10 ms·
Does someone have an ELI5 on what AMD has done since ~2015 to turn their whole operation around like this, and why these new processors are so good? What's the
by Munksgaard 6y ago
Does someone have an ELI5 on what AMD has done since ~2015 to turn their whole operation around like this, and why these new processors are so good? What's the catch?
- speeder 6y agoAMD hired the genius CPU designer Jim Keller, that is it. His previous stint on AMD was when he invented AMD64.
- lorenzhs 6y agoI don't know much about Jim Keller, but a CPU architecture is far too complex for a single person to possible be responsible for such leaps. You can't just take an average team, add one exceptional person, and get results this good. I'm sure he's exceptionally good, but focusing on Keller like that seems rather unfair to all the other engineers who made this possible.
- gameswithgo 6y agoJim has had such frequent success that I think he may have the rare talent of running large technical teams well. I suspect that may be what his talent is rather than CPU design. But maybe both, or maybe he has just gotten lucky. I think the 2 key high level choices AMD made was betting on TSMC, and betting on the chiplet + memory controller design.
- twojacobtwo 6y agoI don't know enough about CPU archtecture to chime in on that aspect, but I can say from experience that a single person with enough weight/respect behind his/her name can certainly get a stuck team moving in the right direction. I don't know what AMD's situation was at the time, but it's possible that no one felt they had the pull to change directions and his fame allows him come in and change the whole dynamic. Just a thought.
- JTon 6y agoHow much do you think a one-of-a-kind designer like Jim Keller makes? I'm looking at his wikipedia page, and he's worked for AMD, Apple, Intel, Telsa. They all want him. I can only assume there's a bidding war on his talent.
- mamon 6y agoMy totaly uneducated guess is: $2M base salary, and $20M in stock options per year.
- dom96 6y agoMy question is: how do you get into this line of work? Does anyone know how Jim Keller got to be such a one-of-a-kind chipset designer?
- baybal2 6y agoWho gets to be a chipset designer, even in times when JK started his career? Just very few people have an opportunity to try their hand at this, and JK was blessed with being able to go from one good team, to another, and mostly isolated from corporate bs. A big part of his success story, was him having this great opportunity to develop his skills. Microelectronics is a career suicide in comparison to software today. I'd say 100 to 1 sieve off rate for people getting into the industry, to ones who actually get to do serious architecture work would be too small. The ratio is likely to be much larger.
- kllrnohj 6y agoKeller left in 2015. It's quite unlikely he had any impact on Zen 3's substantial IPC gains over Zen 2.
- rozab 6y agoAlso worth noting that Keller was at Intel between 2018 and 2020.
- mobilio 6y agoand leaving Intel for "personal reasons"... We only can guess what they can be?
- cbozeman 6y agoHis child has leukemia. I suspect he would have left no matter what though, because Intel is so arrogant they wouldn't listen to him in the manner in which they would to. That was the great thing about AMD; when you're already at rock bottom, you're humbled enough to listen to people.
- kmonsen 6y agoI think this is an important reason, being down and hungry can do wonders for focus and quality. On the other hand doing well over many years means that wrong people got promoted and stay around and grow a big middle management layer that is invested mostly in themselves. AMD either did well or went broke which forces management to take painful but important decisions.
- whereistimbo 6y ago> His child has leukemia. source?
- gameswithgo 6y agoPretty sure the stuff he worked on out extended out to zen4
- jayflux 6y agoI don’t think there’s any catch to it. AMD started again, went with a clean-sheet design, re-hired Jim Keller, then started under-cutting Intel with more cores and decent performance. It went from there really. Unlike Intel who have micro improvements, each generation has been a big leap in performance so it’s been exciting for consumers. Intel did something similar with their core2duos when AMD was on top around 12-13 years ago.
- forest_dweller 6y agoI still use a Core 2 Quad in a desktop I built in 2008 in my office (mainly because it has a proper floppy drive on it). Those processors were such a leap over the Athlon XP 3200+ I had before and it runs a Linux desktop perfectly fine.
- bicolao 6y ago> mainly because it has a proper floppy drive on it What are you using it for? Floppy disks were already very dead by 2008, let alone 2020.
- nitrogen 6y agoNot the OP, but some industrial equipment and musical instruments have floppy drives in them, and it may be cheaper and easier to keep an old desktop running than to retrofit the equipment. E.g. I have an old synthesizer I like that stores patches on floppy.
- forest_dweller 6y agoI have a few old retro computers and a usb floppy drive doesn’t cut it for Amiga and Atari floppy drives (some say it works online but I’ve tried several usb floppy drives and IME it does not). I know there are alternatives to using a floppy to transfer to these ancient systems but for me it works okay.
- gtm1260 6y agoProbably the single biggest reason is the fact that they use TSMC to fab all their chips, and TSMC has had much more consistent node updates compared to intel. Intel's next generation cpu node was supposed to come out this year, and it's been pushed back (possibly due to covid) until the end of 2021 or even into early 2022. Meanwhile, AMD says they have 5nm around the corner.
- kllrnohj 6y agoDr. Lisa Su took over as CEO in 2014 and seems to have driven a re-focusing at AMD, devoting the company's relatively few R&D resources effectively. Starting with large growth in non-PC markets, actually, mostly the consoles. Jim Keller was certainly key in getting Zen up & going, but he left in 2015 so his impact on Zen 3 would be rather minimal to say the least. Combine that with the timing of Intel struggling hard with new process nodes and there you go. A bit of AMD got better, a bit of Intel got worse, and now you've got this landslide result.
- DCKing 6y agoYeah, it's worth noting that Intel's planned microarchitecture iteration on the desktop was delayed by over three years - from 2017 to 2021 - and in 2021 will be released on a larger process than originally planned. AMD's execution looks meticulous by comparison, and they deserve all the praise and profits resulting from it. But their win is inflated by Intel completely imploding. Until next year they're merely iterating on 2015 technology - their Skylake architecture - on the desktop and in servers at least. This is not to say that Intel's 2021 chips will be competitive with this (they won't be) but if Intel with better management would have been at this point in, say, 2018 instead of 2021 this would have played out very differently.
- kllrnohj 6y agoThe 10nm delay is of course legendary at this point, but 14nm was itself also about a year behind schedule. In early 2013 Intel stated Broadwell 14nm would be out that year. It didn't come out until Q4 2014. And Intel's Fab 42, which was originally scheduled to open in 2013 as their first 14nm fab, only just opened this year. Intel ended up needing to re-fit existing fabs to be 14nm instead. So before they faceplanted with 10nm Intel first stumbled with 14nm. 2 generations in a row of issues.
- DCKing 6y agoProper leadership would have taken this risk seriously, and would not have tied up microarchitecture development so much to process improvement. Basing your strategy so fully on Moore's Law keeping up is a recipe for disaster. Clearly Sunny Cove works on 14nm (that's what they're releasing next year!), and they had all the technology available to release that in 2018 too - maybe with lower clocks and core counts. Seems like a classic textbook case of a large coorporation not understanding how to change their ways, ending up being beaten by a more agile smaller competitor. Oh well, hindsight is 20/20.
- numpad0 6y agoI’ve seen someone mention less than stellar memory latency and bandwidth on Zen?
- jeffbee 6y agoIt's true that the Zen parts have higher main memory latency than their Intel competitors. If this is very important to your workload, you should consider them. I think this is caused by the fact that Intel's memory controller is more tightly integrated, while the AMD memory controller is on the other side of some wires from the cores. AMD compensates with huge cache sizes, but every time they increase the cache size they also make the cache slower, which isn't great.
- snvzz 6y agoThe review goes over latencies and how they changed in Zen3, in fine detail. For any practical purposes, Intel is now behind.
- Iwan-Zotow 6y agoINtel screwed up - that was pretty major win for AMD.
- fulafel 6y agoBoth Apple and AMD now beat Intel CPUs. Intel made mistakes.
- chrisjc 6y agoMaybe not ELI5, but I've been following this channel on Youtube for the last couple of years and it does a really good job of explaining the architectural and manufacturing improvements from AMD over the last several years. I really found the yield explanations really eye opening in particular. https://www.youtube.com/c/adoredtv1/videos https://www.youtube.com/c/adoredtv1/videos
- programmertote 6y agoCan you share the title of the video that explains the about yields (if you remember)? I'm not an electrical or hardware engineer, but I'm very curious about it. Thanks in advance!
- chrisjc 6y agoI apologize in advance, but I'm not 100% sure this is the video i alluded to, but here's my best guess. https://www.youtube.com/watch?v=ReYUJXHqESk https://www.youtube.com/watch?v=ReYUJXHqESk
- ksec 6y ago>Does someone have an ELI5 on what AMD has done since ~2015 Let's just go back to 2014/15. And Look at Intel's Roadmap ( Or Slides ). 2016 - 10nm - CannonLake - First Gen 10nm Process. Industry leading Density. 2017 - 10nm+ - Icelake - New Architecture. 2018 - 10nm++ - TigerLake - Optimisation[1]. 2019 - 7nm - Sapphire Rapids [2] Most people would expect the ETA year to be +1. As 14nm itself was delayed. TigerLake has around 15%+ IPC. That would put it ahead of Zen 3. The Intel 7nm is roughly between TSMC 5nm and 3nm. Had Intel shipped that in 19/20, along with Sapphire Rapid. AMD wouldn't even lead, they could barely compete. So to answer your question. It wasn't AMD did some magic. They have been executing, and shipping their product on schedule. ( Although in tech, that in itself is pretty damn impressive ). All while Intel was.... um..... I dont know what they were doing. 10m fiasco has lots of other knock on effect as well. But I guess this is good enough for a ELI5 post :) [1] This Process - uArch - Optimisation name came in later part. In 2014/15 people were still talking about Tick Tock. [2] Sapphire Rapid was always a Server Specific design on the roadmap. But things changes bit. Along with Intel's plan of AVX512