Find our main channel here: ru-vid.com Watch our EVGA Unsung Heroes interview: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-OniFkDc8Btw.html Watch our video with Amit Mehra on the origins of Zen: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-RTA3Ls-WAcw.html Watch our AMD lab tour documentary: ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-7H4eg2jOvVw.html
Why not ask why they didn't continue Phenom? Why didn't they cancel Bulldozer when early performance numbers came out? Was is Hector Ruiz his fault? Also the Jaguar CPU's were very successful. Why not talk about those aswell? Because that success saved AMD due to the Consoles.
I've been getting out on the bike a lot more this year (finally, after 2-3 years of being slammed). Was considering recording more of it again, but not sure yet!
Steve if you only knew how important these videos will be 20-30-40 years down the road. All of these interviews will be looked at in the future like "this is where we came from, this is part of the CPU history".
I can only hope that giving the right people the platform to speak on their experience will be useful to others in the same way interviews of 3dfx, early NVIDIA and Intel, and others have been useful to our team for learning!
@@gnextras imagine having equivalent talks with Jay Miner and his team from the 80s explaining stuff about Atari8bit and Amiga chipsets! Those talks will be priceless in the future.
could have sworn jim keller legit admitted he invented ryzen for AMD. because the timeframe matches up. he went to amd. bulldozer was failing. he made zen. and then left amd and like a year later ryzen launches first generation. from my sources, jim keller made ryzen, but it was arm based. and amd didn't wanna go all in on arm. so they adopted the zen design for x86-64 and that's what we got.
@@goblinphreak2132He said in the video Keller came in but let's not pretend one person is responsible. Keller has specific skill sets that surely allowed Zen to be possible but he doesn't do everything. He's the special ops guy you bring in heading into uncharted territory who doesn't play well with others but he'll keep you alive.
I thought Keller bootstrapped the entire thing although he didn't do it all himself, but often companies hire in top consultants to get things back on track and set the ship right, this happens in software too, some companies run nearly exclusively with contractors or consultants
About 2 years ago I discovered a show called "The Computer Chronicles" hosted by Stewart Cheifet that ran from the 80s right up to the early 2000s, cataloging some of the most interesting time in the explosive growth of the computing industry, and some interviews with very important people of the time (and some even still now). Interviews like this are so important, and especially for Zen, the architecture that turned it around for AMD. Thank you Steve and the GN crew!
I am not an AMD fanboy or any other corporation's fan but some credit must be given to the people of AMD.... I am using and building computers since early 2000s and watching the news of the industry as close as I can, I never encounter such an open approach of a company to talk about technical details, open their lab doors and let its engineers, scientists and designers talk directly to media about the products and tell stories of the development and work of the past days, show prototypes, I mean it is so far away from the competitors' approaches that always use PR department people talking marketing bullshit and nothing about the real important events and just showing favourable slide shows and also thanking Steve all the time (that is the positive side)... Well done AMD and thank you for being so open to us and played your head with Ryzen otherwise we would have still paying 600€ for a 4core CPU with hyperthreading at 3 Ghz turbo in 2024.
Now a few year on, if the x86 ISA lost its market competition I estimate ARM and RISC-V are finally in a development position to scoop up a lot of the market slack and provide indirect competition followed by accelerated development focused on the tasks being performed by x86. That wasn't the case in 2015 for sure. Even if they had a powerhouse ARM or RISC-V back then the legacy-software side was still very attached to x86 on the heals of Window's peak dominance (and Windows its self being programmed specifically for Intel x86 hardware). But a lot of software has become much more portable since then as the importance of diversifying for risk managment is considered, along with the increased availability of non-x86 processors.
Pretty sure Jim Keller had a pretty big hand into Ryzen if not the 3D Chiplet approach, I would love to see an interview with him as well. Pretty cool that you got the interview with Mike Clark, guy got to put his name in the history books with Zen which has been a massive success story with incredible scaling.
there's a really good interview that Jim did years ago that's worth watching if you're able to find it. the 3D chiplet approach isn't new but they were the first ones to make it work. it was a concept being thrown around probably 20-25 years ago aimed at trying to find ways to add more cores to processors but wasn't a feasible option with manufacturing tech at the time. if it was a possibility AMD already had all the building blocks to make it work with their hypertransport which is what infinity fabric is based on.
There's also another interview of Jim somewhere, he says that he's there to enable the team to get the job done, and that the idea was from someone in AMD, as this interview points out, they needed to start a blank sheet design, and Jim probably convinced upper management to do it. The 3D zen CCD is a L3 cache die on top of the CPU die using TSV (Through-silicon via), it is not the same technology as the "chiplet" that allows multiple CCD's to communicate with each other and the memory controller on it's own die to be manufactured on a separate process, potentially by a different company, the later is pretty similar to how old CPUs use to have the memory controller in the north-bridge on the motherboard. I would also gamble that Jim had zero influence over the "3D" part of the design, that required TSMC to actually do TSV for general purpose chips, and not really inside AMD's control, I don't think TSMC were talking about layered designs for general purpose chips till about 2020, previously it was all HBM memory.
It becomes so easy to depersonalise everything and label the product with the brand, but we forget there's people behind it putting everything they have into making these products, dedicating many hours of their lives for the end consumer, so I personally appreciate content like this. Brilliant work to all of you at GN.
I'm still rocking the 1700X. Great CPU, decent platform, its come a long way. OG zen could barely hit 4ghz, and ddr4 support was only 3000mhz but it was a solid product for 2017. Soon I'll be upgrading to zen3. Also its insane to think about a CPU platform lasting 8 years! AMD was super generous with their AM4 socket which makes for a fantastic budget minded system. 5800X3D and DDR4... You could easily build a monster machine for $500/$600 that would wipe the floor with the PS5 PRO.
Hah nice to see some folks still have their original ZEN 1 parts... Back in 2016 the tech world started talking about the upcoming Zen, and i said to the idiot intel femboys that, Zen will be future Legend (i did base my prediction on Jim Keller mostly and his back record), and all i got was laughed on, and most said ZEN will be another FAILdozer and stuff... fast forward to 2017 and ZEN launch, everyone was shocked ZEN was actually really good, (tho IPC level was Broadwell at the time), i was clever enough to sell few months early my AM3+ parts with FX-8350, so i jumped a month or 2 to the AM4 bandwagon with Ryzen 1700 (lost the silicone lottery the chip was not even stable at 3.9GHz, and could not post on 4GHz hah) and highest end Aorus x370 K7 (still have this damn board, still runs rock solid) mobo.... Few years later i moved to Ryzen 3700x, then few years later i moved to now Ryzen 5800x3D, great upgrades... Also had with 1700x, Radeon RX580 8GB, that i sold to miners in late 2021 for double the prices, so that i can upgrade in early 2023 to Radeon 6800XT, and using in the mean time Radeon R9 290, that had about the same performance as the RX580... AMD did not want to give us the x370 first gen mobos microcode update, so we can use ZEN 3 parts on the first gen boards, but they changed that, it was annoying that they did that, but it did not affect me... Only downside of using x370 board with ZEN 3 is you lose PCEex 4.0 support, and that results in 5% lower GPU performance, like 10 FPS, not huge deal but still that is performance left on the table...
just a quick question: Why is the intro logo blue? I mean, the channel logo is green, and in the previous videos, the intro logo was also green. My personal preference would be the colors shifted to green instead of blue.
We're going to stick with green on this channel. The reason is pretty boring: I couldn't find the file for the green logo because it'd been so long! We'll get it in for the next one as the right people are at work this week to locate it.
now you know why they don't like letting engineers talk to the media/public, lol. want 95% of the truth, go to the engineer's. want a bologna sandwich, go to the PR team.
This type of video is amazing, seeing the people behind the tech, listening to the challanges and problems tha happened during development, helps also dismistify the products
I love people like this. So unassuming yet a certified genius. I recently met a person who had a lot to do with the JWST project & he was just a normal, older, nerdy type dude. And he was terrifyingly intelligent. You never know who you're talking to.
G'day Steve & Mike C, While they do not perform well enough for the main channel I❤these BTS informal chats, so thanks for getting back to uploading Engineers videos here
Great interview. I'd only read about Mike Clark, but never seen an interview with him. It's nice to hear his thoughts on coming up with Zen and what it meant to the company.
@GamersNexus To be fair, with all the shenanigans around both Intel and AMD around microcode, Windows versions etc, I can understand why GN are waiting for things to stabilise before retesting. Also, even if the Intel microcode does make things more stable for now, will any existing damage to the chips just get worse through normal wear and tear until failure? There's a good argument to not bother with retesting 13th and 14th Gen entirely.
Started using AMD after Cyrix "died" . From the Athlon Thunderbird, to a Sempron, and then a Phenom 9550, Phenom II x4 965, FX 8320, Ryzen 2 2600, Ryzen 3 3300X, Athlon 3000G (serving as my parents' HTPC) , Ryzen 5 5600 (in my 2nd rig now) and then finally a Ryzen 7 5800x3d (jumping to Zen 5 when the X3d parts are out and MS works out the kinks in their rubbish OS). Obviously I am a AMD man for both CPUs and GPUs. When I was a kid, my parents couldn't afford Intel, and AMD (and Cyrix) were there to save the day. Wonderful memories of using these cpus, and am grateful for them both. Now that I could afford the best from AMD, Intel or Nvidia, I would still go full AMD.
Yay, extras relaunch, I expect to see the following... Computer industry meme collections Trail talks exclusive featuring Kingpin Behind the scenes D&D lore rants mid review An excessively in depth discussion of why cats like computers Visionary projections on the computer industry Random Wendell appearances GN Team talks about their video game history or something like that.
Thankyou to ZenDaddy and his hard working but optimistic team for bringing Ryxen to the market. I have 3-4 Zen Cpus in my pc builds at home, just added a 5700X3D to the collection! Thankyou for your service! 😂👍
This was super fun, even in light of Zen 5 not hitting that double-digits that they're known for. It's a good reminder that there's just people behind these products. They're doing what they can, and it seems like their ethos is at least respectable. And hey, maybe every once in a while they don't quite make it, but if they don't, I'm sure they learned a lot and it doesn't sound like they're happy to play it safe. Very cool interview, thanks Steve (and crew).
If you're able to track down the original team that was responsible for the AMD Opteron, some of those guys were originally on the DEC Alpha team, and some of the ideas found its way into said AMD Opteron line. It would be interesting to listen to the stories about the history of Opteron/Athlon, and then how/why Bulldozer/Piledriver didn't do well, which led to Zen.
Wasn't Jim Keller the father of Zen? (BTW, you need to interview him, he was behind the K7/8 architectures [Athlon/Athlon 64] and the first custom Apple Silicon chips [A4/A5] which are all legendary CPUs)
Because gaming is not very important. They always say zen 5 is foundational but that just means they have yet to build the house. So they need to build the foundation and they have extra resources to build the garage but other parts could only be built next generation. Between Data Center and Gaming they have chosen to design for the data center performance first. There are many things in Zen5 that are incomplete for example their dual fetch decode units only work with SMT. This does not help gaming workload but is very good for high throughput applications.
@@kazedcat Right but they advertised it as good for gaming, and it really isn't. Not really better (or sometimes worse) than Zen4. If that was their position they shouldn't have advertised the chips to gamers.