Announcement

Collapse
No announcement yet.

AMD Ryzen 7 CPUs Shipping 2 March, Pre-Order Today

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • Originally posted by Qaridarium
    world record? YES YES YES YES :-)
    OK then, if you wanna world linux record better use at least 4.9.10+ kernels, otherwise even Zen SMT might be broken

    commit c8cbc219d87cdbe33430b92350cb687b3f2201e6
    Author: Yazen Ghannam <[email protected]>
    Date: Sun Feb 5 11:50:22 2017 +0100

    x86/CPU/AMD: Fix Zen SMT topology

    commit 08b259631b5a1d912af4832847b5642f377d9101 upstream.

    After:

    a33d331761bc ("x86/CPU/AMD: Fix Bulldozer topology")

    our SMT scheduling topology for Fam17h systems is broken, because
    the ThreadId is included in the ApicId when SMT is enabled.

    So, without further decoding cpu_core_id is unique for each thread
    rather than the same for threads on the same core. This didn't affect
    systems with SMT disabled. Make cpu_core_id be what it is defined to be.
    https://cdn.kernel.org/pub/linux/ker...angeLog-4.9.10

    testing stock @65watt TDP (1700) is pointless testing stock @95watt TDP (1700X) on AIR cooling is a good start.
    Testing any of these is good point on Linux, since this is new and things might not be even supported or setuped fine 100% yet as someone might wanna to expect

    Especially one can expect big fail on current released Ubuntus So much about world record, just forget about it really
    Last edited by dungeon; 23 February 2017, 03:29 PM.

    Comment


    • Originally posted by theriddick View Post
      it will be mostly reviewers buying those out is my suspicion. Most people are already rocking intel hardware and won't be migrating. I have a 5820k which is fine, I'm certainly interesting in Ryzen but will wait until a metric truckload of analysis has been done on the mobos and cpus first (like 3-4months from now).
      I'm posting this from a PC with a Phenom X6 1055T that's celebrating his sixth birthday and runs well enough for Starcraft 2, Grey Goo, Minecraft, ffmpeg, HandBrakeCLI, etc.... so these new AMD releases might be a nice upgrade. (I've upgraded GPU since putting this together, but not CPU.)

      But I'm definitely not pre-ordering. If it looks good, maybe I'll buy one later this year.


      Comment


      • Originally posted by beniwtv View Post

        Did not have any of these problems on my two FX 8350 + AMD RX 460/480 PCs with ASRock motherboards. Maybe it's rather the motherboard you were using that was not good?
        I have used Manjaro, SteamOS (now removed, as I don't need it anymore) and Ubuntu,
        For what it's worth, I have the same problems as B.Jay on my system with a Gigabyte motherboard and FX-8320. AMD-Vi errors in the logs and USB problems.

        I'm glad you had a better experience.

        (Edit: for anyone reading both posts, I have two PCs in the house. My main is actually running the older Phenom X6 1055T and the kids use the one with the FX-8320.)

        Comment


        • Originally posted by Qaridarium
          juno please shot your self ! the max XFR clock for the 1800X is 4,1ghz and the max XFR clock for the 1700X is 3.9ghz
          Thanks for the kind offer.
          4.1 is still 100 above 4.0, 3.9 is still 100 above 3.8. WTF do you want from me?

          Originally posted by Qaridarium
          yes i expect michael to benchmark XFR technology on different coolers like AIR vs liquid vs GAS
          Have you sent your tip already to fund the liquid cooling loop?
          Last edited by juno; 23 February 2017, 04:44 PM.

          Comment


          • Originally posted by duby229 View Post

            That's total bullshit. What you just said is fanboyism. No other x86 architecture scales per thread like AMD's CMT architectures. It definitely beyond any doubt has the highest performance potential compared to any other x86 design. By at least 50% and that's huge. The fact that AMD didn't scale it is what the problem was and not the architecture. It was and still is bad ass.

            If AMD scaled up a modern CMT architecture to 8 actual cores with 3 integer units per pipeline it demolish ZEN.
            Yikes, no need to get your panties in a wad, I'm speaking only to the real world performance of the products that shipped. The design may well be brilliant on paper, but the performance of the real SKU's was underwhelming - they consumed more power and had lower integer and floating point performance than the competing intel chips, and that's a fact.

            They were priced accordingly though, so they were actually a very good value in many scenarios. After all, server workloads don't care too much about single threaded performance. The last round of server procurement I did in 2013 was all Dell R815's with 16 core Opterons. We have hundreds of G34 sockets in production right now today because I saw them as the best value at the time. The last batch of 10 servers I bought in 2013 were more than $100,000 cheaper with Opteron than with Xeon. That's a no brainer, saving that much cash.

            Comment


            • Originally posted by Michael_S View Post
              I'm posting this from a PC with a Phenom X6 1055T that's celebrating his sixth birthday and runs well enough for Starcraft 2, Grey Goo, Minecraft, ffmpeg, HandBrakeCLI, etc.... so these new AMD releases might be a nice upgrade. (I've upgraded GPU since putting this together, but not CPU.)

              But I'm definitely not pre-ordering. If it looks good, maybe I'll buy one later this year.
              Nice, my main workstation is a Supermicro H8SCM with Opteron 4386, 16GB of DDR3-1600 ECC, and an original GTX Titan card. It turned 4 years old this month! Personally I'm holding out for the Zen server chips (Are they still called Opteron?) so realistically I'm looking at Spring of 2018 for an upgrade.

              Comment


              • Originally posted by castlefox View Post
                I'll be donating/tipping a solid 5 dollars towards Ryzen getting bench marked.
                Just donated now.

                Comment


                • Originally posted by juno View Post


                  Base clock is 3.4, boost clock is 3.8, so xfr brings an additional 100 MHz.

                  Clocks for multi core boost are still TBC.
                  Perhaps you need to look at XFR as an autopilot that scales the clock, as efficiently as possible, without a brute force boost from 3.4 to 3.8, but 25Mhz increments from 3.4 to 3.9 and beyond depending on one's cooling options. You can run the control to set boost to 3.8 or less.

                  Comment


                  • Originally posted by debianxfce View Post

                    Little math exercise for intel believers, calculate how many cores you see in this picture:
                    All you have to do is look at any block diagram... Literally any one of them.


                    You can obviously see how it takes the whole module to be a an x86 core.



                    Even AMD's own diagrams find it necessary to compare a CMT module to a Zen core because that's what a CMT module actually is.

                    Comment


                    • I think Michael might ask MSI for motherboards.

                      Comment

                      Working...
                      X