Home > News content

AMD/Intel hard hit! Looking forward to the 3nm Zen6 in 2024?

via:博客园     time:2019/8/17 15:01:28     readed:3107

Author: Wan South

With AMD's launch of the 7nm Zen2 architecture of the Ruilong 3000 and EPYC 7002 series processors this year, their competition with Intel is becoming more and more intense. Intel also launched the 10nm Ice Lake processor this year, using the SunnyCove core and the new CPU architecture.

In the X86 market, AMD and Intel have a great advantage to see both products. A good CPU architecture will be the basis of competition. Both parties have announced the CPU roadmap for the next few years.

Pushing a netizen witeken to compare the two roadmaps:


In this roadmap, he regards the CPU core of Intel's original 10nm architecture Cannonlake as the first generation. It has already disappeared with the cannon lake, and there is no actual impact. The Sunny Cove used by Icelake is the second generation CPU core. End with XXX Cove.

AMD has had Zen and Zen+ architectures before. This year is the Zen 2 architecture, just against Intel's Cove 2—— Sunny Cove.

Later, AMD has a Zen 3 architecture, and Intel is Sunny Cove's successor and Cove 3 generation Willow Cove.

In 2021, AMD expects to launch the Zen 4 architecture of the 5nm process, competing with Intel's Cove 4 core Golden Cove.

In the future, it is actually a road map that has not yet been officially announced. Intel is OceanCove, and AMD is Zen5 architecture. It only appears in media reports, and there is no clear information.

As for the Zen 6 vs. Cove 6, it's more speculative now, the shadows are gone, and the 3nm process is just the author's own guess.

In general, the AMD and Intel processor roadmaps made by Witeken are a bit interesting. After many years, the two sides can finally hard-hitting. Every two or two years, there will be a new CPU architecture, and no one will dare to slack off.

However, the specific process and time point are not very reliable. In the process technology (whether or not it is a digital game), Intel is lagging behind the 10nm node. If the 7nm and 5nm nodes do not accelerate to catch up, the competition between the two parties may be AMD will stay ahead if it is misplaced for a long time.


Of course, for Intel at the moment, the most urgent solution is the 10nm process capacity, and the other is the continuous shortage.

Since the beginning of Q3 last year, Intel's 14nm capacity has suddenly been in a hurry. As a manufacturing process with many years of mass production, the 14nm processor is out of stock and most people are puzzled, but the outside is in a hurry and can't be used. .

In order to solve the problem of insufficient capacity at 14nm, Intel announced the following improvements last year:

  • Capital expenditures will reach $15 billion in 2018, an increase of $1 billion from the beginning of the year. The money will be invested in 14nm fabs in Oregon, Arizona, Ireland and Israel to increase supply capacity by 14nm to meet market demand.
  • Intel's 10nm process is progressing, yield is improving, and mass production is expected in 2019.
  • Intel will adopt a customer-first approach and is working with the customer team to align customer needs with supply.

In addition, in terms of processor supply issues, Intel also adjusted the order, priority to meet the needs of high-end processors, including the Xeon and Core series, the priority of the low-end products to the end, and some of the chipsets originally scheduled to use 14nm process For example, H310 and B360 are now also produced in the 22nm process, and new chipsets such as H310C and B365 are introduced.

Intel's 14nm capacity shortage problem has been effectively alleviated this year. From the perspective of major platforms, it seems that few of the information about the eight-generation and nine-generation Core is out of stock. Intel executives have also stated that they will be able to solve this year. 14nm out of stock problem.

But what is the actual situation?Compal foundry Compal Technology said recently that Intel's processor out of stock problem will continue to exist in the second half of this year, and will affect the beginning of 2020, showing that out of stock problems are more trouble than previously estimated.


In addition, Ryzen 5 3550U appears on GeekBench.It looks very much like the 15W low voltage version of the Ryzen 5 3550H.

It is still a quad-core chip with eight threads and 4MB of L3 cache. Geekbench 4 Detection The Ryzen 5 3550 U has a reference frequency of 2.1 GHz and a maximum acceleration of 3.69 GHz.

Simultaneously,Ryzen 5 3550 U includes Vega 9, not Vega 8.

Considering that each Vega CU has a maximum of 64 shaders, Vega 9 should have 576 available. According to Geekbench 4, the Vega 9 operates at 1300MHz.

In terms of OpenCL alone, the 3550U's Vega 9 is 10% faster than the 3550H's Vega 8. Is it AFM that feels the pressure of Intel Ice Lake?



China IT News APP

Download China IT News APP

Please rate this news

The average score will be displayed after you score.

Post comment

Do not see clearly? Click for a new code.

User comments