Current location - Quotes Website - Team slogan - Hardware terminology explanation
Hardware terminology explanation
2 (Computer Hardware)

When it comes to PS3, people can't help admiring its amazing computing power, and its "core" is the famous CELL.

CELL is a processor jointly developed by IBM, Toshiba and Sony. This processor is designed with multi-functional parallel architecture, with multi-task, multi-effect and huge floating-point operation function, and the highest frequency is 4.6 Ghz. "Cell" processors are versatile and can be used at various levels. SCEI will take the "CELL" processor as the core component of the next generation PS3 host, while IBM will build a high-performance computer workstation with the "CELL" processor, and Toshiba will use the "CELL" processor in high definition multimedia (digital TV).

About the summary function of cells:

With the multi-core and multi-thread architecture, it can run multiple different operating systems at the same time.

It has a highly programmable architecture and a highly flexible I/O interface.

The memory and transmission bandwidth of I/O part are greatly expanded.

Real-time resource management system provides a powerful hardware information security protection system for real-time processing requirements.

Using 90 nm technology and IBM's SOI (Silicon on Insulator) process.

How fast the "cell" processor is, it may be known by speed. According to IBM, a single "cellular" processor core can achieve a floating-point computing power of 250 Gflops per second, which is more than 40 times that of the current PS2 host CPU Emotion Engine(6.2G flops), and even the computing speed of a single "cellular" processor is comparable to that of the most powerful supercomputer in the mid-1960s+0990s, or six times that of Nvidia's most powerful 3D display chip.

It can be found in PS3 and IBM's latest blade server BladeCenter H. I believe the application of CELL will be more and more extensive!

give exact information

It contains 9 cores: 1 64-bit PPE control core and 8 identical SPE operation cores.

It has 250 million transistors (DD2 production version). The core area is 235 square millimeters, and it is manufactured by IBM's SOI90 nanometer copper interconnection process, and then by 65 nanometer process.

The working frequency of the first batch of engineering samples is 4.06GHz, and the working voltage is 1. 1 volt. It has been 4.6GHz since then.

It can perform 250 billion floating-point operations (256G Flops) per second.

Built-in 2.5 megabytes of memory, through Rambus's XDR and FlexIO technology, it can exchange 100 gigabytes of data with external memory every second.

It supports grid computing and has a flexible parallel and distributed computing structure.

sli

The full name of SLI is Scalable Link Interface, and two PCIE graphics cards of the same model are used on a motherboard supporting dual PCI Express X 16 through a special interface connection mode. In order to enhance nVIDIA's competitiveness in workstation products, after all, ATi has been nibbling at nVIDIA's market in this field with FireGL series. In the future product line, SLI will become the new highest point. So, what kind of technology is SLI? How is it different from the past multi-GPU technology? Can SLI achieve good results in the market? To answer these questions, it is necessary for us to make a comprehensive analysis and introduction of SLI, and also review the development history of multi-GPU technology.

The history of multi-graphics parallel mechanism can be traced back to 1997. At that time, the graphics card market was dominated by 3Dfx, and the voodoo accelerator card launched by the company in the second half of 1996 became a classic product sought after by enthusiasts. At the beginning of 1998, 3Dfx launched ——Voodoo 2, a second generation 3D graphics card product. At that time, Voodoo 2' s pixel filling rate was 90Mps, and it had the most advanced 3D features such as Z-buffer, anti-aliasing and single-cycle double texture, which greatly surpassed the previous generation and was far behind other competitors. However, the most crazy thing for enthusiasts is Voodoo 2' s "SLI Interleaved Interconnection Technology", which allows two Voodoo 2 graphics cards to run in parallel, and the 3D performance is nearly doubled. As a result, other competitors are far behind.

We know that the parallel operation of CPU is obtained through the parallel execution of instructions, but the situation is different for graphics cards. The graphics card finally generates a rendered 3D picture, which contains a lot of instructions, so how to distribute the work evenly becomes a problem. 3Dfx selects the rendering method according to the frame line of the picture. SLI technology divides the rendered image into scan lines. If Voodoo 2 adopts dual graphics card operation mode, then one graphics card is responsible for rendering the odd frame lines of the picture, and the other graphics card renders the even frame lines. Then the simultaneously rendered frame lines are merged and written into the frame buffer, and then the display can display the complete rendered picture. It is not difficult to see that SLI technology makes the rendering work equally shared, and each graphics card only needs to complete 1/2 workload. Theoretically, the rendering efficiency can also be improved by 1 times, which is the secret of improving the parallel performance of dual graphics cards. SLI is very successful in technology, and enthusiasts have great enthusiasm for Voodoo 2. At that time, if you wanted to play 3D games smoothly at the "high resolution" of 1024×768, the only solution was to use two Voodoo 2 graphics cards and make them work in SLI mode.

In Voodoo 3 after Voodoo 2, 3Dfx did not follow SLI's dual graphics technology, but in Voodoo 4/5/6 era, 3Dfx restored SLI, but the application form was different. Voodoo 2 advocates the parallel operation of two graphics cards, which are inserted into PCI slots respectively and then connected by special cables, but this is not necessary. A single Voodoo 2 graphics card can also work independently, but at a slower speed. In the spring of 2000, 3Dfx introduced VSA 100 graphics chip. At that time, NVIDIA had surpassed 3Dfx to become the boss. In order to regain its leading position, 3Dfx reloaded into SLI technology. VSA 100 can support single-chip, double-chip and four-chip parallel operation. The single-chip version is Voodoo 4, the dual-chip graphics card is Voodoo 5 5500, and the quad-chip graphics card is the famous Voodoo 5 6000. At this time, SLI technology has evolved into the form of single graphics card and multiple graphics chips, which does not occupy two slots, but the internal working mechanism has not changed much. It is still executed by dividing rendering frames and then synthesized in the frame buffer. Due to well-known reasons, these graphics cards have not been widely recognized, and 3Dfx has also gone from decline to death. At the beginning of 200 1, NVIDIA acquired 3Dfx, and SLI technology became history. Although nVIDIA has mastered all the technologies of 3Dfx, it has not carried it forward, but has continued to go its own way. The purpose of acquiring 3Dfx may be just to destroy a competitor.

Then we saw nVIDIA successfully unify the rivers and lakes, and then ATi gradually launched a challenge. GeForce and radium dragon are the most frequently mentioned terms. As for 3Dfx and its SLI, people have gradually forgotten it. Even if people talk about it occasionally, most of them will say that it is an enterprise with improper strategic planning and an expensive and unrealistic technology. In the history of graphics cards, except Voodoo 2, no multi-graphics card and multi-chip technology has been successful. Although ATi has tried and the new XGI has ventured into this field, it turns out that this scheme is not popular with users. However, no one expected that NVIDIA would pick up SLI technology of 3Dfx again. On June 29th, 2004, nVIDIA released "SLI Multi-GPU Technology" with great fanfare, and introduced this technology into the newly released GeForce 6800 and Quadro FX4000 series graphics cards. The name "SLI" is more or less reminiscent of 3Dfx. Perhaps this is what nVIDIA wants, and it hopes to be considered by users to be the same strain of 3Dfx technology. However, if we analyze it deeply, we will find that it has little in common with SLI technology of 3Dfx, and it is basically a new multi-graphics card scheme developed by nVIDIA.

professional

professional

Abbreviations.

1. Public Relations Office

pro 1

KK: []

DJ: []

Mouth = major

pro2

KK: []

DJ: []

Advertising.

1. in favor of, definitely

Noun (short for noun) [C]

1. Those in favor

2. Arguments in favor

The advantages and disadvantages of strike action were discussed.

People discussed the reasons for and against the strike.

3. Yes

a.

1. Positive.

preposition

1. Yes