AMD Ryzen 9 3000 is a 16-core Socket AM4 Beast

Discussion in 'HardForum Tech News' started by sknight, May 10, 2019.

  1. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,291
    Joined:
    Jun 13, 2003
    I get this... but immediately question the judgement if meant to apply broadly, i.e. 'Ryzen 5 3600 doesn't deserve to be tested period'. I don't think that that's where Mr. George is coming from, but if it is, he's wrong. If he doesn't mean it broadly but is rather referencing reader interest and / or his own organization's testing resources, then yeah, I get that. I even get it if he's just disappointed and frustrated that AMD hasn't sent out the big boys.
     
  2. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,442
    Joined:
    Mar 4, 2007
    I'm sure he means for his audience, he doesn't see the point of reviewing a chip that isn't better than a top teir of the last gen. I'm sure they will still test it.
     
    Auer and IdiotInCharge like this.
  3. ccityinstaller

    ccityinstaller [H]ardness Supreme

    Messages:
    4,166
    Joined:
    Feb 23, 2007
    What is a bit worrying is that MsI is equipping it's mid-range boards with both an 8 and 4 Pin CPU Aux power connector....So that should be telling. I would think it's safe to say that the upper end boards from everyone are going to to have the same setup at the minimum with most having dual 8 pins.

    A *Few* x470 boards came equipped with dual 8 pins sockets buy they were overpriced flagship bosrds IIRC.
     
  4. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,442
    Joined:
    Mar 4, 2007
    Dual 8 pin? That's over 450w of power potential! Yikes.
     
  5. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,291
    Joined:
    Jun 13, 2003
    It's a good ceiling, I'll give them that. If you're going to run sixteen cores in a socket and overclock them- perhaps it won't be pulling 450w solid, but having the ability to pull it when needed to maintain stability is certainly a welcome feature.
     
  6. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,442
    Joined:
    Mar 4, 2007
    Yeah, all the 400 series itx motherboards look to have a single 8 pin. Will see if the new itx boards improve on that.
     
  7. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    11,291
    Joined:
    Jun 13, 2003
    With ITX... it may not even be a good idea.

    I get the draw of putting all of that performance into the ITX form factor, but man, that power gets turned into heat and has to get exhausted, and well, it's hard to imagine that being done quietly.
     
    DedEmbryonicCe11 likes this.
  8. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,442
    Joined:
    Mar 4, 2007
    Well even a stock 3850x with a tdp probably puts out less heat than a maxed out 2700x. Will wait and see!
     
  9. NWRMidnight

    NWRMidnight Limp Gawd

    Messages:
    318
    Joined:
    Oct 23, 2010
    Hello??? Do you not understand that the Athlon used nearly identical architecture as Intel, This is not the case with Ryzen. The Ryzen is completely different architecture than intel, or any previous AMD processors (we won't count the failure known as Bulldozer). If you want to try and imply that everything is the same today as it was 15 years age when the Athlon was released, have at it! Because it will only strengthen the opinion that you are out of touch with reality.

    You are also full of crap about accusing me of making excuses for AMD's shortcomings, considering that wasn't and isn't part of the discussion. This discussion is about CB15 being an outlier, which it isn't because you are trying to first compare CB15 results with benchmarks that are not testing the same performance metrics, and are still geared towards Intel architecture. You can deny it all you want, but it is fact. Even Windows 10 has issues and bugs still that crop up with Ryzen, which effect benchmark results.. So how in the world can you expect benchmarks to fully support it if the OS they are being ran on are still having issues? No where have I argued that the Ryzen does't have any short comings.. As we are talking about CB15 compared to other benchmarks when it comes to Ryzen. I really don't care if Ryzen wins or loses, but It has to be with non biased testing metrics that properly and fully support the architecture, which is not the case.

    edit: I was mistaken about Athlon Architecture being similar to Intel's, however see post #411.
     
    Last edited: May 26, 2019
  10. Keljian

    Keljian Gawd

    Messages:
    637
    Joined:
    Nov 7, 2006
    No..
    The athlon did not use the same architecture as the PII/PIII competitors
    The athlon 64 did not use the same architecture as the the P4 competitors.

    Yes they ran similar (and in some cases, the same) instruction sets. But no.. AMD and intel's architectures have been different since the P5 era.
     
    riba2233, juanrga and IdiotInCharge like this.
  11. NWRMidnight

    NWRMidnight Limp Gawd

    Messages:
    318
    Joined:
    Oct 23, 2010
    You are correct, My mistake. I realized that shortly after I posted it and I slowed down to rethink what I said. I actually went to change it, but noticed you responded, so I just added an edit at the bottom of that post, and will just put what I was going to put in it's place here:

    Athlon vs Intel, Compared to Ryzen vs Intel, is a completely different ball game today, specially when it comes to software. 15 years ago, AMD's Market Share vs Intel's was about 45% vs 55% when the Athlon was released, and about 49% to 51% in 2006. What this means is that software developers split their resources between Both architectures nearly equally. We don't have that today. In fact Intel had 82% of the market share vs AMD's 18% Market share in 2017. It is about 79% to 21% today (estimated from current data).. Where is the focus of software developers going to be? where are they going to place most of their resources when making sure that they have all the proper code (instruction sets, etc) in place, towards the 79% or the 21% (82% or 18% 2 years ago)?
     
    Last edited: May 26, 2019
    IdiotInCharge and Darth Kyrie like this.
  12. Keljian

    Keljian Gawd

    Messages:
    637
    Joined:
    Nov 7, 2006
    Passmark (which imo should be install base rather than market share) often lags far behind actual market share... https://wccftech.com/amd-ryzen-intel-core-cpu-market-share-price-report-september-2018/

    You have sales which is market share, then you have install base which is how many people have intel/amd processors, that they may have bought years ago. Good example case of install base is my wife, who is on an intel mac from 2010 at home, which still serves her so well that she's not likely to upgrade in the near future..

    Re architectures...

    I've put some serious thought into this over the 30 odd years I've been around computers.

    Essentially AMD's offerings give you "more" than intel (Bulldozer/piledriver etc not included). In Ryzen, this means more execution units, which is basically the same technological advantage that the original athlon had in 1999 over the Pentium III.

    It can do more, in the same number of clock cycles as the intel equivalent

    What it can't do is make up for the lack of throughput in things like vector instructions (AVX2). This, the cache hierarchy and the CCX bus (infiinity fabric?) are limitations that Ryzen has. For all of its issues, it has a few things that make up for that, namely pure throughput per cycle, better SMT, better power management.

    The way AMD look like they're going to make up for it is by giving you more per socket. More cores, more AVX units, and therefore more processing power. AVX code is typically multithreaded. Doubling the cores, means doubling these units. This may be where we see AMD taking over the content creation crowd when Zen 2 is released. If AMD's units are 20-30% slower right now (which is a decent approximation for Real world workflows) doubling the units/cores should get them to 50% faster than intel in AVX2, taking into account bus/bandwidth limitations etc.
     
    Last edited: May 26, 2019
  13. Balkroth

    Balkroth Limp Gawd

    Messages:
    226
    Joined:
    Dec 15, 2011
    As a dork here and 90% of my stuff is with Service providers, this is gigantic for sdwan stuff, And they're all taking notice.
    Besides cisco has their own deployment, and intel does for uCPE stuff, BIG sp are looking for better open box solutions, but AMD needs someone to come up with their own, or do it themselves, as that is what they are fighting now.

    -Edit Cisco and Intel both have their own sdwan deployments
    -Edit 2 - Cisco has CVIM
    Edit 3 and edge compute- hence uCPE
     
    Last edited: May 26, 2019
  14. cageymaru

    cageymaru [H]ard as it Gets

    Messages:
    19,803
    Joined:
    Apr 10, 2003
  15. mnewxcv

    mnewxcv [H]ardness Supreme

    Messages:
    6,442
    Joined:
    Mar 4, 2007
    Where is the actual announcement taking place?
     
  16. Derangel

    Derangel [H]ard as it Gets

    Messages:
    17,613
    Joined:
    Jan 31, 2008
    According to a video Gamersnexus posted recently there will be info both later tonight at the Computex AMD keynote and then more into in a few weeks at E3.
     
  17. NWRMidnight

    NWRMidnight Limp Gawd

    Messages:
    318
    Joined:
    Oct 23, 2010
    The term Market share can be used for sales or install base. It all just depends on which metric/results you want. But You are correct, it should be market share for sales, and install base for actual in use, to make it less confusing. Obviously, I was referring to install base, which is the metric software developers focus their resources on, as that is where the majority of their income is going to come from and what is going to pay the bills. If they used Sales Market share which is a very short term metric, they would end up dumping resources into failures that could and are used else where. Basically, they are going to be quicker to support the majority than the minority, they are going to focus on where they will make the most income. It's all business 101. Specially when it comes to new architecture.. they are not going to dump a bunch of resources behind something till it passes the test of time and proves itself. This is why Microsoft is still stumbling with Ryzen support, as they are seeing that it's architecture is maturing and it is starting to pass the time test, and now they are scrambling to fix all the bugs to properly support it. In other words, it is now becoming a priority, as each generation of the Ryzen, is only getting better and better.. something Bulldozer never had going for them as it was a failure and never a priority for any software developer. (note I never said a word about being better or faster than Intel, so those foaming at the mouth, relax).
     
    Last edited: May 26, 2019
    Darth Kyrie likes this.
  18. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,003
    Joined:
    Apr 22, 2006
    More info at E3 kind of lends weight to Computex being Ryzen details, and E3, being Navi details.
     
  19. Derangel

    Derangel [H]ard as it Gets

    Messages:
    17,613
    Joined:
    Jan 31, 2008
    Once they get an article version of the video up on their site I can find the exact quote but they said more Ryzen stuff would be revealed at E3 as well.
     
  20. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    9,003
    Joined:
    Apr 22, 2006
    Since they are at computex, you might wait a while. I just watched.

    He is really saying it doesn't look like specs at Computex, with E3/June 10 being the real specification and product stack reveal. July 1st Pre-order, and July 7th on sale and review embargo lift.
     
  21. TheRookie

    TheRookie Limp Gawd

    Messages:
    152
    Joined:
    May 21, 2019
  22. RobCalleg

    RobCalleg Limp Gawd

    Messages:
    129
    Joined:
    Nov 15, 2018
    Hard to understand what Gamers Nexus was trying to say but GN says no specifics today, have to wait til June 10th.

    They are focusing on X570 he says. Maybe one or two processors only today I think he said but not for sure.

    I’m away from home so I can’t link
     
  23. Derangel

    Derangel [H]ard as it Gets

    Messages:
    17,613
    Joined:
    Jan 31, 2008
    Yeah. Sounds like E3 will be the big info dump on the CPUs. It kind of makes sense to have it there if pre-orders are opening up in July.
     
  24. cageymaru

    cageymaru [H]ard as it Gets

    Messages:
    19,803
    Joined:
    Apr 10, 2003
    TSMC's 7nm+ EUV is in production, improves performance by 10%.
    https://www.techspot.com/news/80237-tsmc-7nm-production-improves-performance-10.html

    TSMC already starts N7+ volume production, says CEO.
    https://www.digitimes.com/news/a20190523PD209.html

    Looks like TSMC and AMD are ready to move on from 7nm for the fourth generation of Ryzen.


    N7+ has identical yield rates to N7 and will steadily improve, while also offering a 20% increase to transistor density. There’s also a 10% performance uplift or 15% power efficiency increase. AMD will take advantage of the former in their fourth-gen Ryzen which they’ve confirmed to use TSMC’s 7nm+, while Huawei will most likely take advantage of the latter in their mobile SoC flagships due out in devices at the end of the year.

    While this is all pretty good news for the consumer, it also means that Intel’s got some challenges ahead. They’ve long touted that their 10nm process is equivalent to TSMC’s 7nm, but what about 7nm+? And while some 10nm products will appear in laptops later this year, key market segments, like desktops, high-performance laptops, and server parts won’t receive 10nm hardware for up to three years.
     
    ManofGod, KazeoHin and Darth Kyrie like this.
  25. cageymaru

    cageymaru [H]ard as it Gets

    Messages:
    19,803
    Joined:
    Apr 10, 2003
Tags: