Chapter 980： Private super calculation (below)

The test after the relocation of the supercomputing is completely under the guise of Guo Tailai's intention to test the supercomputing chip he made.

Now the supercomputer deployed by Guo Tailai's data service company is the dawn of 4000a developed by the Institute of Computing Technology of the Chinese Academy of Sciences, with a running speed of 11 trillion times per second. In the early stage, except for occasionally rendering the mt30 assembly demonstration animation during idle time, most of the other computing power is performing the force analysis of the supersonic airliner design scheme required by Guo Tailai, wind tunnel analysis, etc. Full.

This time, through the opportunity of deploying an underwater machine room, Sugon-4000a will also be partially upgraded. Although it has not reached the point of 5000a, it is also a smaller version than 4000a, and the running speed has reached 80 trillion times per second, which is almost an increase. More than seven times.

Of course, Guo Tailai will definitely carry private goods, and put a batch of his own servers into it at the same time. Once the underwater machine room is under water, it will only be opened and maintained for a long time, and Guo Tailai's server is relatively safe. When using it, only a separate line is needed. This is easy to do. Even Guo Tailai can leave a part of the nano-robot in the machine room for remote control via satellite line, which is very convenient.

The Dawning Supercomputer must be delivered after the test is completed, basically don't worry, but Guo Tailai's own supercomputer must definitely be tested, just used to do a rendering for "Avatar" to test the usage.

Guo Tailai said so, and it is still a super-computing rendering. Why would James disagree? Avatar's post-rendering uses Vita Digital's six supercomputers, each running at about 200 trillion times per second, and the rendering work lasted a full year. Adding one more super-calculation can be completed in advance, and it is not harmful to him. And the boss's own super calculation, confidentiality and priority must be taken care of, don't worry at all.

"Okay!" James agreed, without hesitation.

There is no need to worry about negotiating with Vita Digital. Anyway, Guo Tai said that it is only a small part of the small-scale rendering used to test the operation of the supercomputing, and it will not be divided.

Guo Tailai needs super calculation. Not only does the design company need to use supercomputing for wind tunnel analysis and airframe stress analysis, but he also needs a supercomputer to design the aircraft he wants. The monster plane is not suitable for everyone to know in the early stage, let's make it by yourself!

Whether this is Guo Tailai Chu Fei, or James, the body and mind are relaxed, far more satisfying than the ordinary holiday. The three people soon became energetic and engaged in their work.

James continued to do Avatar later, Chu Fei also began to prepare Transformers global propaganda, but this time the publicity work is mainly concentrated in the United States and China, Guo Tailai can also accompany Chu Fei to attend the premiere.

Guo Tailai's work is naturally the same as before, but there is another new job, processing the underwater machine room and installing and debugging his own server.

The twelve-inch silicon wafer columns previously purchased can now be fully used. The chip he is going to make has already been designed and processed.

The design of the super chip carried in the control system is a mature giant chip in the future, 1.2 trillion transistors, a twelve-inch wafer can only make one chip, the largest chip in history, specifically for ai task and super calculation.

This giant chip is the final mature version of wse (wafer 　 scale 　 engine) produced by Cerebras. It has an area of ​​42225 square millimeters, has 1.2 trillion transistors, 400,000 cores, on-chip memory 18g bytes, and memory bandwidth 19pbyte/s , Fabric bandwidth 100pbit/s.

Four hundred thousand cores are connected together through a fine-grained, full hardware, on-chip network, which can provide a total bandwidth of 100pb (100x1000tb) per second. Provide supercomputer-level computing power, high-performance memory and bandwidth in a single wafer-level solution. It can be said that a wse chip is a super calculation, and it is still a super calculation in the next fifteen years, and it also comes with ai learning ability.

Guo Tailai produced 10,000 of these new films, and integrated them according to the integrated system. It can be said that if successful, this will be the most powerful supercomputer in the next decade or so. The overall performance is millions of times stronger than the currently used Dawning 4000a.

Avatar’s post-rendering work used Vita Digital’s six supercomputing operations for a full year, and it may take less than an hour before this new supercomputing. Unfortunately, Guo Tailai's powerful computing power is to ensure the perfection of the monster aircraft he made, not to render special effects for film and television. Guo Tailai also has no intention of revealing this artifact to outsiders, or let him master it first.

The biggest problem with wse is that the tape-out success rate is too low. A twelve-inch wafer can only make one chip. There is a slight error, and this chip is scrapped.

But Guo Tai came here, the problem is not big, although there will also be occasional mistakes, but it is much higher than the power caused by the lithography mechanism. Subsequent packaging is also the same, which has led to almost no one can do this except Guo Tailai, and the chip industry leader Intel is not.

Another trouble is that the heat dissipation is relatively high, and liquid cooling must be used. Fortunately, Guo Tailai is currently directly under water in the computer room, the entire computer room will be immersed in water, a Dushu Lake lake water for cooling water circulation, to ensure that a large amount of operating heat can be taken away in time.

Guo Tailai's processing workshop is kept secret, and no one knows what he is doing inside. Especially after the two chiefs had eaten their meals, they each assigned personnel to strengthen the security level there. Apart from Guo Tailai and the few people around him, no one could enter the processing workshop. In this case, Guo Tailai processed a large number of chips without any pressure.

By the time it came back, it was mid-May. By mid-June, Guo Tailai had processed all the supercomputers and underwater equipment rooms. The underwater project next to the hotel has also been completed, just wait for the super budget to be moved and then enter the water.

It took a week for the data company's super calculation to be installed. After the power-on test was successful, the underwater machine room was sealed with water and successfully placed under water. The water-cooling cycle starts, and Dawning Supercomputer starts successfully and starts to run.

A small part of the rendering work of James was also submitted to Guo Tailai through the satellite network. Guo Tailai started the wse super calculation, installed the rendering software, and started the calculation. Ten seconds later, the two-minute special effects lens rendering was completed. Guo Tailai was in his mind. I saw the picture after successful rendering, perfect!

Good news came one after another, after more than two years of construction, the Iron Man Building was completely completed and delivered. Next, there must be work such as investment promotion, but some shots of the "Iron Man" crew can come here to make real shots, good news.

On June 22, US stocks pulled back high, the stock market plummeted, the Dow fell 2.37%, the S&P fell 2.29%, and the Nasdaq fell 2.07%. The treasure ship club members once again issued a small vote, but no one publicized , Make a fortune. But this time the decline is obviously a bit too big.