AMD Ryzen 5 Cores Are Disabled in Symmetrical Pairs

AMD Ryzen CPU Die

There seems to be some confusion about what CCX arrangement the recently announced AMD Ryzen 5 processors will be shipping with, so we’ve been pestering AMD trying to get that sorted out. After speaking with AMD this afternoon we’ve confirmed that AMD is disabling cores in symmetrical pairs only and that all Ryzen 7 and Ryzen 5 processors use the same exact die. This means that you’ll end up with  4:4, 3:3 ,2:2 CCX arrangement and different amounts of L3 and L2 (512kb/core) cache on some of the processors. No Ryzen 5 series processors will be shipping with a 4:0 or 0:4 CCX arrangement like some were thinking.

Here’s rundown of how they compare:

  • Ryzen 7 1800X: 8 cores (4+4), 16 threads, 16MB L3, 4MB L2, 3.6GHz to 4GHz, $499
  • Ryzen 7 1700X: 8 cores (4+4), 16 threads, 16MB L3, 4MB L2, 3.4GHz to 3.8GHz, $399
  • Ryzen 7 1700: 8 cores (4+4), 16 threads, 16MB L3, 4MB L2, 3GHz to 3.7GHz, $329
  • Ryzen 5 1600X: 6 cores (3+3), 12 threads, 16MB L3, 3MB L2, 3.6GHz to 4GHz, $249
  • Ryzen 5 1600: 6 cores (3+3), 12 threads, 16MB L3, 3MB L2, 3.2GHz to 3.6GHz, $219
  • Ryzen 5 1500X: 4 cores (2+2), 8 threads, 16MB L3, 2MB L2, 3.5GHz to 3.7GHz, $189
  • Ryzen 5 1400: 4 cores (2+2), 8 threads, 8MB L3 (split 4MB per CCX), 2MB L2, 3.2GHz to 3.4GHz, $169

Hopefully that clears things up a bit as some of the things we’ve run across online are wrong. We expect Ryzen 3 will be using a new Zen die, but AMD does not comment on unreleased products.

Take a look at the AMD Ryzen 5 Desktop Processor Sneak Peak below where AMD CEO Dr. Lisa Su and VP of Marketing John Taylor discuss Ryzen and head into the labs for a sneak peek into how Ryzen 5 processors are performing in advance of their April 11, 2017 launch.



  • AzKat


    Ryzen 5 is basically Ryzen 7, with broken / faulty cores that are disabled

  • Marechal Golfinho

    Why the fuc* they did that… I knew ryzen 5 hexa would have to use infinity fabric, but the Ryzen quad? damn damn damn

  • @ndrio

    Damn …. I really wanted this x1600 to perform well. Seems like the sweet spot of price/performance …. I have a feeling though that will be a 🍋 😟

  • Artearius Radey

    why is everyone so willing to shoot these processors in the foot without any real world testing is done…jeez talk about negativity give it a chance

  • Dmitri

    Lol, just curios what they will disable for the Ryzen 3.
    since none of them are hitting over 4GHz, seems like all intel have to do is to make i7 a bit cheaper and continue to rape wallets of those who care about gaming.

    • BaronMatrix

      It looks like R3 will be Raven Ridge…

  • Rod O’Neal

    Is everyone here payed by Intel? Or do you just not care that you’ve been overpaying for years and want to continue?

    • Dmitri

      No man, we just prefer to overpay for a successful/reliable/supported/stable product over paying for a failure. It cost less in the long run.
      I happily payed for Athlon 64 x2 significantly more than Pentium D simply because it’s better. But since than, I found 0 reasons to give my money to AMD.

      • Coach

        Well, Ryzen is the real deal Dmitri. It is no Phenom I or Bulldozer, but a legit CPU that can compete. Try out someone else’s Ryzen and see for yourself.

        • @ndrio

          “Trying” is not enough, sorry. Leaving with it 24/7, OC it, use it to the max is the real deal.

        • Dmitri

          Sorry man, ryzen does not meet my perfromance/features requirements.

          1. mini ITX MB (64GB ram support would be an advantage)

          2. 6-8 cores @4.5GHz + iGPU

          3. good iommu grouping

          4. single die – inter die/socket communication is an expensive thing in terms of performance.

          first company to release such thing, will get my money. until that happens, i’ll continue to “suffer” with i7-4790K –

      • @ndrio

        Amen to that 😇

      • Spedez

        Instead, you have all the reasons to give your money to Putin.

  • I had a feeling they’d f something up… just like they did with the FX series, where their 8 core was more lika a 4 core, with physical hyper-threading… (2 cores sharing one FPU)
    and here they are doing something similar again… trying to keep costs down with shortcuts like these, ultimately doing more harm then good for themselves

    • shumok62

      I am sure they did it for a reason (yields and scalability or some such thing maybe),….and everything I have seen shows the loss in performance is not that big.

  • Michel Outils

    This is bad news. If AMD pushes the 2+2 CCX configuration on Ryzen 5’s 4c/8t it’ll be an insurmountable disaster.

    • shumok62

      Why? Simulated stuff I’ve seen shows good performance.

      • Marechal Golfinho

        because the cores will have to communicate with each other by the infinity fabric instead of the L3 cache

        • shumok62

          It’s not a big sacrifice in performance at all. It might be a big cost saver for AMD though.
          I am sure they didn’t do it for no reason….and since the performance hit is so small…I suspect it was a good decision.

        • Marechal Golfinho

          It’s always faster to cores to communicate with each other by the L3 cache instead of the FSB/IF/QPI ….

        • shumok62

          So what? Are you saying cost is no object? Every processor has compromises.
          I’ll bet if they doubled the cache size they could have gotten more performance…should they have doubled the cache size?
          You make no sense at all.

        • Marechal Golfinho

          So what? I understand that it is a cost saver to use the same die over all the Ryzen R3 R5 and R7 CPUs instead of ordering specific dies… But as the R5 has already 2 CCX on its die, it would be much better to users if it uses only one CCX and its 4 cores (all integrated by the L3 cache), instead of disabling 2 cores on each CCX and making the cores communicate with each other by the Infinity Fabric….

          In my opinion AMD did this to avoid R5 quad-core performance to be superb and overcome the r5 hexacore…

        • shumok62

          No it would not be much better….the difference in performance is just a few percent. I believe the dies also need to be symmetrical, and if you could disable a whole CCX i think you also disable 1/2 the cache on the chip.

          If a ccx has 1 bad core…are you better off disabling the whole ccx or just 1core on the other ccx? I think it’s obviously better to just disable 1 core. I think the logic mostly follows on down….what about chips with 1 bad core on 1st ccx and 2 bad cores on 2nd ccx? So to get the most yields you want to keep these.

          Anyways, I suspect that the R3’s will be 1ccx chips along with Raven Ridge.

        • Marechal Golfinho

          L3 cache >>>>>> Infinity Fabric / FSB / QPI… Its like you requesting some info on your job to a colleague right next to you (L3 cache) instead of getting up to another room to gather information (infinity fabric, QPI, FSB)…

  • Realzob

    aw, i wished 4 cores r5 would be 4:0 to avoid communication betwenn CCx that cause huge slowdowns.. as some youtube reviews shows, r7 with only 1 CCX activated are doing better on a number of games than with all core enable.. so, All R5 will be utter shit because having 2 CCX with less cores in each, inter CCX communications wil happens a lot more frequeltly, causing even more loss in performance..

  • Alex Alexandrewitsch

    I think they can be acitvated

    • Mediasearchprogram

      Good question.. if software disable then possibly yes.

      • Wibble

        They are probably failed 8 core samples i.e. nothing to activate