Hi, this is Wayne again with a topic “Adventures at AMD: Origin of Threadripper”.
Have I got some good news for you if you’re like me and you had some ubiquity gear, that uh was getting a little flaky and you tried to do a warranty and found out that? Well, they don’t honor the warranty when they come in packs, even though it’s discolored in yellowing and they’ve changed Hardware revisions, because there’s some Hardware problems and that sort of fun, stuff or there’s been a firmware update that reset all the Mac addresses on your network again. For no apparent reason we’ll get some good news for you. Ingenius has a tradein program. This promo will let you take your old ubiquity gear and get new ingenious gear for half off Poe switches, it’s the stuff in the fit line. It’S the stuff that I’ve reviewed before I’ve been very happy with. In fact, I was so happy with it. I’Ve been setting up fellow Youtubers with it.
You know Barn CLE, it’s like what. What can I get Barn? Cle, that’s going to be low headache for me rock solid and will give his whole family Wireless literally everywhere. That’S why I chose it. It’S not even part of this ad.
It’S just the thing that I did so uh. That’S enough of that ingenious is sponsoring this video and uh. That’S the promo. They wanted me to promote and I’m more than happy to promote that promo, because uh I’ve had good experiences with ingenious gear, so check it out and now on with the video. It absolutely would disrupt. We had something that was special, that will change the the Paradigm of how people would think of AMD. The world’s most powerful desktop has a nice ring to it. 3D vcash AMD secret weapon, bringing Gamers the best possible performance first in desktop now in laptops, but that’s not what this video is about.
This would never have been were it not for this AMD ryzen thread Ripper. This thing is 16 cores and it launched in an era when we didn’t even have six cores on the desktop as the norm. It was four cores and they had just launched the desktop chip. That’S eight cores, and then this comes along available in 16 and later 32 cores, but if it hadn’t been for this 3D vcash wouldn’t have been a thing, but don’t take my word for it. I think the the success that we had with that initial thread rer products, I think, set the tone for how we look at disruption going forward and I feel like a product like x3d, would not have come to fruition if he had had not seen that success With that early generation, the majority of folks, whenever you want to do something, the first question that comes up is why why do you want to do it right? But why not bring that to the Enthusiast? Why not bring that to the the the consumer and give them server likee performance at at a cost that they can they can do stuff? Why not? Why can’t we do it rather than why are we doing this um and I think that permeated afterwards, the X3, the things that we did were probably a continuation of that us asking ourselves. You know what that. Why not has some value too and it’ll resonate with the people who ask that to themselves every day also got to learn more about the sausage making process? That goes into something like this, because this was an engineering project.
First, the engineers said we can build this. We can turn it into something and it’s going to be awesome. We were in like some meeting and they were talking about Naples right and they were looking through Naples. They were just looking at backend assembly losses and they’re, like oh, we’ll, have a bunch of like these dies that are not good enough to make Naples. What do we do with that? And someone just suggested hey Intel just has like a 8 Core Extreme edition part.
What if we just put two of these together, like that’s an interesting product, let’s look at it further and then that kicked off this whole stream of ideas, and once we started going through that, it became more and more real, and no one could talk about it. Because we didn’t know if we were going to do it, so we didn’t want people committed to it and then somehow it became real and I think a lot of that credit has to go to Travis because he just apparently decided to get in front of Jim Anderson’S car and decided: oh wow, that’s a great way to get a product approved, and then they were able to approach the executives on the marketing side and the company side and convince them to actually productize it. Because I mean, let’s face it Engineers. They always have a lot of ideas. Not all of them can be sold to other people. I do know them feels we had come out of these meetings.
We’D come up with several ideas. Um I happened to be with in Taiwan with Jim Anderson and I said: hey hey. I got some ideas to pitch with you and he goes great uh uh. When can we meet he’s like well I’m driving to the airport tomorrow morning at 7:00? Why don’t? Why don’t we ride together and, I said great, he said good I’ll meet you downstairs, so I got down there early CU, um Jim’s a busy person.
So I want to make sure that that that that I found him and stuff and found his car and I’m sitting outside Jim, come comes running down on the phone gets in the car, um car starts to take off and I jump in front of the car And slam my head down on the hood and said you forgot me he goes yes, I did get in the car and so he’s on he’s on the phone for, like the first half of the drive to uh to the airport. So I had like, I had like 20 25 minutes to pitch three ideas to him: he loved them all um. He came back with the thread, Ripper idea, um and said we got to make this happen and having ex having someone uh like him at the time. That’S got um that can persuade and and influence everyone. Uh was amazing because he just took ownership from that point. He said we’re going to go make.
This is the best thing ever so bringing up a processor like this is a huge involved process. It’S not just taking a server CPU and repackaging it for desktop, there’s, actually a lot of work that has to go into sort of requalifying these CPUs for desktop. You thought, okay, it’s going to be a quick bring up. I still remember the first bring up the first boards are there and they are about to put the thread up apart and me and Travis are standing behind, while everyone is waiting for the part to boot, all me and tris could think about is.
Why is the carrier frame blue right and fun fact? You know why it was blue cuz. Naples is a server part. If you look inside a server, any user replaceable component is highlighted blue. So if you look at Fan tabs dim tabs, if you look at CPU, so the way to get the CPU out was to pull it from the tab and that’s why it was blue.
In the minute. We said that one of these mechanical guys standing at the back and bring like. Why do you guys want to change the color of the carrier frame? I’M, like you don’t understand. This is like the most critical thing and they’re like oh once, the heat sink is on or the socket retention. You can’t see it, I’m like no. A little of it sticks out right.
So it matters right, and it was just that attitude of no compromises and bring up and the performance characteristics in terms of power. Longevity and reliability are uh tuned differently. Let’S say for a desktop part. That’S why you know if you just repackage a server part for desktop, you wouldn’t expect the desktop part to be dramatically faster and it is thread.
Rer has consistently broken world records and, while everyone’s trying to get the to just work and not like B, saw or not do things both of us are trying to break the cinan record right like we have no other business. Being that, like, we are just trying to break the world record, we’re like hey, wouldn’t be cool. If we can push it some more and then because he has a cash background, so he’s giving us all these bits to change the cash performance and buyas it we’re trying to make it go faster and faster and faster. Meanwhile, like this thermal guy comes in like oh, we getting all these thermal shutdowns, so we like – oh, we don’t worry about that they’re like why I’m like we’ll just walk it over to Bill’s lab and just put liquid nitrogen and keep going right.
So that’s a lot of the business being in the overclocking team for us like it’s not our day job, but it’s just something we do out of hey. We are trying to figure out what’s wrong next, so bring up in thread. Ripper was all about pushing it, because the product for the most part was working because it was leveraged. This product line has spawned an entire industry.
You know sort of a revitalization of the workstation industry. Up till this point, the single thread performance on these high-end desktop platforms and workstation products uh, would pale in comparison to their desktop counterparts, meaning that a Plucky little four core is going to run really fast, like a Corvette, but your high-end desktop is going to run Like uh, something that’s a bit slower, you know a tractor trailer or something that’s not really true anymore, with thread Ripper Pro and thread Ripper, with its triumphant return to high-end desktop yeah, but like when we first were were working on thread Ripper it. It was kind of this.
This one shot thing you know where we’re going to you know, have the crown and say hey, we can do it right and we weren’t sure if anyone’s ever going to actually buy it. You know it became like you were a victim of your own yeah. It was selling like crazy and people are and like. What we didn’t understand is that we we we knew people didn’t need all the cores. What we didn’t understand is that people wanted all the cores people were buying. These machines like like you, would buy a hot rod or you you buy some race. Car you’re never really going to track it, but you want that sort of capability.
People were building these hot rod machines for no reason, but because they wanted to and so and they were real use cases as well, obviously, and so because of that now yeah now we had to do kfax and what do we do? Okay, I guess we’ll we’ll put four chips in there and make it 32 core which had its own challenges. Now. My full interview here was over an hour long. I think that’s going to be a patreon and float plane exclusive. A longer Extended Cut version of this, but I’ve also got a lot of other content like how do you handle uh platform errors like platform error handling? Sometimes you hear me say things like platform. First, error handling or operating system. First, error handling in some of my videos. Well, I got to meet the team behind that ask a ton of questions and learn about new features in the Enterprise that AMD is rolling into their server platforms.
Not just that, but I also get to meet the kernel team and learn about the Linux kernel, integration and the patches that are Upstream now, sometimes those make their way to desktop, sometimes not sometimes uh, you know left hand, doesn’t know what the right hand is doing. It’S a sort of a fun thing in all companies. Don’T worry about that, but as far as process goes as far as I can tell, this is one of the most startup feeling Fortune 500 companies that I’ve ever been sort of behind the curtain on it was really uh sort of magical and fantastic and you’d be Surprised how many Fortune 500 companies I’ve been inside, doing a janitorial things or uh unmu things or just you know, sort of helping with this that or the other it it’s. It seemed like one of those uh one of those Mission, Impossible movies. Where you you it’s like, I need. I need a really good guy who can do clocking.
I need one core expert that knows knows how to push the limits on that part. I I need a platform guy who can do this, and so it was just like a a small set of teams. Yeah and the challenge with that is, as you grow and as you get um expectations become higher. How do you keep that spirit, and I think the team’s done a really good job across the board of keeping that Spirit, not only for the people that have been here for a long time, but but when we bring people in we that that spirit is very Um uh is just cross-pollinated, so it it.
You know it’s great. It’S it’s been great to see people coming in from the outside and then you know it’s infectious and it it gets part of their Spirit as well. I’Ve got so much content from this trip. It is just unreal, so stay tuned for the content. You know if you subscribe like that sort of thing and if somebody asks it’s like hey, where did vcash come from? It came from The Winds of the engineering team with thread Ripper first and thread Ripper dominating.
So well, and so completely gave the executive team the confidence in the engineers to be able to pull off something like 3D vcash and probably some other products that are coming down. The pike and big thanks to the AMD team for hosting me and putting up with me and also letting me dive through the dumpsters and uh you know I wish I could have kept the hard drives that I found in the dumpsters those weren’t supposed to be There and wonder this is level one does been a quick look at the AMD thread. Ripper Inception from the beginning, the origin story, I’m signing out.
You can find me in the level one forms .