Two days later, the server purchased by Duke commissioned Kusai to arrive as scheduled. Duke received the arrival notice from Kusai and drove to the company.

"Boss, look, this is the computer you want, to see if it's what you want to buy." Kusai was waiting for the company and saw Duke coming, he quickly took Duke to the side of the server.[]

Duke opened this computer, went to Bios and checked the configuration, it was indeed the 20 core te of two amds.

ama

processor.

"That's it, thank you, Kusai, go and do your business." Seeing that it was indeed something he wanted, Duke sent Kusai away.

After entering the system, Duke took out a USB wireless network card he carried with him and connected it to the server. Now Kerry can only use the Piaotianyi network network, Duke carried a Piaotianyi network card with him.

After quickly installing the driver of the wireless network card, Duke started Kerry and connected it. Soon, Kerry combined the amd te obtained in advance.

ama

The processor instruction set, after countless calculation tests, finally, after about two hours, "the hardware of the new CPU system was completed virtually." Duke's mind suddenly vibrated, and Kerry gave information about the virtual success.

Kerry Virtual Successful Software Te

ama

Processor, fully equipped with te

ama

The original performance of the processor and the entire instruction set are of course different internal implementations. Then Kerry uses this as the core to virtualize a new dual-channel CPU independent system, which can directly run the www.piaotian.comux system on this set of virtual hardware.

Seeing Kerry complete the cloning of a dual-channel CPU server hardware system, Duke downloaded a set of the latest version of the CPU test application and started the test on the virtual system inside Kerry and the actual server around him at the same time. The results showed that Kerry simulated te

ama

System and actual hardware

ama

The system showed exactly the same test results, proving that Kerry's cloning was very successful.

Immediately afterwards, Duke instructed Kerry to complete the simulation of 4, 8, 32, 64 and 128 servers. After this actual simulation test, Kerry fully mastered the hardware simulation of multiple servers. On this basis, Duke began to instruct Kerry to follow C.

Ayxt5 design architecture completes simulation work on the Titan ii supercomputer.

From the perspective of supercomputer design principles, there is nothing mysterious about it. The Titan ii is just a number of 4 20 core tes equipped with 4

ama

The processor and 32gb memory computing nodes are composed of three parts: computing array, acceleration array and service array.

Since all this is just a series of computer nodes, Kerry soon completed the simulation of this supercomputer system, but Duke found that the Titan ii simulated by Kerry seemed to have a serious performance lower than the real Titan ii.

According to the Titan II performance introduced by Hewitt, its stable computing speed reaches 15 petaflops (i.e. 150 million times per second). Now, the Titan II software simulated by Kerry is only about one-thirty of its performance according to a simple multi-channel server.

What's going on?

Duke looked back and carefully studied the composition of Titan Ii. There were a lot of materials on this network. After looking at the computing units used by Titan Ii, Duke realized that he was still thinking a little worse, because in the composition of Titan Ii, the computing array and the service array are composed of computing node machines using general-purpose processors (cpus) and service node mechanisms, while the acceleration array is composed of a large number of acceleration node mechanisms based on the graphical acceleration processor (gpu). It is a heterogeneous collaborative computing of "cpu + gpu".

Because from the perspective of floating-point computing power alone, the floating-point computing power of a GPU is dozens or even hundreds of times that of CPU, and the video memory bandwidth of GPU can reach more than ten times that of CPU, and the latency is lower, and the external data throughput capability is stronger than that of CPU.

Therefore, when designing, if a supercomputer is composed of a GPU of the same size, the floating-point computing power of this node will be increased by a hundred times. The advantage of CPU is that it has strong computing logic and relatively complex data structures.

Considering that whether it is nuclear reaction simulation or climate estimation simulation, a large number of floating-point operations are involved, this "cpu + gpu" heterogeneous collaborative architecture is also adopted in the Titan II system, and teslak20gpu produced by Nvidia is used as the main floating-point operation component.

Fortunately, it is not only supercomputers that use this heterogeneous model, but also many enterprise-level servers to choose from. Duke ordered Kusai to buy one in Boston that day.

In order to ensure that the simulated Titan Ii is consistent with the real Titan Ii in the laboratory, Duke went to the laboratory this time to get the detailed CPU and GPU composition structure information of Titan Ii. This parameter is no secret. So when Kerry completed the software simulation of teslak20gpu, he realized the software simulation according to the Titan Ii architecture again.

16.5petaflops! When the software simulated Titan ii test this result, Duke was excited. Kerry finally realized the reproduction of Titan ii. The system recompiled according to the source code version of the Titan ii system was successfully run on the Titan ii system simulated by Kerry. It can be said that Kerry has now fully implemented the running environment of the nuclear reaction simulation program.

Everything is ready, but the east wind is gone, and the only thing left is how to download this application to Kerry space.

"Duke, you're here." Hewitt was not surprised when he saw Duke who hadn't seen him for a few days. He said hello normally. He knew that his junior brother was a real genius. He was now studying three disciplines with far-reaching differences in materials, computers and nuclear physics. He didn't know how that head grew.

"Hello, senior brother, today I'll come and see if the permission application has been approved." Duke returned politely.

"Oh, there should be no problem. Professor Andrew has been there all this time." Hewitt said. "How is it, I'm thinking about starting research? Have you read the guide to use this system now?"

"I've looked at it briefly. Now there is a small experimental design. If the permissions are not problematic, I want to see how the on-board effect is." Duke said.

"Is that so? Duke, you have really designed an experiment?" Hewitt was a little surprised. Duke's learning speed was really terrifying, because it took a semester to make the first experimental design that could be used on the computer. The guide for big things was too thick, and it took a lot of time to read it all.

"I don't know if it's right. I just followed the guide and wanted to study the operating mechanism of a natural nuclear reactor, so I designed such an experiment." Duke said.

"Natural nuclear reactor research?" Hewitt was stunned and looked at Duke thoughtfully.

Of course, he knew that there was a place on the earth that had naturally formed a natural nuclear reactor. Perhaps it was discovered that such a natural nuclear reactor had happened in a uranium mine in Kafeng, Africa. However, how it was formed, a very small number of scientists have been studying it over the years, but due to limited conditions, there were not many results.

Unexpectedly, Duke chose this relatively unpopular topic as soon as he came. Of course, research on this topic is very meaningful, but there are too many problems in it. The results of the research are only inferred from the general occurrence mechanism.

Since these natural nuclear reactors have completed their historical missions, there is no real natural nuclear reactor found on the earth, so scientists can only conduct research through a very small number of nuclear reactor derivatives. There are too few clues to find. It is really not easy to achieve success on this topic.

"Duke, the direction of the topic you chose is really challenging." Hewitt came back to his senses and continued. After that, he shook his head. This junior brother was really different. Hewitt thought to himself that he gave a negative evaluation for Duke's academic prospects in nuclear physics for the first time.

Since humans discovered the existence of natural nuclear reactors more than 30 years ago, over the years, there have been only a few achievements that can be released. Hewitt does not believe that Duke can be a special exception.

Duke did not understand Hewitt's psychology. Among the few nuclear fission knowledge he received from Bileme, how nuclear fission occurs in natural conditions is the few remaining popular science knowledge. There are all kinds of parameters of this environment, so for Duke, this is really just a very ordinary thing.

It is just that how to show this discovery process as natural as possible is a real challenge. Duke must study some things thoroughly in theory and then take out experiments to verify them.

If the entire natural nuclear reactor formation process is rashly displayed, it will scare many experts and scholars who study nuclear physics.

With Duke's current emotional intelligence, he will definitely not do such stupid things anymore.
Chapter completed!