In the lab: AMD’s Ryzen Threadripper 1920X and 1950X CPUs

Comments closed
    • Jigar
    • 2 years ago

    TR reviewing TR – Nice.

    • Bigbloke
    • 2 years ago

    TR[super<]3[/super<] -> TechReport Totally Reviewing ThreadRipper

      • jihadjoe
      • 2 years ago

      (TR)[super<]3[/super<]

      • Growler
      • 2 years ago

      You left out “Truly Radical”.

      • Anovoca
      • 2 years ago

      “TechReport’s, they reckon Thursday released, thoroughly reviewed ThreadRipper Test Results

    • Solean
    • 2 years ago

    Threadripper 1920X is already in stock in at least 2 countries, Spain and Germany, at german retailer online store Alternate.

    Price for Spain is 869 euros.
    Price for Germany is 829 euros.

    Good job AMD and Alternate.
    _________________

    Edit: Was in stock. Either an error with Alternate, or people just bought all stock in the last 30 minutes since I posted my original comment.

    • Unknown-Error
    • 2 years ago

    Please run a test to see if that linux error can be re-produced on window.

    • SomeOtherGeek
    • 2 years ago

    Are those tin cans!?! Not at all what I commented on awhile ago.

    • orik
    • 2 years ago

    would be very interesting to see a gaming + OBS cpu streaming comparison of both platforms.

    • DoomGuy64
    • 2 years ago

    [quote<]Stay tuned for full launch-day coverage, including gaming results[/quote<] I don't recall seeing the i9-7900X review leading with gaming results, and if TR is not reviewed in the same meticulous manner, that's not going to look like a professional review. People don't buy these chips for gaming, and companies don't make these chips for gaming. The question really is, which chip is right for what workload? Intel has an edge in AVX, so is the TR review going to test big picture or AVX/gaming? The only reason why I feel this even has to be asked, is the atmosphere of both the readers and writers are implying that is [i<]exactly[/i<] how TR is going to be reviewed. Hopefully I'm just mis-reading the room, and this is an inside joke, since a review based primarily on AVX/gaming is not going to look like a honest or quality review.

      • xigmatekdk
      • 2 years ago

      Hell nobody is gonna pay $1000 for an AMD product. Period. The regular AMD shopper is a low budget builder.

        • MOSFET
        • 2 years ago

        Well with your username, you would know.

        HEDT shoppers are not budget shoppers, no matter which company/brand/platform you’re shopping. If a Ryzen 7/5 is what you need, you won’t be paying $1000 for the whole system.

        • Amiga500+
        • 2 years ago

        Your definitely no older than 25 and were not around in the pre-Conroe days.

        • just brew it!
        • 2 years ago

        The whole point of Threadripper is to broaden their appeal beyond “the regular AMD shopper”. They had $1000+ Opterons back in the day, and even had several HPC design wins with those CPUs. The perception that they are strictly a “low budget” option comes mainly from the K6-x and Bulldozer/Piledriver/etc. generations.

        They’ve played in the high end before, now they have a shot at doing it again.

        • Anonymous Coward
        • 2 years ago

        I fondly remember my bachelor days, I bought an A64 FX single core and a year later two dual cored Opterons for over $1k each. Of course that was easier to justify at that stage of desktop computing. Quad cores goes a long ways.

      • auxy
      • 2 years ago

      [quote=”DoomGuy64, paraphrased, “<]A review that makes Threadripper look bad is automatically dishonest and poor quality.[/quote<]PLS. I don't even post here anymore and I'm still tired of you. Grow some damn self-awareness! ( ・`д・´)

        • EzioAs
        • 2 years ago

        Hey, auxy’s back!!

          • Glix
          • 2 years ago

          How could you tell?

          ヾ(⌐■_■)ノ

        • Sahrin
        • 2 years ago

        That’s not at all what he wrote. Intel reviews tend to accentuate the positive, AND reviews focus on negatives. Treating them both in the same context makes sense.

          • chuckula
          • 2 years ago

          Care to show any evidence? If anything TR has bent over backwards to be fair to AMD.

          To the point of going out of its way to completely exclude Intel from large sections of its most recent system guide even though Intel systems were the only ones on the market that would even let you have more than 16 lanes of PCIe connectivity direct to the CPU.

          • ludi
          • 2 years ago

          Before Ryzen was released, AMD was falling further behind in every metric: power, performance, features, cost effectiveness. You can only mumble it gently so many times before it starts to become a glaring deficit in the product lineup.

          Compare that to the response Ryzen and TR are getting, and you can see everyone’s excited again. The only negative I remember from those was the criticisms of Ryzen 3 for attempting to fill a market niche where it needs to either have onboard graphics or else be priced a whole lot cheaper to account for the cost of the add-on video card.

          • K-L-Waster
          • 2 years ago

          Riiight, TR is soooo anti-AMD they gave the R5 line an Editor’s Choice and put RyZens in their suggested builds in the latest system guides.

        • shaq_mobile
        • 2 years ago

        I miss your emoticons

      • Redocbew
      • 2 years ago

      You win the prize for posting useless complaints before the review is even posted. Congratulations. Good job.

        • DoomGuy64
        • 2 years ago

        I’m not complaining about the review. I’m complaining about the [i<]people[/i<] here. The atmosphere here doesn't seem to be leaning towards the objective route. If you guys can't take a serious product serious, nobody will take this site serious. There is already enough problems, doubling down on unobjective review methods just because you want to see AMD crash and burn for your entertainment isn't going to work. This isn't a bubble. Other sites are reviewing the CPU. This site won't be here next year if it can't be objective.

          • Redocbew
          • 2 years ago

          Oh, right. I want AMD to fail, because I don’t make up half baked theories which cast them in a positive light. If I did, then that would make some kind of difference. I forgot that part.

          • K-L-Waster
          • 2 years ago

          The RyZen reviews covered both productivity and gaming.

          They also were pretty objective and generally favourable (R5 even got a gold Editor’s Choice award).

          Not sure where the heck you’re getting the idea ThreadRipper is being set up for a biased report.

            • DoomGuy64
            • 2 years ago

            Read the comments. That’s where I’m getting it from. The review could very well be professional, but you people are insinuating that it not only won’t be, but it shouldn’t be.

            Readers always reflect the bias of the site they’re reading, and if a majority of the readers are calling for a bad review, chances are they have enough experience to know that’s what they’re gonna get.

            That, or this site just has a prolific anti AMD troll problem.

      • chuckula
      • 2 years ago

      I don’t see what the issue is.

      Aren’t you excited that TR went out of its way to not show the 7900X (and now the comparably priced 7820X) beating the R7 1800X in games when that was AMD’s absolute best CPU?

      Aren’t you excited that TR waited for the 16 core ThreadRipper to launch so that there’s no gap in showing AMD’s gaming superiority? After all, with a 60% core advantage and higher base clock speed and the magically delicious RyZen miracle architecture it’s pretty much guaranteed to dominate in gaming benchmarks!

      • natesland
      • 2 years ago

      Save your energy… AMD most likely wins this round.

    • chuckula
    • 2 years ago

    [quote<]Stay tuned for full launch-day coverage, [b<][i<]including gaming results[/i<][/b<], this Thursday.[/quote<] I mean, don't knock yourself out on our account or anything there! We care about your health and avoiding sleepless nights! In fact uh... just run Cinebench and call it day? K-thanks! -- AMD

    • Longsdivision
    • 2 years ago

    Did TR get the whole package everyone else got?

    Would like to see the TR paperweight…would like to see TR bench said paper weight.
    …will be the fastest paperweight ever.

      • ludi
      • 2 years ago

      Not unless AMD issued number 251/250. That’s just fluff anyway, there will be a hundred other review sites showing us how to open a monogrammed Pelican case. TR does numbers.

    • jts888
    • 2 years ago

    My review wish: a few tests with the highest clocking RAM possible (3000-3600 MHz?).

    If even inter-die IF scales with memory clocks, I’m curious how much extra performance can be effectively bought in DIMM form.

      • jihadjoe
      • 2 years ago

      I’m sure they will. TR tested Ryzen with 3866 rated RAM, unfortunately it was early days for the platform and thus only ran at 2933, but timings were sufficiently tight.

    • Captain Ned
    • 2 years ago

    The NDA expires in two days, we have two chips to test, there will be no sleep, the gerbils will still complain, and he’ll be wearing sunglasses.

    Hit it.

      • morphine
      • 2 years ago

      That’d be funny if it weren’t true, Captain.

        • derFunkenstein
        • 2 years ago

        It was funny anyway.

      • TwistedKestrel
      • 2 years ago

      Use of unnecessary violence has been approved

        • Captain Ned
        • 2 years ago

        Thank Bob someone finally cottoned on to where I was going with that.

          • UberGerbil
          • 2 years ago

          “I’ve always loved you.”

      • Anovoca
      • 2 years ago

      I HAVE TO WAIT TWO DAYS?! THIS IS AN OUTRAGE!!!

      • TheMonkeyKing
      • 2 years ago

      Why are you still here typing? Go! Don’t come back until you’ve got some scatter charts in your paws.

      • LostCat
      • 2 years ago

      I wanna complain! I can’t think of anything. Damn it.

        • Redocbew
        • 2 years ago

        You fail at complaining. This is as it should be.

    • maxxcool
    • 2 years ago

    Since this is a Workstation CPU, I would like to see a few MySQL tests, that do NOT fit in the Caches hierarchy.

    I know everyone is focused on testing the cpu to the tiniest detail by doing it all on die on cache, but real world will be loading the thing up with 256gb of ECC and running hundreds of threads.

      • Jeff Kampman
      • 2 years ago

      If you have sample workloads and can give us an idea of what performance metrics to measure, I’d be happy to see whether it’s something we can incorporate into our tests.

        • maxxcool
        • 2 years ago

        Wish I could, i’m that annoying idea guy in the back of the room .. 🙂 I do not have the expertise to provide, but saw the concerns on the other reviews to warrant a thought to toss in the hat.

        • techguy
        • 2 years ago

        I would love to see a sample Plex server benchmark. There’s not a real benchmark, per se, just setup a Plex media server and start streaming Blu-ray quality 1080p video files to multiple clients and tell us how many active streams it takes before it starts to hiccup. Better yet, if you have any 4k videos you can perform the same test with that would be perfect.

        If that’s not feasible, could you run the Passmark CPU test? That’s the benchmark Plex recommends as a basis of comparison for Plex transcode engine performance.

        • blastdoor
        • 2 years ago

        I would totally understand if your initial tests are focused on Windows, but if you ultimately could do a Linux comparison between Threadripper and Intel’s Core i9, I would love it if you would run some multicore R benchmarks and I’d be very happy to make some up for you.

        • gamoniac
        • 2 years ago

        I think for me the most interesting part of having such a powerful workstation with so many threads would be to have several VMs running in full throttle – a DB server, a couple of web servers, a messaging server (RabbitMQ), a file server, a VM with some development IDE (Visual Studio, NetBeans), etc. and still be able to browse the web or game at the same time. it would take quite a software setup though to have a meaningful simulated communication between these servers. A couple beefy NVMe SSDs would probably be good to mitigate potential I/O bottleneck.

      • Srsly_Bro
      • 2 years ago

      Anandtech did an epyc review a month or so ago. There you will find what you seek.

        • maxxcool
        • 2 years ago

        If you check the comments you will see that the size of the DB was very small, and there were concerns about the efficacy of the test since the memory controllers and large portions of ram were not used.

        I looked at it but i am not enough of a DB guy to comment but i get what they are saying and agree it was a odd choice of benching.. they have done LARGE datasets in the past with the old POWER7 review they did so it REALLY struck me as suspect?

        • maxxcool
        • 2 years ago

        not sure why your getting negged, that article has useful data if you exclude the ‘odd’ database test.

          • Srsly_Bro
          • 2 years ago

          Haters gonna hate, i’m just just gonna shake. sup?

        • Kougar
        • 2 years ago

        Anandtech performed minimal to no configuration of tests for that review. For desktop benchmarks it’s one thing, but as a stand-in for enterprise workloads the benchmark results are questionable.

        Back when Anand was at the helm the site usually spent 3-6 months configuring, validating, and testing enterprise workloads before posting reviews.

    • just brew it!
    • 2 years ago

    Kewl. Really looking forward to the article!

    I have a small request. Once you’re done with all of your normal review testing, could you possibly run one of the tests aimed at reproducing the Ryzen [s<]parallel compile segfault issue[/s<] *ahem* "performance marginality problem"? I realize this is probably an obscure enough test that you wouldn't want to include it in your main review coverage, but it would be interesting to get a data point from a trusted source to either support or refute AMD's claim that TR and Epyc are not affected. This is a pretty important issue for a lot of people who do software development on Linux.

    • chuckula
    • 2 years ago

    Kampman better not take any selfies because those chips will rip his threads off!

      • dpaus
      • 2 years ago

      Sorry, but the best I can do is give you an ‘E’ for Effort…

      • Anovoca
      • 2 years ago

      Better be careful with your C+ jokes or Kampman will rip your post off this thread.

Pin It on Pinterest

Share This