Blooming America

The father of the super computer is the side of the laboratory. When you find yourself, you can't sleep, the three points in the middle of the night, Xido. Krevius started with Chuáng, then looked at your design.

"Why is it still not?"

A new generation of supercomputers will use Capricorn as the core.

When Xaha. Kren took this 32-bit micro processor, it was lived by this magical processor, using the latest process, using the best architecture, and has ten times the series. can……

And these, let Xido. Kref insee insee insertion when designing a new generation of supercomputers.

At this time, Kray God suddenly thought of the architecture of the new supercomputer who had discussed with Catherine, and this architecture and the new processor's composition seem to be a good decision!

Thinking of this, Kle immediately organized a human hand and started the development of the supercomputer.

However, in the process of research, Kre is an unprecedented hindrance.

The new super computer uses a turbine design that each blade is a server that can achieve collaborative work and can increase or decrease at any time according to the application.

The form of such a blade server is that Krem is ready to use.

The blade server refers to a plurality of card-type server units in a standard height of the standard height, which is a low-cost server platform for HAHD. This is a high-density computing environment specially designed server, efficiency It turned out to be higher.

The blade server is like "blade", and each "blade" is actually a system motherboard.

These blade servers can launch their operating systems through a "onboard" hard drive, such as X system, UD system, etc., similar to a server server, in this mode, each motherboard runs its own system, service There is no association between the specified different user groups.

However, administrators can use the system software to set these motherboards into one server cluster. In cluster mode, all motherboards can be connected to provide high-speed network environments, and share resources at the same time, serving the same user group. Insert a new "blade" in the cluster, you can improve the overall xìng. Since each "blade" is hot-swappable, the system can be easily replaced and the maintenance time is reduced.

In the last generation of the Galaxy Supercomputer, there is such a concept, you can add a module to your computer.

However, such technology and the current new technologies that are well-known by Krere, it is insufficient.

New technologies can be more complete and skilled using supercomputer's computing power, and increase the efficiency of the machine.

These blade servers have the characteristics of low power, small space, low price, low-cost, etc., while it also inherits some technical indicators of the Galaxy Server, such as the use of hot swaps and redundancy to blade servers. In the design, these design meets the needs of dense computing environments to server xìng; some also effectively improve the server's stability xìng and core network xìng energy by built-in load balancing techniques. Moreover, the blade server can maximize the use of space and fees for the server and provide users with flexible and convenient extension upgrade.

Although this, but design, Krene encountered various problems.

First, this is a new architecture.

Second, the 32-bit processor is also a new architecture.

And at this time, Kle's previous experience, but it is not used. Because this time, he touched is a complete new technology, such a technical order, alone, or even a group, is very difficult.

Kreve realized that he needs a big team, then after a long study, it is likely to achieve its own goal.

One year ... No, maybe two years, three years, your own goals can be achieved.

All previous things must be re-designed, so there is no such time, it is impossible to do it.

In the 21st century, the blade server has become the mainstream of the high xìng to calculate the cluster. In the global super 500 supercomputer, many new cluster systems use a blade architecture. Since the blade server can greatly reduce the number of external cables required, it can greatly reduce the hidden dangers due to cable connection failures, and improve the system reliable.

Catherine has been discussing with Kreth, the two sides talk about a supercomputer composed of such a blade server.

In terms of server, Catherine's classification and past life is almost the same, which is completely Catherrat cat painting.

According to the shape of the shape, the server is divided into three types of tower, rack, and blade servers.

- But actually there are only two kinds, because the third kind of time in Catherine, there is no appearance, which exists in Catherine's notebook and Krene's brain.

The tower server is generally the most seeable, its shape and structure are similar to ordinary opportunities, just slightly more, and its shape has no uniform standard.

The motherboard extension of the tower server is stronger, and there are many slots, and the inside of the chassis of the tower server tends to reserve a lot of space in order to perform redundant expansion of the hard disk, power supply, etc. This server does not require additional devices, and there is not much requirements for placement space, and has good scalable xìng, and the configuration can be high, so the application range is very wide, which can meet the general common server application requirements.

This type of server is particularly suitable for common entry-level and working group-level server applications, and the cost is relatively low. Xìng can meet most of the small and medium-sized enterprise users. The current market demand is still very large.

For example, Catherine's current personal website server is basically using a tower server.

And some small companies, they also buy this server, their facts are not big for such computers.

The higher the needs of the computer is not, the better, the cost is also a very important reference object.

However, this type of server has many limitations, and when you need to use multiple servers to meet the needs of higher server applications, it is very uncomfortable due to the larger server application requirements.

The rack server is actually the product under industrial standards, and its appearance is designed in accordance with unified standards, with unified use of cabinets to meet the company's server-intensive deployment needs. The main role of the rack server is to save space, but can not only take a smaller space but also to manage it.

The width of the rack server is 19 inches and the height is in units of u-in-units. It usually has 1U, 2U, 3U, 4U, 5U, 7U several standard servers.

If it is the 21st century, the popular probably 1U and 2U chassis are just in Catherine's era and different. Because the motherboard is "silly black and crude", the 3u and 4u chassis are almost people's choices, and in many people's eyes, this is already "small".

The advantage of such a server is small, and it is easy to manage, but due to internal space limitations, the expansion xìng is limited.

This server is used in large-scale enterprises that are used in the number of servers. There are also many companies that use this type of server, but the server is delivered to a dedicated server managed mechanism to host, especially the server currently uses this way. .

If you want a better company, that is, the supercomputer is used. The first generation of Cerai now has no production, and the second generation of the Galaxy is the mainstream in the market. Orders have even already 1975.

Even ten currencies, this time there is such a supercomputer, but such a supercomputer is not a regular "server", just because such a supercomputer is strong enough, it can be competed.

After all, at the beginning, Catherine needs to be stable, the cloud system is the typical, such a machine, of course, can be used when the server is used.

But to say, this "orthodox" choice is in Xido. Kray is now studying the system.

"How can I play the efficiency of the entire system?"

According to Catherine's statement, such system efficiency is very high, but Kre is in the study, but it is not the case.

Because the blade server and "Galaxy Module" - are also a separate module of the Galaxy computer, the volume of the blade server is more intensive.

The core of the blade server is a better process better, but because of the relationship between space, it is higher than the "Galaxy module" due to the relationship between space.

Not only this problem, when the first design, Krene found that the architecture they designed, only with the 50% of their expected 50%, more xìng can be wasted.

"Come back ..."

Krere can only think so.

"Maybe ... I should change my thinking ..."

The new core is self-contained, but its own thinking is still on the old computer, why can't you think about thinking?

If you can also play the gpu's xìng, you should be more good ...

......

By the way, the new computer is configured by the server chassis, 4U standard, with GTX670 graphics card, old roll 5 full speed 60 frames ...

Today, three, first more ~

It seems to be eating bad, your stomach is nap ...! .