The toothpaste factory is also frustrated. The newly released processor is like the toothpaste tube is blocked. It is not only hard to squeeze, but also the price is high.
There is even an embarrassing situation that the 11 generation CPU on the desktop is not as good as the 10 generation CPU. ▼
Just when everyone thought that this former semiconductor factory was gradually losing its competitiveness, the new thing announced by Intel on the architecture day two days ago made Tony sit up in his dying illness...
In this special event, Intel not only brought the latest 12th generation CPU architecture alder lake, but also introduced in detail the previously disclosed Intel arc Series High-Performance independent graphics cards.
On the server side, sapphire rapids has been upgraded, and the professional graphics processing has updated the high-end computing card Ponte Vecchio.
Xeon processor and Professional Computing katoni will not be introduced much. This time, we will focus on the 12th generation CPU equipped with a new x86 kernel and Intel's unique display.
In fact, the news that Intel wanted to change to the new architecture alder lake was disclosed at CES held online at the beginning of this year.
At that time, it was said that the alder Lake architecture would use a core design of the same size as the current mobile phone arm architecture, which could bring better performance and lower power consumption.
If the name is not changed, this is mostly the parameter ▼ of the subsequent flagship chip i9-12900
The Intel 7 mentioned here does not refer to the 7Nm process, but the naming of a set of chip process nodes, which is essentially 10nm.
Although it sounds like a bit of a cover up, Intel's 10nm wave does not lose any of TSMC's 7Nm process, and even the density of transistors integrated on the chip is a bit higher.
The most vigorous promotion of alder lake this time is the improvement of the small core.
They directly compare the 12th generation small core with the 10th generation skylake architecture large core. Under the same performance of one-to-one single thread, the energy consumption is reduced by 40%.
Under the same energy consumption of four to two multithreading, the performance can be improved by 80%...
Simply put, the 12th generation of small nuclei, in some cases, have been suspended from the existence of the 10th generation of large nuclei.
For consumers who save machines, if they can buy u with the same performance at a cheap price, both Intel and AMD can naturally yes. Tony thinks this is another significance for Intel to improve its small core.
However, compared with grace of grace Mont, the large core golden Cove is less golden.
Compared with the large cores of the previous 11 generations of Intel processors, the performance of the 12th generation is improved by 19% at the same frequency, and the delay is reduced in the application processing of large data and programs.
On the other hand, compared with the previous generation of AI computing, the new amx unit can improve the original computing performance by 8 times.
In addition, alder Lake architecture also supports new standards such as ddr5, PCIe 5.0 and thread director. It seems that the slots and I / O interfaces on the motherboard will change with this upgrade.
It's reasonable to say that this generation of interfaces can hardly be completely drained of bandwidth. Intel has long been laying out for the next generation, which shows that they admit that they have lost a stubble. It can be considered that they have stabilized their mentality...
At best, it's called war in the future.
Tony felt that Intel's use of a radical small core with a stable large core had vaguely revealed their market goals.
The conservatism of the large core means that the absolute performance of the desktop CPU may not take a big step, but alder Lake's new low-power architecture is very useful in some products with high requirements for CPU power consumption such as the mobile terminal.
Tony guesses again. Whether the new Intel 12 generation u is easy to use depends on the face of the old man Microsoft. If something goes wrong and the windows system is not optimized, it may still repeat the mistakes of the 11th generation.
Yes, you heard right. Intel, a small blue factory with thick eyebrows and big eyes, will also start making independent graphics cards.
When he first heard the news, Tony had a little expectation, but after reading the actual release, he felt that something was wrong.
First, on the architecture of independent graphics card, Intel uses a new Xe HPG microarchitecture.
As the first generation product, Alchemist SOC (literally translated as alchemist SOC, except the end), which is included in the architecture, has a combination of 16 vector engines and 16 matrix engines, providing excellent expansibility and computational efficiency.
In terms of software, none of the old APIs and engines such as unreal engine, DirectX, Vulkan and unity has been pulled down. At the same time, it also supports the light pursuit of DirectX and Vulkan.
Obviously, Intel's independent graphics card is aimed at players and games.
In addition, the arc graphics card also supports a technology called XESS, which can realize the oversampling AI frequency increase technology through the instructions of dp4a to improve the frame rate and stabilize the image quality and resolution.
DLSS: just give me my name. ▼
The name and function are all DLSS of NV family...
What's more interesting is that not long ago, a technology leader who was responsible for NVIDIA light pursuit and DLSS project switched to Intel as deputy director of R & D department...
In this way, it seems that things can be explained...
However, Intel has not been able to recruit such a big player in the process technology of exclusive SOC. They directly adopt TSMC's N6 process node...
Kevin Zhang, senior vice president of business development of TSMC, welcomed the cooperation between Intel and TSMC. ▼
In other words, Intel is currently unable to use its own chip production line to manufacture graphics card chips. Their participation obviously can not alleviate the current supply and demand of graphics cards.
TSMC's chip production capacity is limited. For the graphics card, NVIDIA, AMD and Samsung will share, and I don't know how much will fall on Intel.
Tony thought that Intel's graphics card would make players more likely to buy graphics cards. Now it seems that it's hanging.
Although there may be problems in production capacity in the future, the hard quality of Intel's graphics card itself is still online. If it can be sold smoothly at that time, Tony will try one.
In fact, everyone sees Intel's situation now.
Finally, I found that AMD has occupied the throne of consumer CPU, so I can only consolidate my mobile market and work harder.
Tony didn't know what would happen, but at least they finally lost their old tube of toothpaste and started a new journey.
- THE END -
Original link:negative comment
User comments