Chapter 644: Making chips is not a problem
I guess someone said that the load of Qingda Lazy God has reached the upper limit, won't it cost money to rent a server?
Or just a few more servers?
The external web pages of the lazy god system have actually been transferred to the rented server a long time ago, but the rest really can't be transferred to the rented server.
Lazy worm, little lazy and little million, called super programs by Nanming, because these three programs, must run on ultra-high-performance supercomputing, even if it is distributed cloud computing, it will seriously affect its performance, want to rent a server, unless you rent a few supercomputers, and it is a long-term lease, this has to rewrite its code, so that it can adapt to different computing environments of different supercomputers, and human and material resources do not know how much to consume.
As the first "super program" of the lazy god system, it is also the reason for the birth of the lazy god supercomputer of Qingda Lazy God, and the "lazy worm" is the core and starting point of all artificial intelligence in the lazy god system, which comes from the hands of Tie Dan.
At the beginning of the birth of this terrifying program, known as the Vision of God, if Nanming did not seal it, it might be possible to make all the intelligence services around the world unemployed. If it gets out of control a little more, the closely watched future world of science fiction may arrive sooner. Compared to it, the prism is a hair.
The good thing is that the program is dormant most of the time and is not enabled.
The first real artificial intelligence core "little lazy" in the lazy god system comes from the customer service program written by Tie Dan for his lazy life, and then Dacheng Yu Li Xiang's constantly updated algorithm.
It can be said that Tie Dan, a workaholic, likes manual labor, likes to do rough work rather than mental work, and at this time he doesn't know where to help people deliver food, do housekeeping, and clean the toilet, and single-handedly create the myth of artificial intelligence of the lazy god system.
Then, there is the application of artificial intelligence in the professional field, "Doudou" and its upgraded version "Xiaowan". Although it is the creativity of Wang Ziyang, they all have a lot of code contributed by Tiedan, as well as a common core.
On the territory of lazy god artificial intelligence, although Wang Ziyang is a rising star, it is still too early to surpass the bull man of Tiedan.
Not to mention the iron eggs. At present, Li Xiang's contribution is also much greater than that of Prince Yang.
If this is a pie chart, Tie Dan alone dominates half of the country and eats half of the pie, and the rest Li wants to take it away, leaving Prince Yang with a small corner.
But if you talk about the impact of public opinion, these two people can't compare to Wang Ziyang.
Wang Ziyang released the beta version of Xiao Wan's scientific research assistant in order to combat Wan scientific research, and the effect of the blow was good, but it spread too much. Now the number of users has exceeded the 200,000 mark.
In the key diffusion unit, there is one person in the Kyoto University, and there is basically one person in Qingyang University.
As for other universities and scientific research institutions, there are many people who want to try it, both at home and abroad.
The beta version of the membership was originally used for testing, but the result was also exploited by others, and a lot of registrations were registered, which took up more resources.
Of course, the most extensive features of the beta version though look powerful. But in fact, the resources spent are not particularly large. After optimization, it consumes about the same resources as a life assistant for "lazy life".
But the full-featured "Xiaowan assistant" is a completely different concept. The complexity of the problem and the data retrieved are not at the same level. Not to mention the engineering version of Xiaowan's assistant, as well as the camera, the function of action analysis.
It's much harder to get a computer to understand a picture than it is to get a computer to understand a language, let alone make a moving picture. Fortunately, the starting point of Lazy God's artificial intelligence is facial and action recognition.
Of course, the self-learning characteristics of lazy artificial intelligence also determine that while serving so many objects, it is also constantly learning and absorbing data, and constantly improving its data core. Again, this is a huge resource overhead.
In addition to these versions, Nanming also wants to develop a special version of Qingyang University for Qingyang University, which can be regarded as a benefit for his classmates, similar to the lowest-end membership version, which will require more resources at that time.
But now let alone opening more membership versions, it's almost stuck.
The most important point for the relocation of the lazy god department is to find a place to build a new supercomputer in Dongbin City.
Now that the embargo has been imposed, I am afraid that this progress will be greatly delayed.
What to do?
Nanming is also a little entangled.
Stretched computing resources, embargoed chips, and the pilot tasks that must be completed for relocation......
Originally, I was going to plan the skill tree for the next point, but now I don't have a choice, there is only one possibility, and that is supercomputing!
Aren't you embargoing chips? We'll do it ourselves!
"You gather the people who are still in Qingyang, let's discuss. "Nan Ming Dao.
About five minutes later, Su Wenjie, Su Hongyi, Han Dong, Wang Ziyang, Li Yuncong, and Qin Ye came to Nanming's office and held a closed-door meeting.
As soon as he entered the door, Li Yuncong couldn't help but shout: "Aren't these Yankees just chips? What's the big deal? We designed them ourselves." ”
Su Hongyi sighed, in terms of the chip industry, the Yankees are really amazing.
In this world, there are two kinds of chip companies, one is called Intel and the other is called other.
As the world's most powerful general-purpose chip giant, Intel is two years ahead of the world!
This is still the technology that has been taken out, Intel also has a saying, called the R & D generation, the reserve generation, the release generation, although the current technology improvement is slowing down, but it can still crush for life.
AMD, which was once able to compete with intel, now even sells its own factory.
Everyone knows that i3 is intel's low-end model on the PC platform, but on the PC platform, there is a saying called i3 silent seconds, not to mention i5, i7, and even the embargoed Xeon.
But if it's out of reach, it's not true.
Foundries are all over the world, there are foundries that can manufacture 28nm process in China, the world's largest chip foundry TSMC is also in Wanwan, 16nm technology has long been mature, Samsung 14nm is not too mature but can also be used, take out the design, choose the process, and pay for it to customize all kinds of CPUs.
Now Intel's products are only 14nm, which doesn't seem far away.
It is not manufacturing that blocks domestic CPUs, but the other three major obstacles: 1. design, 2. patent barriers, and 3. general procedures.
The chips of silicon-based semiconductors are completely developed on the basis of other people's systems, and the design concepts and patents are too different, and the versatility is ...... What's the use of not having a system adapted to your chip, and your functions can't be played?
However, this limitation is mostly limited to desktop platforms and supercomputing platforms, and the domestic situation is much better, including chips and architectures, which are all domestic, and there are also supercomputers, which are quite strong.
But the software is far from adapting, and compared with the hardware, the gap in software is desperate.
"Making chips is not a problem for us at Lazy Technology. Nan Ming glanced at everyone and showed strong self-confidence.
I have black technology.
I have black technology.
I don't even need hack technology. The strongest thing about Lazy God Technology is its powerful software capabilities.
"The question is, what kind of chips are we going to make. "Everyone knows that Moore's Law is about to reach its limit. ”
Moore's Law says that when the price is constant, the number of components that can fit on an integrated circuit doubles about every 18-24 months, and the performance doubles.
This process is not sustainable wirelessly, and when the spacing of the transistors is only 7nm, it is even necessary to consider the impact of quantum effects on the chip. (To be continued.) )