Chapter 1718

In the summer of mori, you can't miss Shu No, it is in the discussion with his assistants and partners that E3 exhibition is gradually fermenting.

Unlike the Tokyo video game show for players, E3 is a trade fair for developers, distributors, platform vendors, etc. - but in the future, the nature of E3 has gradually changed and become a game exhibition that is concerned by the public.

The same is true of this world line.

But that was before Sonia set foot.

After senxia began to set foot, E3 was somewhat damaged by senxia.

Every day before the opening of E3, senxia held a grand conference here. This game Carnival directly turned the taste of trade fair into a carnival.

But what does it matter?

Pleasure is the best.

However, some people are so happy at the scene, and some people are not so happy on the other side of the order of Phoenix.

At the same time, senxia has just finished the streaming of gsii for the next generation.

The gsii will use a four core CPU.

It's still glue design, but it's a little bit better than the current design, because each "glue chip" has two cores.

In the design, according to Hitachi's suggestion, the internal part of the chip is split, the IO control chip is independent, and two dual core die (chip core) and a GPU core are placed in it, which turns into a quarter chip.

"Let's see if we can succeed..." Fujihara took a breath.

Fujihara is an old employee who joined the order of Phoenix with Sega's development team.

He once participated in the cooperation plan with IBM, and now he is in contact with Hitachi and other clubs to develop a new CPU.

This is definitely not a simple job.

In fact, at the beginning, senxia intended to integrate all the cores into one die, because in this way, the internal architecture of the chip is the best, the design is simple and the delay is low, which can be said to be a very convenient design.

Therefore, after Hitachi and other companies have come up with this kind of quarternary structure, moriha is not recognized.

In Mori's opinion, except that GPU is developed by other companies and needs additional "glue", it is better to integrate all the others.

This kind of design of four pieces of glue with egg ache can't get into Sen Xia's eye.

But at that time, it was toshara who came forward and convinced mori.

The reason for fujihara's use is simple - yield.

This kind of "glue" dividing CPU is "glue", but the yield of products is very high, and the improvement of product yield represents the decrease of product cost and the improvement of production efficiency.

In the production of CPU, the factory first produced a large round wafer, and then let people cut it and engrave the circuit on it.

In this way, the CPU becomes.

However, when cutting wafers, there is a problem, that is, wafers are round, while chips are rectangular, which means that the corners of the wafer will be wasted, and the smaller the core, the less waste.

It's not just that.

On the wafer, not all places are perfect. Sometimes the core is produced, and one of them may be damaged.

The bigger the chip, the more so.

For example, according to senxia's idea, if a chip with high integration is produced, the cost may be $100. However, if there is a problem on the chip, the chip will be discarded, and the loss will be $100, and the whole chip will be scrapped.

But if it is cut into four small chips, one of the damaged chips will be damaged. In this way, the material cost of $100 may be only $25, and the remaining three chips can still work.

Mori was convinced by this example of fujihara.

Glue is the glue. Just cut down the cost.

So, fujihara gave himself to the pit.

He was promoted.

Promotion is a good thing. But after fujihara's appreciation, the problems to be solved will be great.

"Glue chip" yield and cost can be reduced, this is in the material.

But the design of this kind of chip is not so easy.

If it is a fully integrated chip, the communication within the chip is relatively smooth, which means that the product can save more resources.

The performance is that the chip has higher efficiency, lower latency and easier to resolve internal conflicts.

But if it's glue, the coordination between each die is sometimes a big problem. If one of them is not done well, the delay between chips will increase dramatically, and the work coordination between die and die will also have problems.

Its performance is that it is easy to see a situation in which one core is working while the other several are watching.It's also the lightest side effect. If the design is not good, there may be other problems, such as repeated commands sent to the core, blue screen caused by conflicts, and error reports.

In other words, this glue design not only improves the hardware yield, but also slightly reduces the efficiency of the product. In addition, the requirements for architects are raised by several grades. In fact, the design cost is increased.

And in charge of this part, it is taro fujihara.

Gsii will be listed next year, and whether gsii can have a stronger "heart" depends on the results of this time.

If the trial production of the chip can be successful, then gsii will be found, and the development machine will come out and send it to other manufacturers within the year.

But if not Congratulations, the product needs to be redesigned and filmed. Millions, even tens of millions of dollars, will be wasted.

Because of this reason, so at this moment fujihara will be under great pressure.

Promotion is indeed a promotion, but if this one is not done well, it will be very dangerous.

Fortunately, fujihara knows that he is not the most nervous one.

In fujihara's view, the most square Actually, it's the handheld department.

The handset department is now waiting for the maturity of the new technology. The concept prototypes produced before this are always just prototypes.

What's more, the handset department can't copy the previous success mechanically.

You should know that if the CPU switches the process, even if it is just process adjustment, the design of CPU itself can not be mechanically adopted directly, but needs to be adjusted on this basis.

Only in this way can we make effective use of it.

But because of this, the design is very exquisite.

But that doesn't mean Mr Fujiwara is relaxed.

Because glue is not easy to mix.

Of course, the so-called "glue multi-core" processor does not really fill the CPU with glue, but a kind of package in which two die are placed in one CPU. As mentioned before, it is not listed here for the time being.

Intel was the first to take the lead in commercial glue, as was the Pentium Pro Series released in 1995. Pentium Pro is the first processor in the world to support more than 4GB of "super large" memory, and the first CPU to put the concept of "glue technology" into practice.

However, for ordinary users, Pentium Pro is not necessary at all, and commercial users who need it directly buy the motherboard with multiple CPUs (that is, the motherboard supporting multiple CPUs).

As a result, this ancient glue technology was temporarily submerged in the sea of people.

Until another world line, home's amdx2cpu adopted the "true dual core" scheme.

But that's another dimension.

After I's family didn't touch glue very much, the IBM family has been developing it silently. Otherwise, senxia couldn't ask the other party about this technology.

"It's said that the next generation of PS game console has been developed." At the time of fujihara's rest, colleagues came over with coffee.

"Got the news?" Fujihara was a little surprised.

"Well, after the news of our cooperation with IBM came last year, Sony also found IBM to develop a" big guy ". Mr. shawson speculated that it was the core of preparing for the new generation of PS

In order of the Phoenix, there are many people who are Sen Xia fans.

There are fans of books, games and animation.

These people will also call Sen Xia "teacher Xiasen" behind their backs.

Well, in fact, many of them learned this habit from group A, because many people in group a secretly called senxia.

"Sony's information is so poorly kept secret?" Taro fujihara was a little surprised.

Sony's security capability is not bad.

But What a special channel!

"I expect the document will come down from the top in a minute, and then I can see it."

My colleagues are right.

In the afternoon, fujihara got a confidentiality agreement.

After signing the agreement, he saw some design information about the game console.

But fujihara knows at a glance that this product is not the next generation game machine developed by himself.

Because this machine is not right at all.

The game console named "P" temporarily adopts the setting of embedded FreeBSD Operating System, which is not right - because I use vista system.

The CPU of the game machine is the processor code named "C".

The idea of this processor is not in line with my own.

Senxia's route is similar to the current PC, which is the CPU + GPU route, but it is the integration route.The C processor is a single core eight coprocessor.

Yes, that's the PS3 specification.

It's just another world line PS3.

In this world line, a lot of things have changed.

For example, Sony found IBM to cooperate. In fact, there are some reasons that Sony was stimulated by senxia. Part of the reason for their cooperation is the dual core of senxia GS.

As for whether there will be a PS3 of the same specification in this world line, Sen Xia is not sure, because many of these things have changed - maybe someone has come up with another PS3?

However, although there is no reference value, there is still some reference value, so senxia put out this product that Sony may not have at home for a moment to give you some reference.

C processor is the famous cell.

But Mr fujihara did not know.

However, taro fujihara was a little surprised, because the architecture of this product is very wonderful. It is a single core, but it has 8 spe (equivalent to coprocessor core) CPU.

The main memory of the machine is 256 MB.

"So small?"

Fujihara murmured.

The product line that Sen Xia gives here, next generation gsii, memory is 256 MB.

However, if the PS3 is only 256MB, this product is not an iterative upgrade product. When it is launched in the future, it seems to be a little small.

- you know, when gsiii is on your product roadmap, the memory capacity of the whole product will reach 1GB!

According to fujihara's judgment, gsiii should be right on the PS3

"This should be the original design and will definitely be adjusted in the future." He muttered.

Although some colleagues have disclosed that this document is correct, it may not be the final one even if it is correct. Sometimes, the things to be done during the project establishment and the final product may be quite different

"New types of optical storage media with 50GB capacity This is too scary... "

Does this thing really exist?

Taro fujihara was a little surprised.

The next generation of gsii will adopt high-capacity DVD, which can support 9GB high-density disks, but it is still far from ideal.

But think about it It seems possible.

My handset department is cooperating with Sony. Maybe there is news about the other party's new type of CD?

This is easy to understand, because the current DVD capacity, sometimes it is not enough to watch.

"Up to 1080p output is supported..."

This is easy to understand, because the same requirements are almost the same at home. Senxia requires its products to maintain a good "full HD" support environment.

"Sony, that's what's going on."

Well, Toshihiro Fujiwara has already agreed that this is Sony's product model.

"It's totally different from our development ideas." Tanhara murmured.

Of course, the machine itself is not said to be "Sony". The only one marked in it is that the GPU of this machine is designed by NVIDIA and is still under development.

"On sale in 2006..."

Four years later.

Four years is enough to change a lot of things.

And then He went on to look at the comments above.

Yes, there's also senxia's comment on it.

"The cost is too high, and subsequent research and development of products may lead to losses. We'd better pay attention to the cost of our products. "

This is Sonia's first.

"The memory is too small for the future to foresee, failure."

This fujihara agrees.

"The structure of CPU is too peculiar, which will lead to development difficulties. We must not take such a wonderful path. "

Mori said this, fujihara does not agree. Because this kind of design is really interesting - after all, fujihara is also neon gold, and neon gold likes to develop wonderful architectures

But since the boss said so, it's impossible for toshara to change it, isn't it?

After all, I was promoted before

Well, although the promotion is very busy, he still wants to say How delicious!

……

Two in one meow.

We can't find out the specific development time of PS3. Let's take the time of cell development as a benchmark.

Open in 2005. In 2002, we should have started to develop

www.novelhold.com , the fastest update of the webnovel!