Intel perhaps removes 10nm cpus for desktops [rumor]

Sources from the German website Hardwareluxx show us some strange rumors about the blue team. Intel is said to be abandoning its plans for 10nm desktop processors. This news baffles us a little, but the website relies heavily on its source, since it has leaked reliable information at other times.

According to internal Intel sources, the company could abandon its 10nm plans.

Due to the increasing pressure of the technological AMD , Intel has seemed to have taken a much more aggressive strategy.

According to a fairly reliable anonymous source, the company will go a long way in manufacturing 7nm transistors abandoning 10nm in the process. However, the end of this will take around two years, something that could cost the company a lot.

Despite Intel taking pride in its jump to 10nm, they have admitted that this new micro-architecture is quite problematic. If we add the low frequencies they manage to achieve and the low manufacturing volume they have, this did not seem to converge.

To compensate, we are looking at architectures like Comet Lake-S and Rocket Lake-S , which still hold 14nm . According to the source, the plan would be to increase the core counter in the user field, but without changing too much of the micro-architecture. This would be until 2022, when Meteorite Lake (or with another name) with 7nm transistors came to an end .

Of course, on the other hand, the new generation graphics units (Intel HD Graphics) do seem to be created at 10nm .

Micro-architectures like Tiger Lake-S would be completely discontinued, although their versions for laptops would continue to be developed. According to the statements of the same company in May, we can see on this mobile platform:

• A new CPU architecture New Intel Graphics Xe graphics engine New I / O technology

Processors for servers

In the server section, the Intel Xeon had also planned to use 10nm architectures, but it seems that they will not be affected as much. As the latest news from server processors was fairly recent, we believe the roadmap will not change too much.

Both Ice Lake and Cooper Lake will share LGA4189 socket and not only share this, but will also be on the market at the same time. Although this second still has 14nm micro-architectures, it will have complementary technologies (such as extensions AVX-512_BF16 / BFloat extra) to offer superior performance.

On the other hand, Ice Lake will have Sunny Cove cores, although it will abandon the BFloat16 extensions.

Apparently, Cooper Lake plans to offer us up to 56 cores, while due to the problems with the 10nm, there is only 28 for Ice Lake . So Intel will need other ways to develop 10nm transistors and currently doesn't seem to find any useful solutions. The source talks about a technique called Contact Over Active Gate (COAG), a new way of connecting doors, although it continues to cause problems.

After Ice Lake we would have the Sapphire Rapids micro-architecture built in 10nm ++, but we don't have much information about it.

After all, these are rumors, but we should have a watchful eye to know which of these statements are true.

And what do you think of this strange news? Do you think Intel would do well to follow these rumors? Share your ideas in the comment box.

Hardwareluxx font