AMD Announces Ryzen 7 3700X, 3800X and Ryzen 9 3900X

Discussion in 'HardForum Tech News' started by Pieter3dnow, May 27, 2019.

  1. N4CR

    N4CR 2[H]4U

    Messages:
    3,707
    Joined:
    Oct 17, 2011
    Try it at 8Hz... then get back to me :)
     
  2. AlphaQup

    AlphaQup Limp Gawd

    Messages:
    280
    Joined:
    Oct 27, 2014
    I just got a chance to catch the entire keynote late this afternoon at work, and I was simultaneously watching this thread grow, and I just don't understand how some of you can't be excited about new "series" of players entering the game, regardless of where they MAY (we're all still waiting on official benchmarks/reviews) fall on the performance spectrum. Marketing fluff taken with some salt, the numbers look promising.

    If ANYTHING, can you not appreciate the pricing of this stuff? It might not wear the crown but shit, that damn near almost crown costs half of what the current one costs lol.

    I've never even owned an AMD rig and I'm pumped to see what they have coming out.

    Professional note: I've got designers/creative types jonesing for more cores for simulation/rendering workloads, from your more basic corporate video editing/rendering, to full blown foundry/solidification simulation/model slicing software. More cores, faster simulations, less engineering downtime (the most expensive "component" of any system).

    Some of you guys gotta look past what's best for you and look at other workload types... it's like you can't be excited and skeptical (as you should always be) at the same time. Fun fact, you can.
     
  3. schmide

    schmide Limp Gawd

    Messages:
    208
    Joined:
    Jul 22, 2008
    Can I use a rotary woofer? Even then you're going to need gigawatts to reach 182db. Physics and the conservation of energy are a bitch.
     
    Revdarian likes this.
  4. Lakados

    Lakados [H]ard|Gawd

    Messages:
    1,490
    Joined:
    Feb 3, 2014
    At best from this launch I am hoping that Intel is forced to change their Xeon pricing structure. Their security issues while seemingly terrible are all covered by either OS updates, Bios Updates, Firewall Rules. When these CPU’s hit the bench I am not expecting miracles and I think while really good they will be compatible to existing Intel lineups but I will be extremely surprised if AMD steals the show with out the appropriate Intel response. Intel can afford cuts and losses to maintain share where AMD cannot. That said a drop in upgrade for my EPYC servers or my 2700 would be super swell.

    Edit...
    The 2700 is still doing well but since the purchase my workload has drastically changed and it struggles and I need more cores more than individual clocks.
     
  5. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    10,284
    Joined:
    Jun 13, 2003
    I can- but I'm also not making assumptions, as we don't have detailed independent results, which will be lower than AMD's as a fact of marketing, and we don't have Intel's response, which may erode the value argument. Further, unless you are doing things like rendering, the speed bump from more cores almost completely goes unnoticed after six. If you are, well, I'd expect you to already have an eight-core Ryzen, or Xeon or Threadripper with even more cores.

    So on the surface, it's exciting; we just need to see if the final product is as good as AMD is implying. The first Ryzen was good, if you could get it working- most of the issues have been solved and we'll need to see if they remain solved with Zen2 after unhooking all of the cores from the memory controller.

    I've owned plenty (and plenty of AMD and ATi GPUs), and it's difficult to overstate my frustration with them post Core 2 / 8800GTX.
     
    Trimlock and AlphaQup like this.
  6. Nobu

    Nobu 2[H]4U

    Messages:
    3,027
    Joined:
    Jun 7, 2007
    it's often been the case, recently, with AMD, that their marketing numbers are very accurate. They usually don't tell the whole story (for instance, all you get is a cinebench or blender run and maybe some data to go with it), but those benches have been verified as accurate after release. However, it's always good to be skeptical of marketing material, so I'll be patiently waiting for 3rd party benches and reviews as well.
     
    jeffj7 and Chimpee like this.
  7. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    10,284
    Joined:
    Jun 13, 2003
    Well, I honestly accept their hard numbers with respect to individual benchmarks.

    What I question are the relative numbers they represent, to their own parts and to their competitions, as well as their extrapoliations from their numbers.

    Yup :)
     
  8. AlphaQup

    AlphaQup Limp Gawd

    Messages:
    280
    Joined:
    Oct 27, 2014
    Neither am I, l like 3rd party reviews as much as the next guy, which I did say I'm even more excited for. Marketing will never ever paint a product in a bad light, unoptimized workload, etc. We saw best case today, absolutely.

    1st gen was OK, agreed on the memory issues, 2nd gen was MUCH better, now they're touting 3rd gen on paper to be what looks like a solid product. Lets see how it pans out, I'm not one to double down on my previous statements in light of new evidence. If/when new Intel kills it (for an exorbitant price tag... shots fired on my part :p), awesome, I'll sing their praises as well. I'm speaking about what's on the very near horizon, and so far, the 9900ks is all I've heard about lately, and I thought I read about a 6+ month wait time on that? Same old same old on the Xeon front, nothing new there.

    As I said above, engineer downtime is the most expensive component of any system, ROI is king in a corporate setting.

    Professionally, we're a Dell shop with a few variations (management loves the one size fits all model after all where ever possible.... sigh) and yeah, we have a few exceptions in place right now, but I'd like to expand that. If we can get a "standard" marketing/simulation offshoot of our standards for specific workloads for a decent price (we can't with Intel currently) then yeah, Ryzen 9 has me excited professionally. On the high end of things, I'm looking forward to either TR3, or high clocking EPYC for our foundry simulation workloads, those just love cores period. 40+ core type stuff, many different flavors of Ansys, MAGMASOFT, Xprep, etc.

    I can't speak to frustrations with them post Core2 unfortunately.

    Simply put, I'm just excited for new products!
     
    Revdarian, KazeoHin and IdiotInCharge like this.
  9. TheRookie

    TheRookie Limp Gawd

    Messages:
    130
    Joined:
    May 21, 2019
    In other words, you are just speculating.
     
  10. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    10,284
    Joined:
    Jun 13, 2003
    In other words, I can take AMD's numbers, and say 'less'. I didn't speculate as to how much.
     
  11. TheRookie

    TheRookie Limp Gawd

    Messages:
    130
    Joined:
    May 21, 2019
    ...then, how can you know that it is "slower" when you are just speculating?
     
  12. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    10,284
    Joined:
    Jun 13, 2003
    Because marketing.
     
  13. Red Falcon

    Red Falcon [H]ardForum Junkie

    Messages:
    9,871
    Joined:
    May 7, 2007
    My body is ready.
    The Jaguar in my sig is not. :D
     
    Flexion and Darth Kyrie like this.
  14. TheRookie

    TheRookie Limp Gawd

    Messages:
    130
    Joined:
    May 21, 2019
    What about it?
     
    gigaxtreme1 likes this.
  15. NightReaver

    NightReaver n00b

    Messages:
    23
    Joined:
    Apr 20, 2017
    He's saying that since it was the benchmarks that AMD decided to use during a marketing event, it's possible that in real world testing the gains won't be as huge. Aka always take what ANY company claims during ANY event with a healthy dose of salt.

    Hence why in my own opinion, sure the numbers look great, but I'll wait for a good number of tests to be ran before I go reaching for my wallet. I don't have the budget to buy new setups on a whim and hope the event numbers align with real life.

    On the other hand, it can be speculated these chips can be OC'd over the event specs, so it really depends on how optimistic you are.
     
    IdiotInCharge likes this.
  16. TheRookie

    TheRookie Limp Gawd

    Messages:
    130
    Joined:
    May 21, 2019
    He mentioned it as fact, something that he could not possibly know unless he has an review sample.

    That's what I am pointing out.
     
    Revdarian and N4CR like this.
  17. NightReaver

    NightReaver n00b

    Messages:
    23
    Joined:
    Apr 20, 2017
    Idk, to me it comes down to conviction. One thing I've noticed while lurking here is convictions run high. I keep seeing these arguments run in a huge circle, both sides convinced they're right. Instead of agreeing to disagree and waiting for real reviews to come out, people just keep hammering away at the same points. I for one don't trust marketing slides, they're almost always trying to show product in the best light possible. My reaction to them is always "That's really cool, hopefully reviews will back that up".

    I don't see his comments as fact, just strong conviction that marketing is part BS, which is hard to argue against.
     
  18. Snowdog

    Snowdog [H]ardForum Junkie

    Messages:
    8,810
    Joined:
    Apr 22, 2006
    Sure companies choose benchmarks that work in their favor, perpetual AMD favorite is Cinebench, because it is an embarrassingly parallel workload, that responds almost linearly to more more cores, and works exceptionally well with SMT...

    It would be questionable if this was all AMD showed, but they showed a fair depth of benchmarks, including gaming. So it looks like a fairly solid improvement across the board for Zen 2.

    Of course it would be prudent to wait for 3rd party reviews, but this looks like a solid upgrade, and even more of threat to Intel's desktop business.
     
  19. NightReaver

    NightReaver n00b

    Messages:
    23
    Joined:
    Apr 20, 2017
    Oh no doubt. Trust me, I'd love to replace my 1600 with a gleaming new 3700x. Once I see a few OC results (got a water loop just begging for some heat output), I'll be right on that trigger.

    I'm just saying, you can't blame someone for being skeptical or marketing slides. Though this huge of a node jump should be offering some significant changes.
     
    jeffj7 likes this.
  20. notarat

    notarat [H]ard|Gawd

    Messages:
    1,697
    Joined:
    Mar 28, 2010
    I've been pretty lucky with my AMD purchases thus far, as far as overclocking is concerned, but I'm still pre-ordering for a new system build as soon as the 3000 series is available, marketing numbers be damned...

    My 1600X runs at 4.195GHz, all-core, 24/7 at great temps but it'll be the one I replace since the new 3000 series are approaching 4.6GHz with less power draw. It's a no-brainer to upgrade when they're released because I know I will see a major increase in performance, regardless of what the "marketing" numbers are. And I don't have to worry about some patch coming out that'll kill 30% of my performance (because Intel)

    My only question is whether or not to go with ITX X570 or normal ATX X570 with this build...I'm liking the sheer variety of boards being offered and it'll probably take far more time picking the board than to pick which new CPU I get.
     
  21. elite.mafia

    elite.mafia Broke Back [H]

    Messages:
    11,397
    Joined:
    Aug 23, 2004
    I wish my 1700x would do 4.2
    I haven't been able to crack 4ghz. I'm rock solid at 3.95ghz though.
     
  22. Nobu

    Nobu 2[H]4U

    Messages:
    3,027
    Joined:
    Jun 7, 2007
    I gave up on overclocking my 1600. I don't know if it is my mb, bios, memory or cpu, but it's just not stable. I'm pretty sure it's my ram, but for now I don't have time to troubleshoot.
     
  23. notarat

    notarat [H]ard|Gawd

    Messages:
    1,697
    Joined:
    Mar 28, 2010

    When I first built my 1600X system I used Patriot Viper 3400MHz DDR-4. The ram ran 2466MHz because it wasn't even on the compatibility list for the board, and I was lucky the system booted at all, let alone worked... (I'd picked up the Patriot RAM for a different build that fell through, and had it on hand when I built the 1600X so I figured I'd use it to save $$$)

    After several months my main system (4770K) pooped itself so I was forced to use the 1600X. I did a BIOS upgrade on the board in the hope that Asus had finally gotten off their lazy ass and added an option to disable the built-in BlueTooth on the board (They still haven't...)

    After the BIOS Upgrade I loaded optimized settings and rebooted and the Patriot RAM was running at 3200MHz with low latency settings. I decided to try overclocking it and it booted up at 4.4GHz at 1.4v but wasn't stable in benches so I backed off to 4.195 and it's been like that (at 1.37v) ever since.

    Have you checked recently (in the past month) for a BIOS upgrade? I'm wondering...since it's an Aorus board, maybe they added a new BIOS revision to support Ryzen 3000 which may improve RAM compatibility like I found by accident with my CH Hero VI

    I see the latest version is F31 and came out on 2019/05/06
     
  24. Nobu

    Nobu 2[H]4U

    Messages:
    3,027
    Joined:
    Jun 7, 2007
    Not yet. Think I'm on F21 (or something like that) if not an older one on the backup bios thanks to GB's buggy recovery feature on this board. I'll probably stick with it until I get a 3xxx series ryzen. it works, just crashes in games randomly. Will probably "refresh" windows at that point also.
     
  25. Meeho

    Meeho [H]ardness Supreme

    Messages:
    4,303
    Joined:
    Aug 16, 2010
    They only showed gaming improvements compared to Zen+ and not how it does against Intel like they did for Cinebench, which seems suspiciously selective. Reported gains do indicate there will finally be gaming parity with Intel, though, so that gives me hope. Intel will probably keep better OC gains, but that should be much less important compared to current situation.
     
    Araxie likes this.
  26. elite.mafia

    elite.mafia Broke Back [H]

    Messages:
    11,397
    Joined:
    Aug 23, 2004
    It's probably your RAM. Is it on the QVL? A lot of x370 boards don't seem to want to run ddr-3000 but will run stable at 2933
    My 1600 ran about 3.85ghz no problem.
     
  27. OrangeKhrush

    OrangeKhrush [H]ard|Gawd

    Messages:
    1,490
    Joined:
    Dec 15, 2016
    That is according to you
     
  28. FlawleZ

    FlawleZ Gawd

    Messages:
    790
    Joined:
    Oct 20, 2010
    And what kind of significant jumps has Intel made since Q1 2011? I'm not seeing any.

    I suggest you take a look at the law of diminishing returns. That will always have a role in technology. The sooner you can accept that the sooner you can realize just how good these Ryzen 2 chips are looking so far.
     
    N4CR, Darth Kyrie and LightsOut41 like this.
  29. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    10,284
    Joined:
    Jun 13, 2003
    You're going to make the claim that these numbers are not marketing numbers?


    Please be my guest :D
     
  30. OrangeKhrush

    OrangeKhrush [H]ard|Gawd

    Messages:
    1,490
    Joined:
    Dec 15, 2016
    Cinebench and pubg, but yeah it was for marketing the new generation product
     
  31. IdiotInCharge

    IdiotInCharge [H]ardForum Junkie

    Messages:
    10,284
    Joined:
    Jun 13, 2003
    ...from the vendor themselves.
     
    travm likes this.
  32. Meeho

    Meeho [H]ardness Supreme

    Messages:
    4,303
    Joined:
    Aug 16, 2010
    Me neither.
    Which is it, diminishing returns or just that good? You can't have it both ways.
     
  33. Rockenrooster

    Rockenrooster Limp Gawd

    Messages:
    232
    Joined:
    Apr 11, 2017
    Bruh............ dat latency...

    AMD-Ryzen-3000-AIDA64-Memory-Benchmark-1030x1002.jpg
     
    Revdarian, alxlwson, Verado and 9 others like this.
  34. Rockenrooster

    Rockenrooster Limp Gawd

    Messages:
    232
    Joined:
    Apr 11, 2017
    Looks like L2 and L3 are slower as far as GB/sec than a 1700@4ghz bench I found, but L1 is way faster and the Latency on L2 and L3 and RAM are waaaaaaaaaaaaaaaaaaaaaaaaaaaaay better
     
    Flexion likes this.
  35. Nobu

    Nobu 2[H]4U

    Messages:
    3,027
    Joined:
    Jun 7, 2007
    I think we found where that ipc boost came from, at least part.
     
  36. Gideon

    Gideon 2[H]4U

    Messages:
    2,155
    Joined:
    Apr 13, 2006
    Hmm that is quite impressive if it's true.
     
  37. OnceSetThisCannotChange

    OnceSetThisCannotChange Limp Gawd

    Messages:
    130
    Joined:
    Sep 15, 2017
  38. Silentbob343

    Silentbob343 [H]ard|Gawd

    Messages:
    1,754
    Joined:
    Aug 2, 2004
    What was the latency on Zen and Zen+?
     
  39. MMitch

    MMitch Gawd

    Messages:
    667
    Joined:
    Nov 29, 2016
    Quick google seems to point to double latency on memory.
    If this is true it will change a lot of things. Those numbers look better than Intel (memory portion) from the quick search I did.
     
    Silentbob343 likes this.
  40. GoodBoy

    GoodBoy [H]ard|Gawd

    Messages:
    1,332
    Joined:
    Nov 29, 2004
    15% ipc improvement is something AMD has sorely needed for awhile. How close does this get them to Intels' IPC (fully side-channel mitigated Intel)? Once these come out, can't wait to see some comparisons.

    If the IPC becomes closer to Intel, and they are hitting >4.5Ghz boost, at those core counts and prices, it is really amazing! And it is good for all of us, whether you buy Intel or AMD. If you buy AMD, here is a worthy upgrade, if you buy Intel, chances are your pricing just dropped a few hundred $$..

    You guys arguing that IPC isn't important are mistaken. AMD has been behind Intel in this area for years... if they catch up, it means real competition.

    The IPC is also more important than ever, as the shrinking transistors manufactured with silicon have hit the speed barrier already. So overall performance improvements can only come from more cores and better IPC (which is a single core performance metric based on a single clock cycle). And we all know that many games do not benefit all that much from large core counts, so the single core performance (IPC x Ghz) matters most.

    So, question for those of you already running these newer AMD cpu's and the Boost speed. The 3900x is reported to have a 3.8Ghz base with a 4.6Ghz boost speed. How does that work in real world experience? I mean is it only a few cores that get boosted to that speed, or could all of the cores boost to 4.6Ghz simultaneously? Like if you were running something on all cores?

    Excited for some reviews once these come out. Hopefully we can get some comparisons to intel for a variety of workloads from somewhere. The raw benchmarks aren't bad but do not translate into gaming performance directly.
     
    Last edited: May 30, 2019