 And we're here at the Linaro Connect 2018 here in Hong Kong, and who are you? Yeah, I'm Pao Wang. I'm from Peking University, China. And I'm also the leader of Thailand Development Group of GCC, Green Computing Consulting. So, Green Computing Consortium, it sounds like very important. We don't want to use too much power, right? Yeah, sure. That's about energy-efficient stuff. So, we believe that even for servers, we do not need much power consumption. We do not want power-hungry servers. We need energy-efficient servers. And in that way, we believe that ARM 6.4 is a good choice for the server processors. It's very important because so many people want to use a smartphone. They want to use a computer. They want to have more cloud services. Yes. The demand is going up very fast. So, if we just continue to make big energy services, it's not going to work, right? Yeah, sure. So, let's look at this. And in the PCH, in the PCH server, X86 was dominant. That was before, like this? That was before. And ARM was worse at the smallest scope here. But when we move to more modern age, and ARM grows up, and X86 is also okay. But in the future, even for example, in the cloud computing age, we believe that there are three trends, multi-core, heterogeneity, and integration. And what is the most important thing is that the power consumption. And nowadays, energy efficiency is very important. So, we believe that if we want to reduce the power consumption at servers or at data centers in cloud computing age, we need to use say, say, for stuff. So, it's very important because there's a lot of things happening. You're talking, for example, about the 7nm process. It's basically more slow. Maybe it's slowing down a little bit. But there's now more cores. Yeah, more cores. A little software. Yeah, yeah, yeah, software. So, that's okay. That's another trend. We need parallelism to support more workload. But I think even then, the power consumption is very important at the server side. So, our focus here is energy efficiency, the power consumption. So, I've been doing videos about the ARM servers for many, many years. So, there's ARM and now Qualcomm, High Silicon. And there's also KVM solution. Can you say something about how do you work with them? What are you looking for with them? Okay, we have a common idea to gather a group here. And you can see that those ARM is an IP wonder. And Qualcomm and High Silicon are system or chip wonder. And those are system wonders. And we also have Chinese government to support others. Yes, this one. This is a Ministry of Industry and Information Technology in China. And so the MIIT, the Ministry of Industry and Information Technology, just supports all these wonders to gather a group to build a specification to make the ecosystem healthy in China. Are you saying, is it possible that maybe China will say, we want ARM servers now and will happen? Is that how it works or it's not how it works? I think that's an option. Option? Yeah, that's an option. You need to keep the options open, right? Yeah, sure. You want to develop everything so if it's an option, you should be seized. I think that's my personal idea, right? At the government side, the government will not say that you should use 86 server or ARM server. But the government will say that ARM server is also good and it's power efficient. So the government will say that you should develop these terms at least energy efficiency side. Hopefully China is doing a lot of renewable energy to keep the pollution levels lower, right? Yeah. There's a lot of change happening, right? Yeah, sure. Maybe it should be kind of like a rule. Every WeChat message, maximum power, like some kind of thing where you don't want the cloud company, maybe they have so much money, maybe they don't care to pay more money for power. I think they will. But they should care. Yeah, they should care. That's the responsibility of the government, right? To regulate, maybe regulate. Yes, regulate. The regulation is very important. Just like maximum power per cloud? Yeah, sure. I believe that. Per unit of the cloud, yes. Performance. And it's like the cars, when they do the cars, it needs to be more efficient. Yeah, yeah, yeah. It should be the same for the cloud, data centers, right? So you know that there is a list called top 500, top 500 super computing, super computer. And there is also a list called top green 500, green top 500. And I believe in my information that the Chinese government just pay more attention on the energy efficient. Because China has the biggest supercomputers. Yeah. But also the most energy efficient? I don't know. It's possible it's in the top, right? Yeah, but at least I know, I think that the government has some idea that they want to make the servers or the supercomputing most power efficient. Because as far as I understand, the supercomputer Chinese one is many small cores. Yeah, sure. And so it's parallel supercomputing, a lot of parallel. Yeah, yeah, yeah. And the power efficient, somewhat the power efficient. But I don't think that's the top one. And maybe you can talk about what is the organizational structure right here? That's the organizational structure of GCC, green computing, current consulting in China. So we have general meeting of members, advisory committee, board of directors, technical committee and secretary tyrant. But the most important, we have... Not mostly, sorry. Yeah. But far the important working groups are standards and tests and a system enabling and pilot program, pilot development and IPR. So we can see the details there. So secretary tyrant is the permanent department which is responsible for organization, operation and working group support. And the standards and the test working group defines GCC technology standards and test the specification and test and the verification to a wide industry spirit. And the system enabling group co-operates with an inaugural to promote the system enablement via R&D and to development. And we have a pilot program, working group, promotes application demonstration for both OTC and enterprise customers to drive ecosystem build up. And the IPR working groups, my working group manages IPR related issues case by case. And I'm the leader of Thailand development working group. So we establish the Thailand development programs in academy to ensure the sustainable development of green community industry. So that means you are recruiting students that can understand the armed service for example? We educate students to understand the armed service. And we bring the materials to the campus to give courses. So it's going to be very interesting how to optimize the software on the arm. Yes, yes. How to get the most performance per watt. Yeah. Optimize software, make better software. Open source, everything open source? Yes, everything is open source. But the first thing is that we need a specification. We need a unified specification to unify, to cooperate, to coordinate all those system wonders of the armed servers. Otherwise, you know, in 1886, okay, a very good thing about 1886 is that it's unified work, right? Everything that you write, the unified kind of program. Each system wonder, you just provide very similar systems. At least in terms of the specifications, very similar. And then you do not worry about the compliance, right? In this world, compliance stuff. So that means we have the specification of the armed servers and what we actually, we propose the wide paper of the armed servers. So this is something like this. We release green computing standards, wide paper at the World Internet Conference in December last year. So what does it say, the wide paper? The wide paper just talks about how to build a system level for each system wonder to build a unified standard of the armed servers. That's for the system wonders, like the Nobel high silicon. One other thing that I wonder, very big companies that are big cloud companies, they're making millions of servers. So they have big partnership with Intel. So maybe there's also a big business aspect. They cannot just switch to armed because Intel will be angry. So I think many of them maybe have to be careful but they're testing, but if it's ready, they will switch, but they know when they switch, Intel will be angry. Sorry, I don't know if it's any relevance because there's not only the green, there's also the business. Like that kind of green, the dollars. I think that's the, this is my personal idea, right? My personal idea. I think Chinese, at least the Chinese government pay much attention to green computing, right? So this is called the green computing consulting. And right now, at least right now, armed server is the best in my personal belief. Armed server is the best for green computing. Is it a big ecosystem? Not because of the big ecosystem. I mean for the power efficiency features for the armed servers. And Intel is powerful, right? But I believe Intel should pay more attention to power, hungry servers to make the energy efficiency more energy efficient. I think otherwise, you know, suppose we have a boundary, we have a maximum boundary of the limit, maximum limit of power consumption. And what if Intel cannot reach this? They have to think about it. Yeah. And if ARM help Intel think about that, it's also good. I think that's also good. It's actually a success for ARM if they can force Intel to make a greener server. Yeah. Because we aim at technologies, right? We aim at the human future. We do not aim at a single wonder, right? So I think at least that's my personal belief. Because I guess that in China every year the cloud computing is doubled or something. It's growing like crazy. So that means in five years it's going to be a hundred times bigger. That's what it means, right? Double, double, double, or something. Or ten years or something. That's a big challenge. Yeah, sure. And I think the energy efficiency is at least among the most challenging stuff. So it's very urgent. Yeah, very urgent to think about this. Your work is urgent. So how do you feel about Linaro? How's your collaboration with Linaro? How long you work together with them? Okay, so let me check. So we work together with a system-enabled working group to coordinate with Linaro to promote a system-enablement R&D and a two-development. So Linaro is, I think that's the... Linaro has ERP platform, right? So we work with Linaro. Linaro can also help us to promote the GCC ideas there. So I've been doing Linaro videos for a long time but it would be exciting to see even more collaboration, faster collaboration in the open source. Yeah, open source. I guess Linaro can play a nice role in just more and more engineers like a very advanced, crazy, good engineer all over the world working together. Yeah, sure, sure. We also talked about collaboration in the Thailand Development Program. So in that program, Linaro will provide us some materials for some trainings in the conference to make the university students to understand what should be good power-efficient software open source software and the specification how to write the programs with more power efficiency. Maybe there should be a route to go through for the students where they can maybe, at the end, maybe they get a job in Linaro or some other company that do open source. It's the interest of Linaro to find the next 100 engineers, right? Yeah, and maybe they would build their own companies. Yeah, it's open source, right? Yeah, that's open source. So they can do anything they want, right? Yeah, yeah, yeah. I think they need to be funded somehow. I think they need to understand this, right? I think the students need to understand what is the idea of open source, the idea of power efficiency. I think that's the most important. At least that's the idea of the Thailand Development Working Group. So how many students are there in China working on this? How many students are interested in the ARM servers? Thousands or no? I think maybe... How many students you have worked with? I have. How many you work with? I have several tens of students. Myself. And it's growing. Yeah. Myself. This is not for this working group. That's in my lab. Yeah, in your lab. But then there's more and more interest, right? Yeah, yeah, yeah. I think all the students in China are interested in energy efficiency, energy efficient computing. But open source, especially open source stuff. But the important thing is that they need some guidance to understand more or understand more accurately what is a good side of open source, a good side of energy efficiency. And I think one stuff is that for ARM servers, we really do not have much more, many ARM servers in compers. We have actually 86 servers. But ARM would like to, or the system wonders would like to donate some ARM servers to the university. Yeah, we need that. You need some Qualcomm servers, some K-Vim Thunder X2 servers. Yeah, yeah, yeah. And high silicon. And high silicon. Yeah. And the novel stuff. So hopefully there will be some, even some affordable development boards. Yeah, yeah, yeah. We should make them available. Those development boards. Yeah. We need that. High priority. Yeah, yeah, yeah. Then the students would understand, okay, those are good things. And they will learn how to use these things. And not like before. Before we only had actually 86 servers. Then they did not have opportunity to emulate. To emulate, yeah. They did not have any opportunity to see whether there are other opportunity in ARM server or in other architecture. So maybe also it's very important for some, I don't know if it's Baidu or Alibaba or somebody you should make, or another company makes some servers. Yeah. And you can put your software on them. Yeah, yeah. There should be more and more of that also. Yeah. Like for, you know, like test servers. Yeah. And need many of them. Yeah. We have, see this. We have standards and... Actually, we are building an ARM-based green computing based on a test platform called CEISI. And we will have two test labs. And then we install or we develop applications on those test labs. Students can access the internet. Yeah, yeah, yeah. Internet. That's open. That's open to the public. It's coming very soon. I think that would be pretty soon. Nice.