Chapter 595: Gather your comrades, I have something good!
Lou Xiao'e had no objection to this suggestion. She probably wouldn't be able to cook the meal today, but she didn't want Zhendong to starve.
Gao Zhendong brought the rice bowl and the couple went straight to Shazhu's house.
"Hey, why didn't I ask you for a favor yesterday? You will be busy eating it today. I really have you."
In front of his own people, Silly Zhu's mouth was still so cruel. He was busy adding food while teasing Gao Zhendong with a smile.
"Hey, by the way, if you have any debts, just collect them as soon as possible, haha." Gao Zhendong didn't take it seriously. He was one of the most relaxed places in Shazhu. This kid has his own intentions, but he doesn't need to focus on his own people.
"Zhendong, what do you think our department asked me to do today?" Lou Xiao'e, who was tutoring He Yuyu on his homework, looked up while He Yuyu was doing calculations.
Gao Zhendong probably guessed it, and his movements were quite fast. But he didn't give the answer directly, which would ruin Lou Xiao'e's interest.
He didn't know that as soon as the technical department of the No. 3 Factory brought the materials, Lou Xiao'e's department quickly completed the information, except for the technical information that needed to be studied.
The head of the department looked at the National Defense Work Committee looming in the material, and felt even more happy. Even though this thing didn't seem to have high technological content, its specifications were not low.
"What are we doing to you?" Gao Zhendong pretended to be stupid.
"The head of the department said that I studied well and asked me to take part in a microbiology project to help the teachers." Lou Xiao'e was just vomiting, but now she calmed down and talked about it.
Gao Zhendong looked surprised: "Ah? Really? That's great, it's amazing. I don't even understand microorganisms." His tone was like coaxing a child, but Lou Xiao'e was still very proud of it.
Silly Zhu was very excited when he heard this: "Microorganisms? By the way, your school is Beijing No. 2 Medical University, right? What a coincidence, it should be our subject."
Lou Xiao'e also thought it was a coincidence: "Really? Then why didn't I see you?"
Silly Zhu smiled and said: "I am responsible for the cooking technology for that subject, and the comrades over at Zhendong are responsible for other technologies, so the one who went over was from the technical department of their third factory, and I was still cooking noodles in the factory."
< br>
A few people were chatting and laughing, and Lou Xiao'e expressed her learning experience: "Zhendong, don't talk, I looked at the knowledge about microorganisms, it's very interesting."
What she studied should be a major similar to clinical practice in the future. She didn't learn the knowledge of microorganisms, but since she wanted to enter this subject, she had to make up for it.
This study piqued Lou Xiao'e's interest. This small world is so interesting.
Bureau seat: It’s so fun.jpg.
Gao Zhendong smiled and said: "If you are interested, just learn. There are so many skills that you can't overwhelm yourself."
Lou Xiao'e nodded vigorously: "Well, find me a book."
There are not so many books on microbial research.
For Lou Xiao'e, if Gao Zhendong is indecisive, it will be over, and there is no need to think too much about anything else.
Not studying? There is Zhendong.
No direction? There is Zhendong.
No teaching materials? There is still Zhendong.
Gao Zhendong agreed: "Okay, you look at what you have now, and I will look around for you later."
Silly Zhu couldn't get in the middle of the conversation between the two, but He Yuyu was fascinated by what he heard. "Sister Xiao'e is so amazing, I want to be like her."
While eating, Gao Zhendong discovered that one dish was noodles and still salty.
Damn it, boy, get even with me.
——
The DIP sockets from Factory 1274 cannot wait in a short time. After all, it still takes time from design to mold to production, but the IC principle circuit board in Gao Zhendong's hand has already been soldered.
Although it was a bit ugly and a lot of wires flew out, I finally debugged the circuit.
The specifications of the entire circuit board are roughly the area of a DJS-60D, 50*50cm, but the entire area is not used up.
Gao Zhendong connected the circuit board to the DJS-59. There is no serial port and there are quite a lot of wires connected. The parallel port is like this. There is at least one wire for one bit of data. It would be even more exaggerated if the addresses are not reused.
Fortunately, Gao Zhendong has already designed a simple peripheral bus for DJS-59 and 60, which can partially reuse address and data lines. The situation is slightly better, but the price is that it will reduce the speed and increase the programming complexity of the underlying driver.
Gao Zhendong first compiled a simple program on the DJS-59. There is nothing special about this program. It is not even targeted at the circuit board, but at the DJS-59 itself.
The program is very simple, multiply two single-byte numbers 15 and 15, 100,000 times! The result is 225, which can still be placed in one byte. While calculating, calculate the total time taken.
They are multiplied separately, and the two multipliers each time are the same two 15s, instead of multiplying 15 times 100,000 times. The number of 15 raised to the power of 100,000 is not enough to be described in astronomical numbers. Of course, compared to Abnormalities such as Gregorian numbers and Tree3 are still insignificant.
The program was simple and compiled quickly. Gao Zhendong entered the compile and run instructions.
The program started running. Compared with the speed of the computer, it can be said that it took a long time. After taking a look at the time, Gao Zhendong made a "tsk" sound in his mouth and shook his head. He knew the general range of this result. It was definitely ugly. Now he just wanted to confirm the specific number.
Then Gao Zhendong started to write another program. This program was a bit difficult to write, mainly because it required direct operation of the underlying hardware.
The operations are the same, repeat 15*15 100,000 times.
Moreover, Gao Zhendong must deduct the communication time in the program, because this is the DJS-59 communicating to the test circuit board, controlling it to perform calculations, and then transmitting the data back.
The operation may be fast, but the communication time is long. Gao Zhendong's purpose is to deduct these communication times to roughly calculate the total time required for operation.
Fortunately, this thing is not too complicated, and Gao Zhendong's movements are not slow.
An hour later, Gao Zhendong completed the compilation and started the program.
This time, the time was longer than DJS-59 calculated. Gao Zhendong had expected it and was very patient.
There is no way, the communication speed of the peripheral bus is too far behind the internal bus.
After a long time, a number popped up on the screen. Gao Zhendong glanced at it and smiled, this is what he wanted.
Compared with using DJS-59 for calculation, after deducting the communication time, this IC experimental circuit board is ridiculously fast at the same frequency.
——Twenty-five times!
This is what Gao Zhendong wants.
——Single-cycle hardware multiplier!
The DJS-59 does not have a hardware multiplier, not to mention a single-cycle hardware multiplier, not even a multi-cycle hardware multiplier that controls the adder to implement multiplication.
The former can complete the multiplication calculation in one instruction cycle, but the latter may not. For example, the 51 microcontroller can complete a multiplication operation in 4 instruction cycles.
Needless to say, the former is fast and expensive, while the latter is slightly faster and slightly more expensive.
The number of DJS-59 is huge. Without a hardware multiplier, it can only control the adder and register step by step to complete the multiplication operation through program instructions.
For DJS-59, it takes about 15 instructions to complete an 8-bit integer multiplication operation, and there are many multi-cycle instructions. Taken together, it takes about 25 instruction cycles to complete a multiplication operation.
And this is only the case of 8-bit. If it is expanded to 16-bit or 32-bit, this number will increase.
As for floating point numbers, the situation is even worse.
This is the weakness of DJS-59, and also the weakness of most early general-purpose CPUs. It cannot perform more advanced mathematical operations.
But for signal processing, multiplication is the basic operation and the most commonly used operation.
This is the reason why general-purpose CPUs are very weak in signal processing. It can't be calculated, it's a problem in the mother's womb.
But Gao Zhendong’s next step, everything, and the final goal is basically directed at radar. With DJS-60D, even if it is IC-based, it is still a weakling and not good enough.
So Gao Zhendong was thinking about making a hardware multiplier by himself, and his goal was a single-cycle floating-point hardware multiplier. He praised the Internet and his mentor. In his previous life, he had implemented this thing in VHDL, and he knew its structure clearly. very.
VHDL, Hardware Description Language, one of the IC design languages on CPLD/FPGA, can directly design the IC hardware structure.
Although the final result of VHDL is based on logic units, which is different from the direct connection using transistors or logic gates, the algorithm it is based on is Gao Zhendong's fault, and it can still be exchanged.
This is why the circuit board of this multiplier is so big. If it is just a hardware multiplier implemented by an adder, then just add it to the DJS-60D. The increase in cost and area is not much, but accordingly, The performance increase is not much. This needs to be done, but it is not enough for Gao Zhendong’s needs!
What Gao Zhendong is doing is a 16-bit floating-point single-cycle hardware multiplier, which is absolutely far ahead.
Gao Zhendong happily changed the multiplier of the second program to two 16-bit floating point numbers and started the calculation again.
This time, he did not torture DJS-59. This kind of calculation was really tormenting the old guy DJS-59, although it was still much faster than humans.
Soon, the calculation results and time consumption jumped out.
Gao Zhendong checked the calculation results he had prepared, and they were completely consistent. As for the time, it was the same as calculating the multiplication of 8-digit integers.
This is the charm of a single-cycle hardware multiplier. As long as it does not exceed its design specifications and the numbers involved in the calculation are within its design range, no matter what it is calculated, it will be the same speed - a single instruction cycle.
Gao Zhendong burst out laughing. Although this was just an intermediate result, just a principle circuit board, for quite a few comrades, the modification was actually usable. After all, quite a lot of radars have volume requirements. Not high.
And compared with the previous back-end signal processing circuit, this 50*50cm board is definitely compact!
Gao Zhendong happily ran to the confidential room and asked for the phone number of the leader of the National Defense Working Committee.
On the phone, he only said one sentence to the leader of the National Defense Work Committee: "Leader, please summon comrades with sufficient level to work on radar. I have good things!"
(End of chapter)