'반도체 사대천황' 짐 켈러가 대한민국 반도체에 대하여 말하다! (짐 켈러 텐스토렌트 CEO) | 광화문 초대석

Your Role has been change has changed tring us from CTO to CEO so what is the most significant change in your role and what concerns you the most well yeah about in the beginning of the year I became CEO um so then I have to run the whole company and then last year we raised some money on a on a note so that was a lot of work to do and we've been working on bringing our products to Market and rewriting some of our software to make it more efficient and clean so I'd say just got much more diverse and then we also built out a bigger sales team with David and customer support and then started building the proper operations and supply chain team and that was a lot of work we it's it's very different from the core engineering focus of you know build the the best product and work on Hardware software we had to expand our our scope but are you still involving in designing work or CH architecting work or yes the software there's operating system group a compiler group a low-level bare metal software team that work for me directly um and then on the hardware side so team uh CPU design team AI IP team work for me and I talk to them all the time about technical stuff I don't do design at the moment but I have lots of meetings with them about what we're doing while we're doing it I do lots of design review I'd say well Korean readers and Korean people at not really aware of tense tent so could you please uh explain your journey as a chip architect and how you came to be known as a living legend of Chip Chip design yeah I'm not really sure how that part happened so I started my career well my second job was at digital equipment I was there for 15 years and I worked I learned to be a computer architect on the back 8800 in the 80s and then I worked in the alpha design team at and I was co-architect of the second and third Alpha chips which at the time were the world's fastest processors they were in the Guinness Book of World Records I don't know how that happened really and then I went to AMD where I was the architect of K8 and I also contributed to the K7 design and then the x86 6 Force spec in hyper transport and some point I found out I had fans on a nonte there was a forum which had very long discussions about my CPU designs and Architectural decisions and then I worked on four generations of Apple products and one MacBook Air and I think they were also great products and people really like that and then I went to AMD and I was ran the Zen project and I was one of the principal Architects but also I built the the team and organization and methodology for Zen which is very famous in the PC World and then Tesla I was the first employee in the autopilot Hardware group and we built a hardware 3 chip which drove a car in 18 months which was a uh let's a world record for going from one person to driving a car that was pretty big then Intel I was SVP and I had a team of 10,000 people which is super fun I read it on on another interviewing article yeah yeah so so somewhere along the way people know that I guess one thing you recently mentioned was that you kind of pro provoked some Ultimus to it then you can make it cheap you can do it cheaper so so first of all I don't believe like the news read ran articles saying he wanted to raise $7 trillion well so we do two things we build an AI processor which uses RIS 5 processors inside of it to feed the instructions to the tensor engine M right and those processors are more about how they work than whether they're better or more efficient so the tensor processor is special but the RIS 5 engines in there we used them because we could build them the way we wanted to RIS five as a computer architecture is a specification right it's an open source specification so arm is owned by arm and Intel architecture is owned by Intel and AMD but RIS 5 is something anybody can do anything with so the reason I like it is I'm a computer architect I can build a new processor that's different and I think better than anybody else which we intend to do but it's not so much because it's risk five it's because it's open source and we can do what we want that's the key the more people that work on something the more chance there is for Innovation the architecture is great obviously T we've got a team with a history of delivering high performance compute but the fact that they can own something and they can customize it the way they want for their workloads it's really important to them then how did you establish the relationship with Korean company so I've been working with Samsung since I think 95 I did a project with a a digital for Alpha Four phone chips at Apple autonomous driving chip at Tesla and several chips with them at Intel and they were all successful okay so we get along pretty good apparently M and there's lots of really great people okay because oh speaking of the system LSI um well actually Samsung is developing own application processor AOS but it's not really popular so exos is not really popular gain attention or popularity uh so actually Korean readers or Korean there is a kind of atmosphere that ACD this failure of Samsung so can you give them on kind of advice to improve the AP I'd have to go through like building this technology is really hard okay and you have there's lots of pieces that go together to make the product and then there's also the business proposition and then sometimes you don't win the business for reasons besides the the technology so so you know you mentioned I think Jim talked about a very long history with with Samsung I think we're also really impressed you know I've seen the uh Korean company's adoption of new technology Ai and stuff like Risk five per it's been incredible so what's really impressed me is that you know we've had you know when we made some announcements what we were doing we we had good relationships with a lot of Korean companies but they were very proactive in adopting the technology some of the fastest companies in the world in terms of the adoption which is awesome so that's something I think that's also helped us gain traction here because you know we're doing AI we're doing risk 5 these are in the news a lot you see what's happening in Nvidia you see what's happening you know with risk 5 so I think the adoption and their willingness to try do technology has been pretty impressive as well okay uh may I ask when when you when did you start your um business the relationship with hunda yeah it was over a year ago oh some people I knew in Korea introduced us oh I see and then they wanted to make an investment and they're building all kinds of technology I I didn't know they build Automotive components bus and Dynamics obviously great cars I went I got a factory tour this morning how was it good mostly good I saw some room for improvements but they make very good cars do do you think that hyai can be a next Tesla I think they're going to be better than him a great car company I don't know if anybody's going to be a next Tesla why would you be a next Tesla okay they should be themselves oh and I think they're very good at that they make a lot of different products yes don't copy somebody then you're a follower I don't think they're followers I think they're oh you don't think they are oh that's interesting look at their cars they're pretty good you know one more company so LG as well LG was actually T torr's first publicly announced IP customer oh yeah so yeah you know I think we've had a lot of success in Korea with LG Hyundai Samsung and we work really closely with many software companies here as well so yeah it's a good market for us could you explain the importance of software in the auto automotive industry in recent era yeah so there's a big movement like they say the software defined vehicle because it used to be about a car and didn't have software right and then they had a little bit of software on a little screen for your GPS or maybe your temperature control but now modern cars they have an infotainment system they have a GPS system they have uh maybe ades they have wireless radios software updates and some features like a Tesla we put a fast computer in the infotainment system so you can play games apparently while you're driving so all those software components you know need compute and then the the devices in the car like your brakes and your steering are getting smarter so then all that software has to talk to each other so then the car itself has a really fast Network in it where all the different software components talk to each other and then when you write lots of software you need a really good development environment you don't just build software on the car anymore you build softwares in the data center test it build it and then deploy it to the car so that's a really big change for a car company cuz they didn't used to have any programmers and now they have a lot you have had multiple options and foundary so why did you choose samung overpowering on a chip pretty soon that's based on tsmc and then when we were looking for a partner for our next chip we wanted to go what's that quazar qu yeah so we wanted to build it with multiple chips and chiplets and there's a lot of design work around the chiplets the package the D dii the IP and when we talk to Samsung they have a really good team for like in-house at Samsung to bring a lot of that technology together like the reason I chose Samsung for the Tesla part is they were really full service place they did packaging design some of the physical design back end they provided some of the IP some from xnos and and that whole package is really strong I just read uh um they uh your LinkedIn post that you have kind of relationship with it rapidas in Japan too so yeah in what role they are uh are they different so in Japan we have a number of customers for our Ai Ai and CPU technology and they introduced us to rapidus and then the Govern the Japanese government has a very aggressive plan that's true to invest in both The Foundry technology chip design and big systems Japan often has l the world in HPC Computing so we got talking about what their plan was and I really like it and then some of our customers are going to use that technology so we're going to work together to do that I want to ask you how can Samsung drive or survive in this era some uh chip Wars I don't know they're pretty big foundaries they lead the world and in dams which is really good they have really good process they're in a lot of products people like to turn everything into a horse race Samsung's been around for a long time I see they have really good technology really good Engineers I like working with them because they have a good relationship and they deliver and they work well with us so I think that's one of the keys for business long run is good relationships and you know work together make a partnership to make a product I see though the South Korea Harvard industry is pretty pretty much uh flourished am I right um but the country is just started is just beginning to nurture the software industry to so could you do you have any advice for the government officiers or companies to fostering this industry like software industry well I'll support startups support startup okay there a lot of you know if you want a big sof forare a company you have to start with lots of small ones right um we like to hire interns I'm going to work on hiring some interns in Korea next year and and the cool thing about that is the student come and learn they also do a lot of work students are really fast to get up to speed but then that affects like what do they go back to school to learn and it also some people say I figured out by being an intern what I really want to study so it affects their course to study um but the AI wave has given lots of students a reason to go into software I'm sure that's happening here [Music]

Channel: 연합뉴스경제TV

Views: 55,652

Keywords: 짐켈러, 짐 켈러, 켈러, 짐 켈러까지, 팀켈러, 팀 켈러, 티모시 켈러, 티머시 켈러, 포스트잇, post it, 천재, 천재 반도체 설계자, 인텔, 애플, amd, 테슬라, 삼성, 암드, cpu, 씨피유, cpu설계자, 프로세스엔지니어, 천재엔지니어, 라이젠, 애슬론, 아이폰, 엑시노스, 엔비디아, jim keller, 한국경제tv 뉴스, 증권, 경제, 한국경제tv, 뉴스플러스, 뉴스+, 김민수, 삼성전자, 삼성전자 주가, 삼성전자우, 삼성전자 배당일, 삼전 주가, 삼전주가전망, 삼전 배당금, 삼성전자 주가총액, 삼전 주식, 삼전 실적, 삼전우, 삼전 배당일, 삼전 반도체, 삼전 반도체기술, 삼전 반도체사업, 삼성전자 반도체, 삼성전자 반도체 연구소, 반도체주, 반도체관련주, 반도체주 전망, 반도체관련주 전망, 반도체 전망, 삼성 ai, 삼성 ai포럼, 반도체, 반도체 지식, 삼성반도체, 반도체전공정, 시스템반도체, 반도체생태계

Id: v_HbgT6GzAw

Length: 13min 30sec (810 seconds)

Published: Sat Mar 09 2024