Fang Ziming was worried. Boss Qiu was the investor, so he had no choice. He originally wanted to invite Mr. Wang to come and reconcile, but Mr. Wang didn't give him face. What should he do? There is no solution.

But when Qiu Baijun reached out to the international version of the dubbing, Fang Ziming realized that it was time to meet another big boss...

...

The conscience of heaven and earth, Wang Anran is not afraid of face, he is really busy now.

The graphics card business unit, the CPU business unit, the chip production, the Finfet structure and mathematical model are all things he has to worry about. If he had not stepped down from the CEO of the three major companies, he would have been busier.

Even so, Wang Anran hasn't slept with peace for a long time.

Today, he came to the CPU business department. Thanks to Huaxin Technology's good planning, he has made a complete set of R&D plans in the current products and their optimization, future product development, future product demonstration, and a complete set of R&D plans.

Faced with Intel's aggressiveness, the CPU business unit did not show panic, and all matters developed in an orderly manner.

The only difference is that 800 nanometers, 500 nanometers, and 180 nanometers of 40 million-level transistors, the three manufacturing processes are moving forward together.

The 800-nanometer Athron first-generation pro model has begun mass production, and it is almost released. It is benchmarked against the third-generation Intel Pentium 586 processor. The results of the flow test are in line with the tone and are completely defeated. The performance parameters are the highest, which is nearly 27%.

The 500-nanometer manufacturing process is a benchmark Intel Pentium 2 processor. In Wang Anran's impression, this processor is about to be born.

As for the processor of the 180-nanometer manufacturing process, it uses the 180-nanometer manufacturing process that Texas Instruments is debugging.

Project cost calculation, after mass production, the cost of each qualified processor will be as high as US$3,000, equivalent to more than 20,000 yuan.

The market I face is naturally high-performance computing centers and rich players who have mines at home.

For Wang Anran, high performance is far from enough. On this processor, he also wants to break a world record, which is the 1ghz frequency mark.

There are two ways, one is to continue to explore the potential of the assembly line and deepen the existing assembly line from level 18 to level 32.

Another way is to use the thin instruction set as much as possible on the instruction set and under unnecessary conditions to reduce the delayed waiting caused by complex instruction sets.

At the product demonstration meeting, Wang Anran put forward several key points.

"The 180-nanometer manufacturing process is provided by Texas Instruments for us. This is an opportunity to demonstrate Huaxin Technology's super technical strength in the field of CPU design. It is an opportunity for us to officially stand at the same weight as Intel and surpass it..."

Wang Anran didn't know how to inspire people. He just expressed his true feelings and had been silent in the shadow of Intel for too long. Now the time, place and people are in harmony. We need to design and design, manufacture and manufacture, and have advantages that Intel does not have, that is, the 180-nanometer manufacturing process of Texas Instruments, which Intel cannot master.

The audience was filled with enthusiasm, and Wang Anran saw it and continued: "Today is a product demonstration meeting. I will give you a few indicators, let's discuss it together."

After Wang Anran finished speaking, the underground technicians opened the notebook to record.

Historical experience proves that Mr. Wang never talks nonsense at product demonstration meetings, and strangely, most of the opinions he gives are the core technologies of new products.

Sure enough, Mr. Wang’s first words made them pick up big materials.

"This time, we will hit the frequency of 1ghz..."

The scene was silent, and they all looked at Wang Anran with their eyes wide open.

The new technical director Gu Liang asked the first question: "Mr. Wang, the CPU frequency is 1ghz, so how much is the front-end bus frequency and how much is the frequency doubling design. The current North Bridge chips, PCI buses, memory and other buses may not keep up with the speed of the processor, which will form a huge bottleneck and cannot play the performance that the processor should have."

"good, this question is well raised."

Before Pentium, the front-end bus was equivalent to the CPU frequency. After Pentium, the CPU frequency increase speed was much higher than the speed of the front-end bus, which introduced the concept of frequency doubling.

As for the 1ghz CPU frequency proposed by Wang Anyran, people in the industry quickly realized that the bus frequency must be greatly increased in order to exchange a large amount of data with the North Bridge chip to exert the performance of the CPU.

This question is easy to explain. Suppose you use Ryzen 3900 processor, which is not equipped with DDR fourth-generation 3200 memory, but the first-generation DD memory, with a performance loss of at least 80%.

The reason is very simple. Data exchange cannot meet the needs of processor computing, which means that the processor cannot feed it at all.

After the front-end bus frequency increases, the corresponding PCI configuration cannot keep up, so the bottleneck always exists.

Wang Anran took a sip of water. It was obvious that inertial thinking, the frequencies of the front-end bus of the FSB and the PCI bus have always been the same. In fact, there is no need for FSB to wait for the device bus. If you can run faster, you will naturally have to run faster.

This involves memory technology. The front-end bus is closely related to memory. For example, in the Pentium 4th century, the northwood core processor, the 800mhz FSB front-end bus and the dual-channel DD were completely integrated, and the PCI bus interface was only 133mhz.

As before, although Wang Anran knew the standard answer, he was unwilling to say it directly, otherwise his engineers would be less fun. After all, not everyone likes spoilers.

Wang Anran looked at Gu Liang and said, "Director Gu, what did you think?"

"The CPU frequency is increased to 1ghz. According to previous estimates, the front-end bus must be increased to 400mhz. The memory area is now 166mhz. After upgrading to the 180nm manufacturing process, the gap is not too large. The equipment bus can be synchronized, which requires the progress of the entire industry."

Wang Anran took a sip of water and said as if nothing had happened: "If the industry does not progress, shouldn't we CPU develop?"

Wang Anran said this a bit heavy, and Gu Liang scratched his hair awkwardly.

Being firm in the idea of ​​keeping one's own words and not innovating is a dead end for IT/IC companies. As a technical director, you should understand this more. What the team needs to do is solve the problem, rather than being led by the problem.

This is a point that needs to be hit.

"The CPU frequency is higher than the front-end bus, why is the frequency of the front-end bus and the device interface consistent?"

After a clear hint, someone finally thought about this problem.

Senior engineer Yu Xiaotao said: "Mr. Wang, we can try to introduce qda frequency doubling technology between the front-end bus and the device interface."

Once a layer of window paper is pierced, it will be meaningless.

Wang Anran nodded and said, "QDA frequency doubling technology is just a technical measure to adjust the relationship between the CPU main frequency and the front-end bus and the device interface frequency. We need to do more than this."
Chapter completed!