd abandons the chips package structure on ryzen 4000 series, encapsulates CCX module and IO module on the same chip, and if bus connects CCX and I / O. The core display module is composed of 8 groups of transverse cus. It is worth mentioning that ryzen 4000 core display uses Vega's 3D operation module, Navi architecture's encoding and decoding engine, and display feature module. Its performance is 59% higher than that of the previous generation.
▲ APU timeline
The bus bandwidth and intra core delay of the third generation mobile ryzen are significantly reduced due to no longer using chips. With the addition of 7 nm technology, the chip's energy efficiency ratio has been further improved.
Of course, single chip packaging also has its disadvantages. Due to the chip size limitation, CCX can only have two groups at most, that is, eight cores. Due to the need to make room for Vega, L3 cache and Cu units are also not available.
With the great leap forward of architecture and process, the performance of ryzen 4000 is 15% higher than that of previous generation IPC, the performance of single thread is 25% higher, and the unit energy efficiency is doubled directly. 15% of them are from the improvement of IPC performance, 17% from the improvement of design and 47% from the improvement of 7Nm manufacturing process.
For notebook, the performance of processor depends not only on its core number and frequency specification, but also on its TDP and unit energy consumption ratio. The lower the unit energy consumption ratio, the longer the boost time and the more sufficient the performance release.
Thanks to higher power consumption ratio and single core performance, the single thread performance of ryzen7 4800u with the same TDP of 15W is 25% higher than that of 3700u.
In terms of interface and channel support, ryzen 4000 with zen2 finally supports the latest Wi Fi 6, 4xpcie channels, nvme, usb-c, lpddr4x4266 memory, which is a short board of a's laptop.
At the CES conference, amd has compared Intel's latest 10 generations of low-voltage processors with 15 W low-voltage ryzen 7 4800u. According to AMD's test, 4800u has the advantage of rolling grade in the multi-core test of cinebench R20 and 3D mark.
On the standard pressure processor used in the mainstream game book, the 8C / 16t ryzen 7 4800h also easily defeated the 6C / 12t i7-9750h in the performance test. Even the i9-9880h, which is also 8C / 16t, was defeated by ryzen 7 4800h.
Now, amd has announced the ryzen 9 series with top performance, including 4900h and 4900hs, both of which are 8C / 16t, with more Cu units, improving the basic frequency and boost frequency, directly benchmarking core i9, challenging the position of intel mobile market leader.
According to the official test results of AMD, ryzen 9 4800hs (35W TDP) is better than i9-9880h of 45W TDP in cinebench R20, video decoding, image rendering and other multi-core projects.
Amd has launched an all-round attack and the mobile market will change
Step by step, AMD is finally waiting for this day. In the past two years, we have seen the rapid rise of AMD in the PC market, with its share rising step by step. However, in the notebook market, amd still can't threaten Intel's market position in the light and thin notebook and high-end performance notebook because of its poor reputation and backward performance power ratio, so it can only attack the middle and low-end market by price performance ratio.
The advent of the Ryzen 4000 series finally changed this situation, Zen 2 the structure of TSMC 7 nm process, so that the mobile three generation Ryzen in the case of unchanged performance, power consumption ratio is better. The excellent paper performance not only gives consumers "AMD Yes」," but also gives manufacturers more supply chain bargaining power, shaking Intel's" arbitrary "position in the mobile market.
In the past, only tens of thousands or even tens of thousands of notebook computers were eligible for the top performance of 8-core 16 wire processors. Now ryzen puts this specification within 10000 yuan to promote the progress of the industry, and at the same time, it does bring more benefits to consumers.
A single branch is not a spring, a garden full of colorful spring.
Now AMD with ryzen 4000 in hand has planted flowers in Intel's back garden. How Intel will deal with this year will be determined in the second half of the year.
Before the war, I want to applaud AMD: "amd yes! "