TechTechPotato
TechTechPotato
  • 318
  • 8 618 424
EPYC for Desktop: It's finally here! (and cheap too)
Ever wanted a cheap entry into the world of EPYC? AMD is announcing its first DDR5 entry-level processors, based on the consumer AM5 platform but with support for BMC, DDR5, RAS features and all that good stuff. The entry level silicon is AMD's cheapest EPYC CPU to date.
[00:00] Intro
[02:36] Introducing EPYC 4004
[04:04] EPYC 4004 specifications
[05:50] EPYC 4004 line-up
[07:15] EPYC 4004 vs. the competition
[08:07] Vendor partners
[09:48] Socket and platform
[11:39] Retail when?
Need POTATO merch? There's a chip for that!
merch.techtechpotato.com
more-moore.com : Sign up to the More Than Moore Newsletter
www.patreon.com/TechTechPotato : Patreon gets you access to the TTP Discord server!
Follow Ian on Twitter at IanCutress
Follow TechTechPotato on Twitter at TechTechPotato
If you're in the market for something from Amazon, please use the following links. TTP may receive a commission if you purchase anything through these links.
Amazon USA : geni.us/AmazonUS-TTP
Amazon UK : geni.us/AmazonUK-TTP
Amazon CAN : geni.us/AmazonCAN-TTP
Amazon GER : geni.us/AmazonDE-TTP
Amazon Other : geni.us/TTPAmazonOther
Ending music: ua-cam.com/video/2N0tmgau5E4/v-deo.html
-----------------------
Welcome to the TechTechPotato (c) Dr. Ian Cutress
Ramblings about things related to Technology from an analyst for More Than Moore
#amd #epyc #ecc
------------
More Than Moore, as with other research and analyst firms, provides or has provided paid research, analysis, advising, or consulting to many high-tech companies in the industry, which may include advertising on TTP. The companies that fall under this banner include AMD, Armari, Baidu, Facebook, IBM, Infineon, Intel, Lattice Semi, Linode, MediaTek, NordPass, ProteanTecs, Qualcomm, SiFive, Supermicro, Tenstorrent, TSMC.
Переглядів: 60 740

Відео

[7] Was Sundar Correct at Google I/O?
Переглядів 5 тис.28 днів тому
We've been foiled! Google decided to count up the number of times they said AI in the 2024 Google IO keynote and mention it on stage. Are they correct? We did our own analysis. Need POTATO merch? There's a chip for that! merch.techtechpotato.com more-moore.com : Sign up to the More Than Moore Newsletter www.patreon.com/TechTechPotato : Patreon gets you access to the TTP Discord server! Follow I...
The Magic of Transistors: TSMC's Path to A16!
Переглядів 80 тис.Місяць тому
TSMC is celebrating 30 years of the North America Tech Symposium, and this year they have a ton of announcements. Headlining is A16, the process node beyond TSMC N2, featuring Super Power Rail technology. We also have new packaging in System-on-Wafer, enabling wafer scale like designs similar to Tesla Dojo, and there's also COUPE, TSMC's new optical co-packaging technology. At the Symposium, th...
$40m Goverment Supercomputer, Sold for $480k ??
Переглядів 15 тис.Місяць тому
I have the need... for performance! The question is, do I get a supercomputer, or a few GPUs? Here is the auction: gsaauctions.gov/auctions/preview/282996 [0:00] Lotsa performance [0:46] The Auction, The Specifications [2:35] E5-2697v4 [3:55] No Shipping! [4:40] Can you recycle and make a profit? [6:10] How to get a top supercomputer [8:00] Photos of the system, and leaky This system is still #...
Intel Vision 2024 Keynote: But it's all AI
Переглядів 6 тис.Місяць тому
Need POTATO merch? There's a chip for that! merch.techtechpotato.com more-moore.com : Sign up to the More Than Moore Newsletter www.patreon.com/TechTechPotato : Patreon gets you access to the TTP Discord server! Follow Ian on Twitter at IanCutress Follow TechTechPotato on Twitter at TechTechPotato If you're in the market for something from Amazon, please use the followin...
Intel's Newest $350 Million Machine
Переглядів 189 тис.Місяць тому
In order to print the smallest structures in silicon, a massive ASML Twinscan machine is needed. Current generation technology can create line widths of 13nm, and those machines cost $150m. The new generation can go down to 8nm, and cost a massive $350m! In this video, we tour Intel's Oregon Fab where they've completed installing the world's first commercial High-NA EUV machine. It's in the cal...
Google Cloud Next 2024 Megacut: But It's All AI
Переглядів 4 тис.Місяць тому
The latest in our megacut series is from Google Cloud Next 2024, featuring Google Cloud CEO Thomas Kurian and Google/Alphabet CEO Sundar Pichai. It's all about AI. Need POTATO merch? There's a chip for that! merch.techtechpotato.com more-moore.com : Sign up to the More Than Moore Newsletter www.patreon.com/TechTechPotato : Patreon gets you access to the TTP Discord server! Follow Ian on Twitter...
My First Time at a Chip Design Conference: SNUG
Переглядів 39 тис.2 місяці тому
Join Ian as he recaps his Synopsys user group (SNUG) experience, including what SNUG is, what/who Synopsys are, what they showed off, and how it is going to change EDA and simulation going forward. Machine Learning and Artificial intelligence is only going to play a larger part in chip deisgn as designs get more complex - the issues with thermal, electrical and magnetic interferences are going ...
Making AI More Accurate: Microscaling on NVIDIA Blackwell
Переглядів 15 тис.2 місяці тому
The focus on AI hardware has been speed and power efficiency. One of those tools is quantization - using fewer bits and doing more calculations. It usually comes at a cost - accuracy. Accuracy is important in AI. In order to get the best of both, NVIDIA has integrated microscaling - fixed offsets for small FP4 batches of numbers. Here's an explainer! Need POTATO merch? There's a chip for that! ...
2024 GTC NVIDIA Keynote: Except it's all AI
Переглядів 6 тис.2 місяці тому
Does the biggest AI company in the world say AI more often than other AI companies? Let's find out. AI AI AI AI AI AIAI AI AI AI AI AIAI AI AI AI AI AI AI AI AIAI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI AI Need POTATO ...
[35] Ian Interviews Mark Wade, CEO Ayar Labs
Переглядів 8 тис.2 місяці тому
One of the limits of how big AI scales in the future is how fast the hardware can talk to each other. Ayar Labs, a silicon valley based startup, has an in-package optical interconnect solution, involving optics-in-silicon for faster chip-to-chip connectivity. I spoke with CEO Mark Wade about the technology, the roll-out, and what it all means! www.ayarlabs.com [00:00] Intro [00:45] Who or what ...
The Biggest Human Calculation in 100+ Years (ft @standupmaths )
Переглядів 6 тис.2 місяці тому
It's official. I've eaten too much silicon. I had to go technology-free for a week to get it out of my system. This is the story of my rehab - how I participated in the biggest human-only calculation in 100 years, organised by @standupmaths. Watch the @standupmaths video here! ua-cam.com/video/LIg-6glbLkU/v-deo.html Many thanks to them for letting me use clips from the video :) It's not every d...
4,000,000,000,000 Transistors, One Giant Chip (Cerebras WSE-3)
Переглядів 120 тис.3 місяці тому
The only company with a chip as big as your head, Cerebras has a unique value proposition when it comes to AI silicon. Today they are announcing their third generation Wafer Scale Engine, called WSE-3. Built on 5nm, this chip increases the cores to over 900,000, has four trillion transistors, and doubles training performance over WSE-2. Each system costs a few million, but the price hasn't gone...
[34] Ian Interviews Aart de Geus, Synopsys
Переглядів 25 тис.3 місяці тому
[34] Ian Interviews Aart de Geus, Synopsys
Customized and Bespoke Hardware, but for AI
Переглядів 10 тис.3 місяці тому
Customized and Bespoke Hardware, but for AI
NVIDIA GTC 2024 Preview (Win an RTX 4090?)
Переглядів 10 тис.3 місяці тому
NVIDIA GTC 2024 Preview (Win an RTX 4090?)
[33] Ian Interviews: Renee James, Ampere Computing CEO
Переглядів 26 тис.3 місяці тому
[33] Ian Interviews: Renee James, Ampere Computing CEO
[32] Ian Interviews: Pat Gelsinger, Intel CEO
Переглядів 52 тис.3 місяці тому
[32] Ian Interviews: Pat Gelsinger, Intel CEO
Podcast 12 ~ Rebellions, Intel, Ceremorphic, Tenstorrent, NVIDIA, AMD ~ The AI Hardware Show
Переглядів 5 тис.4 місяці тому
Podcast 12 ~ Rebellions, Intel, Ceremorphic, Tenstorrent, NVIDIA, AMD ~ The AI Hardware Show
Ep 12 ~ Rebellions, Intel, Ceremorphic, Tenstorrent, NVIDIA, AMD ~ The AI Hardware Show
Переглядів 29 тис.4 місяці тому
Ep 12 ~ Rebellions, Intel, Ceremorphic, Tenstorrent, NVIDIA, AMD ~ The AI Hardware Show
Podcast 11 ~ Tenstorrent, AWS, Kalray, FlexLogix, AMD, Ambarella ~ The AI Hardware Show
Переглядів 3 тис.4 місяці тому
Podcast 11 ~ Tenstorrent, AWS, Kalray, FlexLogix, AMD, Ambarella ~ The AI Hardware Show
Unboxing the Tenstorrent Grayskull AI Accelerator!
Переглядів 47 тис.4 місяці тому
Unboxing the Tenstorrent Grayskull AI Accelerator!
Ep 11 ~ Tenstorrent, Amazon, Kalray, FlexLogix, AMD, Ambarella ~ The AI Hardware Show
Переглядів 22 тис.4 місяці тому
Ep 11 ~ Tenstorrent, Amazon, Kalray, FlexLogix, AMD, Ambarella ~ The AI Hardware Show
Podcast 10 ~ Synsense, Intel, NVIDIA, Quadric, Esperanto ~ The AI Hardware Show
Переглядів 2,8 тис.4 місяці тому
Podcast 10 ~ Synsense, Intel, NVIDIA, Quadric, Esperanto ~ The AI Hardware Show
Ep 10 ~ Synsense, Intel, NVIDIA, Quadric, Esperanto ~ The AI Hardware Show
Переглядів 18 тис.4 місяці тому
Ep 10 ~ Synsense, Intel, NVIDIA, Quadric, Esperanto ~ The AI Hardware Show
Podcast 9 ~ Intel, Xilinx, Qualcomm, Encharge, Alibaba, Furiosa ~ The AI Hardware Show
Переглядів 3,6 тис.4 місяці тому
Podcast 9 ~ Intel, Xilinx, Qualcomm, Encharge, Alibaba, Furiosa ~ The AI Hardware Show
Ep 9 ~ Intel, Xilinx, Qualcomm, Encharge, Alibaba, Furiosa ~ The AI Hardware Show
Переглядів 31 тис.5 місяців тому
Ep 9 ~ Intel, Xilinx, Qualcomm, Encharge, Alibaba, Furiosa ~ The AI Hardware Show
Seven Silicon Highlights from SC23. And a Llama
Переглядів 29 тис.5 місяців тому
Seven Silicon Highlights from SC23. And a Llama
How fast can you transfer data between two GPUs?
Переглядів 11 тис.5 місяців тому
How fast can you transfer data between two GPUs?
AMD in Servers: CPU, V-Cache, and GPU 💻 #sc23
Переглядів 12 тис.5 місяців тому
AMD in Servers: CPU, V-Cache, and GPU 💻 #sc23

КОМЕНТАРІ

  • @randar1969
    @randar1969 Годину тому

    Correction on the title , Intel buy $350 million chip making machine from ASML. Now i seems like they made it. ASML HQ isn't even in the states although admitted it does contain american patented parts and Intel does own ASML stock like TSMC and samsung.

  • @najeebshah.
    @najeebshah. 17 годин тому

    You all are a bunch of drama queens honestly 🤦‍♂️ especially GN and how holier than thou they act, mfrs used LTT name to boost their views and didn't even do a apology video when the investigation proved madison was acting maliciously

  • @sepolopez6706
    @sepolopez6706 21 годину тому

    Excellent purchase from Entel. This machine is incredible and Intel's progress forced TSM and Samdung to also place orders with ASML so as not to be left behind.

  • @raphtal
    @raphtal День тому

    I went to elementary school and middle school with this women. She is kind as she is smart. I love watching her talk.

  • @tomilan6001
    @tomilan6001 2 дні тому

    are you an ee

  • @holyknighthodrick5223
    @holyknighthodrick5223 4 дні тому

    How does LLVMpipe perform?

  • @digitalizeddeath
    @digitalizeddeath 4 дні тому

    AMD last entered this realm back in the Piledriver days with the AM3+ Opterons. I have an Opteron 3365 8 Core Chip. They offer much in the same as with being aimed at the low end server area. Nice to see them coming down this way again. Im definitely upgrady as my Opteron is a bit long in the tooth and even though ill be going from an 8 core to a 4 core 8 thread. The performance will be so much better.

  • @corneliusantonius3108
    @corneliusantonius3108 6 днів тому

    ASML Velhoven The Netherlands !

  • @cheezyuser
    @cheezyuser 6 днів тому

    That subtitle mis-spelling had me thinking, thats a whole lot of energy for Iraq...

  • @dervolker
    @dervolker 7 днів тому

    as an SMB this is interesting. do we know about system integrator going forward with this already? (not yet finished viewing, maybe it‘s in there after the 50% mark :) )

  • @RonHeusdens
    @RonHeusdens 8 днів тому

    i bet that was an evening editing well spent🤣

  • @Mr11ESSE111
    @Mr11ESSE111 9 днів тому

    poor Intel ,they milks 20 billion from government for factories of chips and this here and tgey still behind of AMD

  • @Mr11ESSE111
    @Mr11ESSE111 9 днів тому

    Its money from government and Intel still behind of AMD and now in month comes zen5

  • @Bladernner2049
    @Bladernner2049 10 днів тому

    Buy intel

  • @Maximilian-sd9xy
    @Maximilian-sd9xy 10 днів тому

    Hello, can you stack them to a cluster?

  • @VideogamesAsArt
    @VideogamesAsArt 12 днів тому

    I love your die-shots background!!

  • @RicoElectrico
    @RicoElectrico 12 днів тому

    Meanwhile me designing IP on N28 ;_;

  • @parkerbond9400
    @parkerbond9400 12 днів тому

    6 lasers coming together... Sounds like the Death Star...

  • @ConsistentlyAwkward
    @ConsistentlyAwkward 14 днів тому

    They are going to showcase 16 Tb/s interconnect this year so maybe a follow up interview?

  • @user-rq4el5nc6v
    @user-rq4el5nc6v 15 днів тому

    Can this be used in audio applications like clocks with a 10m input??

  • @hozachief
    @hozachief 15 днів тому

    Im new to the channel. Have you made a video comparing a Dojo chip to the latest Nvidia chip or Dojo wafer to Nvidia GPU?

  • @barryrose9452
    @barryrose9452 16 днів тому

    Too much me and not enough “we”

  • @ConsistentlyAwkward
    @ConsistentlyAwkward 16 днів тому

    Ibm is so interesting how does this chip relate to the NorthPole chip they announced last year? ik that chip is an inference only ASIC is that research related to this chip or are they completely separate? will they have a training only chip next 😅

  • @ConsistentlyAwkward
    @ConsistentlyAwkward 17 днів тому

    So i obviously don’t have a good understanding of this but how are they getting an advantage with using general purpose cores? where are they making up the difference in parallelism is it just easier for software?

  • @powertreadssupremacy
    @powertreadssupremacy 17 днів тому

    A mouse reviewer is incompetent and therefore unqualified if they check mouse feet for protective coverings and then criticising the feet for abnormally high friction without having the experience anf common sense to look at or touch the mouse feet. A gpu cooler reviewer is unqualified if they knowingly and solely test it on a gpu that isn't compatible and comes to a conclusion based on that review. An investigative journalist is incompetent and therefore unqualified if they subversively manipulate their audience, or at the very least are a propagandizing journalist. Qualification =/= a piece of paper.

  • @hill2hell
    @hill2hell 19 днів тому

    The catch is the CPU is "affordable" but in order to be able to activate and utilize all the EPYC features you need $500+ for a mobo and it'll have a short list of ram that can fully utilize the speed of the CPU which would be at the higher end side which would most likely be around $200+ per kit. But yeah, atleast you saved some money on the CPU right? 😂😂😂

  • @miinyoo
    @miinyoo 19 днів тому

    Anastasia and Linus both jabbed in one video with a wacky camera. The black mirror was real! How does it end? What comes next?

  • @falman-5653
    @falman-5653 20 днів тому

    Me me me me. They need better a ceo. Has no ambition.

  • @CataclysmZA
    @CataclysmZA 20 днів тому

    As a consumer, I'd love it if my next system properly supported ECC and was super stable, but had more affordable pricing.

  • @nextcurve
    @nextcurve 20 днів тому

    Yeah, bumped into Ayar Labs last year at SEMICON West. Cool tech, cool founder and CEO.

  • @user-sx6nd8zl5k
    @user-sx6nd8zl5k 21 день тому

    They left aside the avx512 instructions that are very necessary for the AI ​​due to the stupidity of the E-cores, that is why generation 13 and 14 of Intel suck, and the worst will come with generation 15 of Intel that will no longer have Hyper -Threading, one step forward and two steps back in chip engineering .x86 is dead, it died along with Moore's law

  • @viktoreisfeld9470
    @viktoreisfeld9470 21 день тому

    Great job on this video! It's quite the skill you have to pull key insights from people during interviews. I learned more about where AI hardware is going from those two interviews than I did from reading hundreds or articles, videos, and talks/presentations.

  • @seansingh4421
    @seansingh4421 21 день тому

    So Angela White in 16K Res ? 😅😅😅

  • @ShiroKuroh
    @ShiroKuroh 21 день тому

    Glad to see they're adding to something between the HEDT and Home PC Market, but dollar to dollar, how would this compete vs the old Epyc 7551P with 7 slots? I get a new feature set. higher ipc on boost clock with lower power, but the fact is how many PCIE lanes available, and no motherboard for this set I look at has 7 usable slots, even if 4 were only x4? I don't feel the consumer home market is being as well met outside of saving some heat and power out of the wall would take over 3-4 years to make up the price difference, and that doesn't even account for ram costs. Is the higher IO worth it when we can't build our NAS's, video rendering boxes, etc due to lack of 7 physical slots on most boards available.

  • @galileo_rs
    @galileo_rs 21 день тому

    28 lanes makes this mostly useless in comparison to real Epyc.

  • @DavidSiebert
    @DavidSiebert 21 день тому

    I really wish they would produce and SOC of this with a BGA. If they could just make it in the US that would be great. Now if they would sell just the CCUs and allow a custom IOcomplex that would be even better. I have a project that could really use this in an SOC or even better the ccus. Or allow outside folks to add they own chipletts and packaging.

  • @MT01123
    @MT01123 22 дні тому

    No Linux support. Only supports Windows 10 && 11 so just another useless AMD release.

  • @jouwklappik
    @jouwklappik 22 дні тому

    And Thanks to VDL ETG Eindhoven, for making the Vessel and OPframe. 😎

  • @roinois
    @roinois 22 дні тому

    25:25 aged real bad. Hope this is a learning experience for you.

    • @phillycheesetake
      @phillycheesetake 22 дні тому

      The tough guy routine doesn't really work for him does it? Lol.

  • @patrickprafke4894
    @patrickprafke4894 22 дні тому

    So what Intel did with Xeons. This is awesome! I'm in!!!

  • @Alpine_flo92002
    @Alpine_flo92002 22 дні тому

    Sad thing is its still basically missing all parts that most people would need of a cheap CPU like that

  • @WantMore-mc8dx
    @WantMore-mc8dx 22 дні тому

    A new HP micro server? :)

  • @Im_Nosferatu
    @Im_Nosferatu 22 дні тому

    In a couple thousand years we'll be laughing at how these machines used to take up a whole building

  • @bits2646
    @bits2646 22 дні тому

    I was kinda missing AMD equivalent of Intels Xeon E/D/W series, really good deal that AMD is going now to SMB and entry products

  • @golfinglibrarian
    @golfinglibrarian 22 дні тому

    But how well do they run RandomX? ;)

  • @le_potate3861
    @le_potate3861 23 дні тому

    I don't get this launch. There's literally zero reason to buy EPYC 4004 over Ryzen 7000. What a stupid product.

  • @nagyandras8857
    @nagyandras8857 23 дні тому

    Red flag.. only 2 memory channels and only 192 gbyte max ram. Not gona work. Needs 4 channels. Needs 512 GB ram support. Needs more Than 28 pcie lanes.