1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Review: AMD Ryzen 7 1800X processor

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Mar 2, 2017.

  1. user1

    user1 Maha Guru

    Messages:
    1,392
    Likes Received:
    447
    GPU:
    hd 6870
    Seems to follow what the stilt has discovered

    [​IMG]

    2.0-2.2ghz is the bottom of its voltage scaling curve

    Also

    [​IMG]

    https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/
     
  2. dwiewolverine

    dwiewolverine Member Guru

    Messages:
    163
    Likes Received:
    0
    GPU:
    Strix GTX 980 Ti SLI
    AMD is back!!:banana::banana::banana:

    Intel suck our $ for years..they now should be serious when throwing a new chip (innovation-technology-capability-price)...:)
     
  3. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,004
    Likes Received:
    137
    GPU:
    Sapphire 7970 Quadrobake
    This makes me believe that this is going to be a killer notebook APU. Especially if it comes with Vega, which I bet will do similar stuff at lower frequencies.
     
  4. ender79

    ender79 Member Guru

    Messages:
    124
    Likes Received:
    0
    GPU:
    Zotac GTX 970 ~ 1455 Mhz
    I also have a laptop with i7 6700HK and I keep it undervolted all the time with XTU. At 3.15 Ghz with 0.912 V. So for mobile parts is normal to work on lower voltage. Ryzen looks to be in the same range of voltage like Skylake and both are made on 14 nm node

    I can't wait to see mobile parts of Ryzen. To bad the initial release was made only with 3 models. We have months to wait for desktop 6 and 4 core parts and mobile... We even don't know nothing about
     
    Last edited: Mar 6, 2017

  5. ender79

    ender79 Member Guru

    Messages:
    124
    Likes Received:
    0
    GPU:
    Zotac GTX 970 ~ 1455 Mhz
    Keep in mind, Intel GPUs are so weak, on the other hand, AMD make more powerful GPUs, including that ones will be fitted into APUs
     
  6. Alessio1989

    Alessio1989 Maha Guru

    Messages:
    1,386
    Likes Received:
    232
    GPU:
    .
    The full system @fixed 3.0GHz clock was about ~100,3W during Cinebench, ~73.3W less then default clock settings.

    @3.2GHz (fixeed clock) and 0.95V was about ~114.3W (full system).

    This with a MSI X370 XPOWER GAMING TITANIUM. You can read the rest of the system specifications here: http://www.bitsandchips.it/9-hardware/8117-preview-amd-ryzen-la-rinascita-di-amd?start=1
    Note that now Fottemberg is testing the CPU for the review with an updated BIOS.
     
    Last edited: Mar 6, 2017
  7. user1

    user1 Maha Guru

    Messages:
    1,392
    Likes Received:
    447
    GPU:
    hd 6870
    14nm lpp will be really quite good for apus and is very good for zen server chips.

    Makes you really wonder what amd is doing with tsmc 16nm finfet.. :infinity:
     
  8. ender79

    ender79 Member Guru

    Messages:
    124
    Likes Received:
    0
    GPU:
    Zotac GTX 970 ~ 1455 Mhz
    https://www.techpowerup.com/231268/amds-ryzen-cache-analyzed-improvements-improveable-ccx-compromises


    I'm right 100% , Ryzen R7 is not a truly 8 core processor is a dual quad core on the same package , with that 2 CCX linked by a bus with only 22 GB/s ? WTF ? L3 cache latency is 3 times higher than for a i7 6900K (17.3 ns VS 46.6 ns)
    So for gamers the problem with R7 will not be solved with BIOS updates . Of course there are some tricks : run the main thread of the game on one CCX and AI of the game and others to the second CCX that means a new game patch only for R7, but will never be in the same league as Intel even with that tricks.

    R5 with 6 cores will suffer the same issue in games.
    Ryzen with 4 cores will never suffer from this problem, will be the sweet spot for budget gamers.

    Hilbert, please run games with affinity in task manager from core 8 to 15 . Pleaseeeeeeeeeeeeeeeeeeeeeeeeeeeeeee
     
    Last edited: Mar 6, 2017
  9. Denial

    Denial Ancient Guru

    Messages:
    12,343
    Likes Received:
    1,529
    GPU:
    EVGA 1080Ti
    Did AIDA update to read Ryzen cache correctly?

    Also Ryzen's L3 cache isn't used the same way Intel's is. Ryzen only uses the L3 when it's evicted from L2.

    Intel stores everything from L1/L2 into L3 at all times. So even if the latency is higher, how the cache is being used may effect whether the increased latency even matters for performance. If it's even higher. Last time I checked AIDA needed updates to read it correctly and AMD provided it's own cache latency for review purposes and iirc the numbers they provided are similar to Intel's.
     
  10. ender79

    ender79 Member Guru

    Messages:
    124
    Likes Received:
    0
    GPU:
    Zotac GTX 970 ~ 1455 Mhz
    You understand that CCXs are linked with 22GB/s bus? That is hurting games and L3 cache.....
     
    Last edited: Mar 6, 2017

  11. HonoredShadow

    HonoredShadow Ancient Guru

    Messages:
    4,266
    Likes Received:
    10
    GPU:
    MSI 980Ti+KS800 TV
    https://youtu.be/ylvdSnEbL50

    Very interesting video about how the tech press got things supposidly wrong about testing CPUs including Ryzen and the conclusion about Ryzen could be wrong too.

    I think we as gamers might be getting it wrong too...
     
    Last edited: Mar 6, 2017
  12. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,004
    Likes Received:
    137
    GPU:
    Sapphire 7970 Quadrobake
    Any link about that? How do you know it?
     
  13. Denial

    Denial Ancient Guru

    Messages:
    12,343
    Likes Received:
    1,529
    GPU:
    EVGA 1080Ti
    https://www.techpowerup.com/231268/...yzed-improvements-improveable-ccx-compromises

    This paragraph:

    Seems like something developers could work around.
     
    Last edited: Mar 6, 2017
  14. eclap

    eclap Banned

    Messages:
    31,497
    Likes Received:
    2
    GPU:
    Palit GR 1080 2000/11000
    Oh dear, interesting what ender79 is saying. I'll keep an eye on this. Seems like somewhat innovative design on AMD's part.
     
  15. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,004
    Likes Received:
    137
    GPU:
    Sapphire 7970 Quadrobake
    That number sounds way too low to me. What is their actual source?

    Also the AIDA64 cache latency numbers are wrong, which kind of throws all of this under the bus as a theory.

    [​IMG]
     

  16. Turanis

    Turanis Maha Guru

    Messages:
    1,422
    Likes Received:
    155
    GPU:
    Gigabyte RX500
    You know is a kind of conspiracy against Ryzen.

    Ryzen can defeat Intel i7 6800k or 6900k.Intel can't accept this and they strike with their lap dogs from mass-media or some users. ;)

    Bussiness as usual.Thankfully we still have free Internet.
     
    Last edited: Mar 6, 2017
  17. Denial

    Denial Ancient Guru

    Messages:
    12,343
    Likes Received:
    1,529
    GPU:
    EVGA 1080Ti
  18. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    7,004
    Likes Received:
    137
    GPU:
    Sapphire 7970 Quadrobake
    They literally say in the article that the tools can't measure L3 properly.

    I also still await a confirmed source on the only 22GB/sec interconnect, which sounds very peculiar to me. The only source I have managed to find is a Reddit thread where someone literally pulls the number out of their a*s.

    They basically pull the 22GB/sec number from this slide:

    [​IMG]

    What they suppose is that the 32bytes/cycle connection of the L3 to the Data Fabric is done in Memory clock speeds (so 1.2GHz for DDR4-2400). I can see how that would be the speed to the memory controller (although even that is a supposition, nobody says that the memory controller has to operate as a piece of logic at the same speeds as the memory), but it tells me nothing about the frequency that the L3 connects to the Data Fabric itself.

    For all I can see, it's the speed that the L3 actually runs on, which would be the CPU speed and that would bring up the interconnect to a much more reasonable 66GB/sec, possibly bi-directional at that, from the other CCX. I can't see how on-die connections over an interconnect build for that can be slower than DDR4.

    EDIT: It looks like I was wrong. I'm still not 100% sure it's on the memclock speed, or even that what we see is how two CCXs communicate between them, or a vertical slice of how a CPU with a single CCX would operate.
     
    Last edited: Mar 6, 2017
  19. ender79

    ender79 Member Guru

    Messages:
    124
    Likes Received:
    0
    GPU:
    Zotac GTX 970 ~ 1455 Mhz
    Innovative? Hahahah is a sh-it, is a intel core2quad in AMD vision. Is the worst thing that can happen to a modern CPU
     
  20. ender79

    ender79 Member Guru

    Messages:
    124
    Likes Received:
    0
    GPU:
    Zotac GTX 970 ~ 1455 Mhz
    Most of us wanted an eight core ryzen R7, AMD give to us a dual quadcore.... And yes, gaming will not be fixed with a BIOS update,it is simply impossible
     

Share This Page