Submit a news tip



How Wii U maintains high efficiency with low power consumption

Posted on October 10, 2012 by (@NE_Brian) in News, Wii U

In making Wii U, the development team had a particular goal in mind. The staff wanted to maintain high efficiency and performance while keeping power consumption low.

How the team accomplished this was explained in the latest Iwata Asks. Nintendo president Satoru Iwata, Genyo Takeda, and Ko Shioto delved into this topic.

Takeda explained that the Wii U uses a multi-core CPU. With it, “data can be processed between the CPU cores and with the high-density on-chip memory much better, and can now be done very efficiently with low power consumption.” The console also makes use of a MCM, which houses the multi-core GPU, CPU, and LCL chips. Shiota noted, “By putting LSI chips in this small package, the power necessary for communication between LSI chips drastically fell.”

Iwata: What was the key to achieving low power consumption and high performance this time?

Takeda: First of all, adoption of a multi-core CPU8 for the first time. By having multiple CPU cores in a single LSI chip, data can be processed between the CPU cores and with the high-density on-chip memory much better, and can now be done very efficiently with low power consumption.

8. Multi-core CPU: A central processing unit (CPU) with multiple processing cores (hardware for running software) for executing commands within a single chip package.

Takeda: And that we adopted an MCM.9

9. MCM (multi chip module): A component with multiple silicon chips called bare die mounted on a single substrate.

Takeda: This time we fully embraced the idea of using an MCM for our gaming console. An MCM is where the aforementioned Multi-core CPU chip and the GPU chip10 are built into a single component. The GPU itself also contains quite a large on-chip memory. Due to this MCM, the package costs less and we could speed up data exchange among two LSIs while lowering power consumption. And also the international division of labor in general, would be cost-effective.10. GPU (graphics processing unit): Also called a graphics chip or video chip, this is a special chip for rendering computer and game console displays.

Iwata: A big challenge this time was putting silicon chips made at different semiconductor plants into one package. Shiota-san, as the person who actually had to make that happen, what hurdles did you encounter?

Shiota: The LSI chips were made at different companies, so when a defect arose, it was difficult to isolate the cause. In defect analysis, it was inside the MCM, so figuring out the problem was incredibly difficult.

Iwata: When it’s actually running, it’s all inside a single box, so you can’t easily observe what is happening.

Shiota: Right. We really drew on the wisdom of Renesas11, IBM12 and AMD13, who cooperated with us. To isolate the problem we devised a way to have a minimum amount of signal travel outside of the MCM, so we could verify the problem with the minimum amount of overhead.14

11. Renesas Electronics Corporation: A semiconductor manufacturer with its headquarters in Chiyoda Ward, Tokyo.

12. International Business Machines Corporation: A company providing computer-related services and products. Its headquarters is in New York State.

13. Advanced Micro Devices, Inc.: A company that offers development, manufacturing and sales of computer related products. Its headquarters is in the state of California.

14. Overhead: Excess burden in processing that is ancillary to the original process.

Iwata: But it must not have been easy to reach that point.

Shiota: No. We made decisions as we gathered data on our past experiences, but we didn’t notice some things until we actually ran it, so we had to give feedback and repeatedly slog through such areas.

Takeda: They were all different companies, so when it came to defects, it was like, “That isn’t our responsibility.”

Iwata: Usually when there are defects, you would fix it so it doesn’t happen in the first place. The moment that programmers run a program they have made, they hit the key thinking, “Of course this will run!” And when you tell them doesn’t run, they think, “It must be a problem somewhere else.” In the same way, when you pack in chips made by different semiconductor manufacturers, it’s only natural for everyone to think, “The problem must be somewhere else.” Shiota-san, how did you handle that?

Shiota: Simply put, I adopted a policy of “Prove your own innocence.”

Iwata:Oh, that’s interesting! (laughs)

Shiota: First, with regard to the LSI chips before inclusion in the package, we tried to establish a way to test without missing anything, and each company devised an extremely robust method. That way they were able to greatly lower the possibility of defects. They were kind enough to provide the all-important information for defect analysis based on the solid data that had built up.

Iwata: Did that process go smoothly?

Shiota: It took time to reach that point. At first, there was talk to the effect that if we put in a certain process, we wouldn’t be able to make a lot, or that investment in production facilities would balloon. But if you put your heads together, solutions present themselves, so by a little fiddling with existing test facilities, we found an extremely efficient test pattern and managed to do it.

Iwata: There still aren’t many instances where all these key components of a system is built into a single chip.

Shiota: There are a few, but there aren’t many examples of something mass produced like this with a CPU and GPU of this caliber in one package. This is the MCM substrate.

.

Iwata: This single chip contains the console’s heart. They were separated to two chips in the GameCube and Wii. Is the reason you focused on an MCM because you thought the results were worth it?

Shiota: Yes. As Takeda-san said, lowering power consumption has been our position since the GameCube. By putting LSI chips in this small package, the power necessary for communication between LSI chips drastically fell.

Iwata: Compared to power flowing between chips in separate physical positions on the board, you can get by with less power inside a small module. The latency is also reduced, and the speed increases.

Shiota: Yes. And by putting them in a single small package, we can make the footprint on the CPU board smaller. For the contribution it would make to casing miniaturization too, I wanted to do it no matter what!

Source

Leave a Reply
Manage Cookie Settings