Stories
Slash Boxes
Comments

SoylentNews is people

posted by janrinok on Saturday February 11 2017, @08:19AM   Printer-friendly
from the start-saving-your-pocket-money dept.

ShopBLT in the US, and Kikatek in the UK, let slip pricing details a bit early.

http://hexus.net/tech/news/cpu/102322-uk-us-prices-amd-r7-ryzen-processors-spotted/

VideoCardz reports that in the US an outfit called ShopBLT (sounds like a sandwich shop) has published a trio of R7 Ryzen chip prices, with some other accompanying details. [...] The absolute top end AMD Ryzen 7 1800X is currently listed at US$490, the Ryzen 7 1700X at $381, and the Ryzen 7 1700 at $316. Remember US prices don't usually include state tax which varies depending where you live.

[...]

In the UK we have a screenshot of a listing of Ryzen CPUs from trade seller Ingram Micro. These listed processors seem to have been taken down, but luckily VideoCardz took a snap. You can see the top end 4GHz AMD Ryzen 7 1800X was listed at GBP £365, the Ryzen 7 1700X at £283, and the Ryzen 7 1700 at £235. These are ex-VAT prices so you have to add 20 per cent, unfortunately. That makes the AMD Ryzen 7 1800X £438 by my calculations. In the listings WOF seems to mean 'without fan'.

-- submitted from IRC


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 0) by Anonymous Coward on Saturday February 11 2017, @09:22AM

    by Anonymous Coward on Saturday February 11 2017, @09:22AM (#465724)

    Seeing AMD offering 4 cores min CPUs makes me wonder if it had given up on single-threaded performance.
    After using these CPUs for 10 years I'm fairly convinced that they are both excessive and at the same time inadequate on the desktop.
    Most software hasn't moved past single-threaded design. Having realized this I'm now getting an i3 with the best single-threaded performance available, for $170.

  • (Score: 2) by butthurt on Saturday February 11 2017, @09:42AM

    by butthurt (6141) on Saturday February 11 2017, @09:42AM (#465728) Journal
  • (Score: 5, Insightful) by The Mighty Buzzard on Saturday February 11 2017, @10:39AM

    by The Mighty Buzzard (18) Subscriber Badge <themightybuzzard@proton.me> on Saturday February 11 2017, @10:39AM (#465736) Homepage Journal

    As a rebuttal, I offer you this: MAKEOPTS="-j17" emerge world

    --
    My rights don't end where your fear begins.
    • (Score: 2) by Techwolf on Saturday February 11 2017, @03:19PM

      by Techwolf (87) on Saturday February 11 2017, @03:19PM (#465782)

      This is why I always point out the Intel chip "fake" cores all the time. It single threaded performance is good. Threaded performance is better when only two or four threads are used. Any higher and performance drops as the number of threads outnumber the number of real cores. This is for Intel desktop chips, that max out at four cores. There zeon line however, can go up to 16 cores if you have the money. AMD desktop CPU have 8 real cores vs. Intel 4 cores plus 4 fake ones.

      • (Score: 2) by The Mighty Buzzard on Saturday February 11 2017, @03:26PM

        by The Mighty Buzzard (18) Subscriber Badge <themightybuzzard@proton.me> on Saturday February 11 2017, @03:26PM (#465787) Homepage Journal

        Looks like the Ryzen ones are going to have 8 real cores and hyperthreading as well so efficiency should top out at building with 17 threads. Two for each core and one to take up any slack. After building Firefox several times trying to get a good PGO build today, I can readily appreciate that.

        --
        My rights don't end where your fear begins.
      • (Score: 2) by turgid on Saturday February 11 2017, @07:54PM

        by turgid (4318) Subscriber Badge on Saturday February 11 2017, @07:54PM (#465863) Journal

        In years gone by, intel CPUs scaled very poorly in multi-processor machines because of a poor front side bus architecture. AMD brought out Hypertransport in 2003 with the Opteron, which was basically the same interconnect that Cray, Sun and SGI used on their big SMP boxes. The intel equivalent didn't come out for another 5 years, Quickpath in 2008.

        Having many cores is great, but you need the I/O bandwidth to keep them fed. You also need to think about locality of memory. In some cases it might be faster to have more physical CPUs, since each comes with its own memory controller, e.g. two 4-core chips vs. one 8-core chip.

        Hyperthreading (or "virtual cores") is a trick to hide some memory latency and utilise some processing resources when another thread is stalled. You can expect a few tens of percent extra performance from adding a virtual core, I have access to some very powerful workstations (intel 8, 12, 16 cores + hyperthreading), but unfortunately don't have tine to play with them to investigate fully. We're always running out of RAM, not CPU power, though. We don't have any big AMD systems. I'd love to compare them.

      • (Score: 0) by Anonymous Coward on Sunday February 12 2017, @01:09AM

        by Anonymous Coward on Sunday February 12 2017, @01:09AM (#465948)

        Running "/usr/bin/time nice -n 20 git gc --aggresive" on the Mesa3D repo reports 600-650% CPU using a Nehalem 4 core - 8 thread CPU. At the same time the machine is playing music and doing other light things too, which have higher priority so will kick git when they ask for cpu slices. Looking at htop I can see the start and end of the process is not parallel but "time" measures everything (see https://en.wikipedia.org/wiki/Amdahl's_law [wikipedia.org] ), meaning that in the parallel zone, using 8 threads, it must be well above 6.5x factor.

        Maybe one day I should play with "cpuset" cmd (and "cpupower" to limit the turbo) and do a 4 cores vs 2 cores with HT. Intel HT hasn't been the crap of the first versions for ages, at least if the coder knows how to implement parallelism.

        Yet I hope AMD has some good CPUs with Ryzen, Intel has been sleeping for too long. We need more Athlon/PIII and Athlon64/Netburst times.

    • (Score: 4, Interesting) by Azuma Hazuki on Saturday February 11 2017, @06:35PM

      by Azuma Hazuki (5086) on Saturday February 11 2017, @06:35PM (#465847) Journal

      As a rebuttal to your rebuttal, I offer this: https://blogs.gentoo.org/ago/2013/01/14/makeopts-jcore-1-is-not-the-best-optimization/ [gentoo.org]

      tl;dr: -jX should have X equal to the number of threads (not cores), NOT numthreads + 1 any longer. And you should limit the system load to 1 * X as well. This is especially true if, like me, you are using the MuQSS or BFS schedulers, apparently.

      --
      I am "that girl" your mother warned you about...
      • (Score: 2) by The Mighty Buzzard on Saturday February 11 2017, @08:19PM

        by The Mighty Buzzard (18) Subscriber Badge <themightybuzzard@proton.me> on Saturday February 11 2017, @08:19PM (#465872) Homepage Journal

        Interesting. Cheers.

        --
        My rights don't end where your fear begins.
        • (Score: 2) by Azuma Hazuki on Saturday February 11 2017, @08:39PM

          by Azuma Hazuki (5086) on Saturday February 11 2017, @08:39PM (#465877) Journal

          Competence means staying on top of new developments :D I love this Ivybridge mobile i7, what a huge jump from that old Core 2 Duo...

          --
          I am "that girl" your mother warned you about...
          • (Score: 2) by The Mighty Buzzard on Saturday February 11 2017, @10:06PM

            by The Mighty Buzzard (18) Subscriber Badge <themightybuzzard@proton.me> on Saturday February 11 2017, @10:06PM (#465912) Homepage Journal

            I was seriously considering building a Ryzen box and retiring my Phenom II x6 for a few minutes but I just can't bring myself to give AMD (and whoever they decide to share it with) a back door into my system. I mean, do you trust Trump not to force them to give him the keys?

            --
            My rights don't end where your fear begins.
            • (Score: 2) by Azuma Hazuki on Sunday February 12 2017, @05:45AM

              by Azuma Hazuki (5086) on Sunday February 12 2017, @05:45AM (#466026) Journal

              ...uh...what? 1) you're supposed to blame Obama 2) if you think Intel's any safer you're nuts 3) no one in the Puzzle Palace gives a shit about your bestiality-watersports-bdsm-electroshock torture porn habit.

              --
              I am "that girl" your mother warned you about...
              • (Score: 2) by The Mighty Buzzard on Sunday February 12 2017, @11:58AM

                by The Mighty Buzzard (18) Subscriber Badge <themightybuzzard@proton.me> on Sunday February 12 2017, @11:58AM (#466101) Homepage Journal

                I know you want to think I'm all "yay Trump" so you can argue with me but that's never been the case. I just hated Hillary worse.

                Intel's even worse, which is why I'm sticking with my Phenom II x6 until the wheels fall off and hoping Libreboot gets done with neutering the ME on ARM by then. I mean there's no way RISC-V will be useful by the time this box dies on me and I'm unable to find replacement parts.

                No, but Obama certainly gave a shit about me enough to decide veterans with anti-governmental tendencies needed to be watched [fas.org] as potential dangers. To be fair, he was absolutely correct. Those who've sworn to defend the Constitution are always going to be a danger to men like him and Cheeto Jesus. So I'll just go ahead and not be spied upon any more than necessary, thanks.

                --
                My rights don't end where your fear begins.
                • (Score: 2) by Azuma Hazuki on Sunday February 12 2017, @10:05PM

                  by Azuma Hazuki (5086) on Sunday February 12 2017, @10:05PM (#466315) Journal

                  Hah, half of that actually made rational sense, for once. Buy up Thuban CPUs and compatible motherboards from ebay if you're that worried; there's plenty around.

                  --
                  I am "that girl" your mother warned you about...
  • (Score: 4, Insightful) by turgid on Saturday February 11 2017, @10:51AM

    by turgid (4318) Subscriber Badge on Saturday February 11 2017, @10:51AM (#465737) Journal

    Most software hasn't moved past single-threaded design. Having realized this I'm now getting an i3 with the best single-threaded performance available, for $170.

    Horses for courses. It's pretty trivial to utilise all cores from the shell. For those of us with large amounts of data, many cores are useful.

    • (Score: 3, Interesting) by rleigh on Saturday February 11 2017, @07:55PM

      by rleigh (4887) on Saturday February 11 2017, @07:55PM (#465864) Homepage

      Exactly this. While people go crazy over single-threaded benchmarks, I specifically bought an AMD FX-8350 despite its "bad" single-threaded performance and power consumption because I valued having 8 cores for building stuff and doing complex computation. Single-threaded performance might be "poor", but I really don't care that much. If it does things slower but 8 times in parallel, I'm still getting better performance overall. If the new CPUs have 16 threads then if it offers a concrete benefit, I'll definitely look at upgrading.

      The other thing people often forget is how amazing all modern CPUs are. When comparing CPU benchmarks, the graphs are often badly scaled, missing out the zero point, to show the relative variance but not the absolute scale; when observed as absolute values from zero the differences would appear to be minimal in some cases, yet people get massively worked up over the minor differences.

  • (Score: 2, Informative) by Anonymous Coward on Saturday February 11 2017, @10:52AM

    by Anonymous Coward on Saturday February 11 2017, @10:52AM (#465738)

    Most software hasn't moved past single-threaded design.

    Maybe, but the operating systems are multitasking since long ago so they can take advantage of multi-cores for running several single-threaded programs in different cores.

    • (Score: 0) by Anonymous Coward on Saturday February 11 2017, @11:08PM

      by Anonymous Coward on Saturday February 11 2017, @11:08PM (#465931)

      I'm shocked more people don't understand this. I have a laptop running 112 processes and over 1600 threads right now.

      When a thread hits the fan, the CPU utilization will be 12.5% or maybe 6.25% instead of 25%.

  • (Score: 0) by Anonymous Coward on Saturday February 11 2017, @03:57PM

    by Anonymous Coward on Saturday February 11 2017, @03:57PM (#465795)

    given up on single-threaded performance.

    I guess you didn't hear that they improved IPC by 40%. Try researching before you post.

    • (Score: 2) by jasassin on Monday February 13 2017, @03:32AM

      by jasassin (3566) <jasassin@gmail.com> on Monday February 13 2017, @03:32AM (#466433) Homepage Journal

      given up on single-threaded performance.
      I guess you didn't hear that they improved IPC by 40%. Try researching before you post.

      What does inter process communication have to do with single threaded code?

      --
      jasassin@gmail.com GPG Key ID: 0xE6462C68A9A3DB5A
  • (Score: 2) by jasassin on Monday February 13 2017, @03:03AM

    by jasassin (3566) <jasassin@gmail.com> on Monday February 13 2017, @03:03AM (#466429) Homepage Journal

    Damn. Look on newegg for i5 3100. You can get a complete refurbished HP with 2TB and 4GB for $135.

    --
    jasassin@gmail.com GPG Key ID: 0xE6462C68A9A3DB5A