Meet AMD Ryzen: The processor that should stress Intel® HWzone
ComputersProcessorsFeatured articles

Meet AMD Ryzen: The processor that is supposed to pressure Intel

updating: Great event of Behind us, and he gave us some demos and some new hopes for cores And the first models to be based on - but not all the answers unfortunately, with price tags and specific launch dates remaining confidential.

Get updates from everyone in TelgramGet updates from us all at TelgramJoin the channel now


In the chip developer, they made a direct comparison between the top-of-the-line Ryzen with eight physical cores and 16 logical cores and Intel's latest octagonal Core i7-6900K (which carries a massive price tag of 1,100 dollars, Specify at every opportunity), and demonstrate a similar and even better level of performance in the Blender processing program, HandBrake video coding software - and in-game 1, which runs in conjunction with a Titan X monochromatic pair of cards in maximum settings and resolution Maximum of course.

amim amim1 amim3 amim2

All of this sounds interesting, especially if you assume that the processor is May carry a price tag lower than that of 50 In this case) plus much more competitive prices for the motherboards, due to the use of the platform New to all price categories at the Sunnyvale company) - but we keep wondering about mainstream models that will be more relevant to most consumers with four-core processing, and their capabilities compared to mainstream processors with far more sane price tags from the competitor. We have no news in this area, unfortunately, and we seem to need to wait until the CES 2017 exhibition, where the picture will become clearer some more.

amim6 amim5

The original article: Just hours before its big online event, The first find their way to the network and give us technical data that were missing until now

Very soon will begin AMD's new Horizon event, Which will finally give us an official look at the processing capabilities and gaming capabilities of the E- The new and the first commercial models to be based on - but even before that happened, modeled slides that came online via the videocardz website provide us with some interesting technical details about , The new trade name to get used to.

Ryzen will be the name of the Summit Ridge family processors, it seems - and the new flagship model will contain 8 physical processing cores, 16 logical processing cores, 20 megabytes of cache that are split into 4 megabytes of ultra-fast memory Cheat L2 plus 16 megabytes of memory cheat L3 - and 3.4GHz or higher working frequency, which may indicate that in Are still unable to fully realize the full potential of the architecture and the latest production process on 14 nanometers, Compromising slightly to withstand the thermal shell they have already decided on.

Great new brand
Great new brand

A number of additional slides show some of the new generation's features and technologies, including an up-to-date information processing mechanism between the cache and the processing pipes, a very interesting XFR mechanism that promises to provide maximum maximum operating frequencies in systems where the cooling will be advanced and allowable, And accurate that they will also be able to receive Maximized depending on the environment in which the processor is run, and according to the tasks that will be activated at any given time.

See more details very soon.

zin4 zin5 zin3 zin2

Source images: videocardz.com


Tags

47 תגובות

  1. No doubt I'm going for the 500 dollar processor instead of the 1100 level processor is not bad at all and there is also an option in the 350 dollar ...

      1. This is not official e-information through the rumors from the sites but I tapped myself that all the rumors so far have become reality

  2. No, it's all leaks and speculation ...

    What's more and more as ZEN sees more and more performance, I think it will become more and more expensive, a company that likes to be the affordable but not sugary option.

    In the demo today he introduced a few single FPS over 6900K in BF1 and was worth it in two super heavy performance tests on the processor, all without turbo yet…

    1. Good morning Eliran we understood ... you are sick of intel probably Neydia too, let me say that the only loss is in your pocket your right as long as you do not hurt anyone and will say that without amd you would not buy a processor at 1100 dollars but at 2000 dollars. The amd market worker creates competition and good competition for consumers

  3. If these results continue while tests What will remain for Intel is the 10 processor cores that comes as 3 + affordable computers -

  4. My first processor was AMD K6 and maybe the next ..
    Provided that the CPU division does not nimble liars like their video card division ..

    Which is a very big condition.

    1. Meaning liar nerds like NVIDIA's 4GB by chance?
      Or maybe the 1080 cards that go up in flames?

  5. Some advertisements in the past were with promises and at the end of a fall.
    No motherboards yet
    No one knows yet what it will be

    A bit of a shock that this processor competes with Intel's 6900
    Intel 4 memory channels versus AMD's 2
    Unknown OC

    1. I'm not sure but I think you have a small mistake that the 2 channels do exist but the bony boards and down b350 and down ... I think x370 is indeed 4 channels

      1. True you have a big mistake!
        Intel 6900 series has 4 memory channels and not 2
        It's already basic

  6. Quote of Idan P.

    4 channels זיכרון Don't give anything away in the home market…

    Sound & Video Processing + UPSCALE Is It Anything?

    Processing more information is nothing?

    Check yourself

  7. First of all thank you very much for the quick article.
    What we would do without you, so thank you and appreciate!

    The only real friend here in the comments is SIM!
    It's all YAP YAP YAP!
    We will wait for unbiased tests!

    And I wish it was real with a rally price _ (but there are speculations that the price will not be rally)

  8. The processor looks impressive and promising.

    A little disappointment that there were no bouts of ADOBE.

    I also liked the example in the DOTA game, The guy is both playing and recording online at the same time, Intel's processor seems to falter near -RYZEN.

    If that's what RYZEN's multitasking performance looks like, it's not ...

    Jump to frame 47: 55

  9. Quote of Idan P.

    What's funny about that?
    I see great differences, you brought a link to the article and did not specify what to see in it?
    You have brought an article from 2015

    Is 40 lines this nonsense? Check out how much AMD gives?
    40 lanes (PCIe Gen 3.0) available

    Video Transcodin - The graph does not lie
    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,10.html

    Transcodig - the graph is not lying
    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,10.html


    Non-lying graph - Corona Ray Tracing

    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,10.html

    3DMark 06 CPU Test
    3DMark Vantage CPU Test
    Performance System Memory: Memory Read Test

    Testhttp://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,11.html

    the 4K Ultra HD resolution Dark Knight Rises trailer, this is a MP4 H.264 file and you can see that the CPU load is at roughly 30% with multiple cores showing utilization.
    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,15.html

    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,22.html

    1. They have an argument between the memories of 2 channels and 4 channels between each other there is no significant difference and this is according to the link by C. World. Bottom line of the same article can be waived bandwidth (4 channels to 2) for extra processor cores…

  10. Quote of ttuuoo

    Is 40 lines this nonsense? Check how many AMD Give?
    40 lanes (PCIe Gen 3.0) available

    Video Transcodin - The graph does not lie
    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,10.html

    Transcodig - the graph is not lying
    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,10.html

    Non-lying graph - Corona Ray Tracing

    http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,10.html

    40 PCI-E routers are also lacking in the home market.

    It does not matter whether it is 2013 or 2010 with the Triple Channel of the 1366, even then there was no difference and today in 2016 there is no difference.

    And obviously the graph does not lie, they have more cores and therefore the performance is higher not because of the memory bus.

    PS in the last comment showed above all doubt that you have no idea what you are talking about, so please next time do a little more ...

    If you want to compare memory performance you need to compare the same processor exactly with email and once with QUAD (as compared to what I brought) otherwise it is irrelevant.

    1. "40 PCI-E routers are also lacking in home market use.

      It doesn't matter if it's 2013 or 2010 with the Triple Channel 1366, even then there was no difference and today "
      Some nonsense you write
      And writes….

      "And obviously the graph is not lying, they have more cores and therefore the performance is higher not because of the memory channel. |

      Because they have a lot of memory and lots of cores, they perform in jobs that require memory and cores !!!

      Believe me if you stay here for 20 a year, you will not understand anything yet and continue to write no matter

  11. I needed a computer so I bought I3 6100 as a temporary computer until it comes out in the next few months, I will buy at the first minute and say hello and not see Intel with their tick and talk strategy.
    Someone has to sue them for the sheer amount of electronic junk they produce with all their Socks replacement every six months, what have they renewed since the USB3 and SATA6Gbs releases?

  12. Completely agree, the CPU market has been deserted like talking in the last 4-5 years, even during Intel's anesthesia with the Pentium 4 was not that boring, just one big yawn so the ZEN must put AMD back in the game for the benefit of us all.

  13. Being strong in multithread is hard, being strong in single-thread is much harder. I want to see the new processor in games known to be more dependent on this situation against Intel's 7700K.

    How will the first perform tasks? Hope to know as close as possible to launch, second? Maybe I'll get a little before the launch, do not expect any surprises

  14. Is it just me, or is the presentation of the processor with an Enoydia card and not AMD itself very puzzling?
    And no, don't tell me 'because the Titan X is just the strongest' is not a reason to tag your other competitor's monster performance.

    1. Drivers and they will give a punch in the next show is a war between Noydia and if enough ... Who will bring out the ace, last

  15. It's good to have competition in the market, imagine what would happen if there was exclusivity to one of the companies.

  16. This is interesting but as they said here it is hard to believe that AMD will give the same performance level at 50% cost. Beyond that, the final consumer price is often rising due to demand like we saw on the RX 480 and throughout the Noydia Pascal line.
    Most interesting to me is whether there will be a six to eight core supply around a budget of 350 dollars. Completing hardware encoding capability (H265) is also something that has been delayed for too long (amazing that every Chinese SOC has been using hardware and Intel and AMD have been delayed for so long).
    If we can get them performing at 70% cost (that's my bet) that would be amazing!

  17. Intel has no problem lowering price. In fact, AMD should be hard-pressed because its manufacturing defects are significantly higher than Intel's.

  18. At the processing level it looks great!
    The graphics don't buy me! Because this is one picture shading the other good lighting.
    Well done to AMD.

  19. When you do too much FINFET ……. So go up in flames!
    Why is Gigabyte the only one that changed the VRM and CHOKS system? _ (Because that's why its 1070 or higher tickets are the coolest in areas that are destined for pyromania and calamities such as menstruation and capacitors)
    Because Gigabyte learns from the past !!!
    While the others stay with the same hardware and require more…. Gigabit has spent money on engineering and modification of the production line, and also asks for slightly less than the others!

    And most annoying is that the others have removed the sensors from these areas so you don't know the exact temp .. (But there are benches for those who will look good)

    In FINFET, in the Israeli summer, the difference between the core and the VRM ranges from 17 degrees to 23 degrees!
    So anyone who sees his core about 82 degrees ... who knew VRM was coming to 100!

  20. ^ Blah blah blah blah blah blah blah… ..

    Pardon me, some gagging on nothing. The reason for the EVGA cards Certain Turned out to be the case is that EVGA did not put proper cooling on the voltage stabilizers, and that's it. A small piece of aluminum and a cooling pad solve any potential problem.

  21. Quote of Idan P.

    40 PCI-E routers are also lacking in the home market.

    I would like to add that it is useful if we say you have 2 cards in SLI that you just have to work on X16, and also 2 NVME cards.

    Don't know if it can still be called home use (probably not standard), but it's products sold in regular stores in Israel so let's say :)

  22. We will ignore the fact that there is no noticeable difference between working at X16 and X8 when it comes to video cards and look at the allocation of PCI Express channels to components:

    16 tracks / channels / routes to the first expansion slot for the video card (the PCI Express X16). I do not know what the source of these channels will be in the case of AMD, but in the case of Intel they are due to the processor and not the chipset (the processor is connected directly to the graphics card on 16 PCI Express channels, while the other components are connected using the PCI Express channels to the chipset).

    6 channels for SATA connections. I don't think anyone investing in a motherboard designed for a video card set would like to have fewer SATA connections.

    6 channels for USB connections. Just like in the case of SATA connections - you want as many USB connections as possible.

    It's already 28 channels. Even if you have 40 channels possible, you only have another 12 left, so you can't assign more 16 channels to another expansion slot for a video card.

    Now suppose you are interested in a panel with 2 M.2 connections, each with 4 PCI Express routes assigned. This means that you have fewer 8 tracks, meaning there are only more 4 channels left.

    Of course you wouldn't want a motherboard without some PCI Express X1 expansion slots, so you've already reached about 40 channels in total.

    Now suppose the AMD processor will provide 16 PCI Express channels to connect the main video card directly to it. If you want at least 6 SATA and USB connections plus 2 "fast" 2 connections, it's already 6 + 6 + 4 + 4 = 20 channels that require the chipset. All this without adding the PCI Express channels assigned to the expansion slots. Currently only Intel's most powerful chipset, the Z170 provides such a number of channels and all the "weaker" chip sets are less satisfying.

    Now suppose, according to the rumors (For example), The new AMD processor will provide 36 PCI Express channels. We will take 32 channels that will be connected to two video cards in X16 each. This leaves 4 PCI Express channels free. These channels must be used to connect the processor to the chipset (and most likely will be needed). This means that the chipset will need to provide at least 24 PCI Express channels on its part (20 channels as I explained earlier and another 4 to connect the processor to). Is it possible? Probably so. Would it be lucrative for most people? Probably not.

    Anyway, I guess we will soon find out exactly what will happen with AMD's new processors and chipsets and how good their performance will be.

  23. Quote of elikh

    By what you say, everyone who has it Processor With 40 lines, and two graphic cards, Can't run both with 16 lines? can not be…

    A. First of all - it depends on the design of the motherboard. Even if the processor gives you 100 PCI Express channels, it will not help if the board moves only 16 channels to the expansion slot intended for the graphics card.

    B. An ordinary user has no ability to affect the number of PCI Express channels assigned to each component (not upward creep). So even if the processor alone gives you 40 PCI Express channels and the chipset gives you another 20 PCI Express channel, that does not mean that you yourself can determine how many PCI Express channels will be allocated to each expansion slot and each M.2 connection.

    Just for clarification - this is the case today, to downsize when it comes to Intel processors:

    http://ark.intel.com/compare/81761,90590,90592,90595,90591

    The H110 chipset provides 6 PCI Express channels, the B150 chipset provides 8 PCI Express channels, the H170 chipset provides 16 PCI Express channels, the Z170 chipset provides 20 PCN Express channels and the XXN XX chipset

    About processors:

    http://ark.intel.com/compare/88188,93338,88191,93339,88195,94189,94188,94196,94456

    Skylake processors provide 16 PCI Express channels and Broadwell-E processors designed for the high end home market (40) provide PCI Express channels with the exception of the Core i7-6800K processor which provides "only" 28 PCI Express channels.

    Just be clear - the 100 series chipsets (H110, B150, H170 and Z170) use the Skylake processors, while the Broadwell-E processors use the X99 chipset.

    Assuming (reasonably) that the better performance will give a video card that is directly connected to the processor (and not one connected through the chipset) then to be able to play two video cards in X16 this requires that only the processor alone will provide 32 PCI Express channels. The only processors that meet this requirement, as of today, are the Broadwell-E processors (again, for the Core i7-6800K processor), which are quite expensive. The problem is that the chipset for them, the X99, only provides 8 PCI Express channels, so if out of a total of 48 PCI Express channels, 32 channels will be assigned to the graphics cards, 16 will leave channels to be shared among all other motherboard components (and especially the M. connections). 2 and card expansion slots).

    Quote of none77

    Shi, to my understanding sata, USB Unrelated to pcie channels, other pciex4 slot ports, etc. Yes.

    right. I'll elaborate a bit more - in Intel's chipset, at least those from the 8 and 9 series, and also those from the 100 series, there's something called I / O Port Flexibility. This actually limits the number of connections that can be assigned to each type, including SATA connections, USB connections, M.2 connections, and expansion card connections. The Intel chipset 100 series actually has up to such 26 ports (depending on the chipset model) that you can assign between USB 3.0 (up to ten) connections, SATA connections (up to six), a wired network card connection (I'm not sure if Limitation to one network card only) and 20 PCI Express channels. Of course this means that there must be some overlap between the components because 10 + 6 + 1 + 20 = 37> 26. For example - Ports 7 to 10 can be assigned (each) to a USB 3.0 connection or PCI Express connection, while for the 19 port, you must choose between a SATA connection, a wired network card, or a PCI Express connection. In practical terms, this means that the SATA connections, as well as the wired network card, come at the expense of the number of available PCI Express channels, and if you want to realize more than 6 USB 3.0 connections on the motherboard will necessarily come at the expense of the number of PCI Express channels. Of course, connections that require 2 or 4 PCI Express channels (which are actually SATA Express and M.4 connections) actually require 2 or 4 ports. All of this is on page 26 in the Intel chipset booklet on page 26.

    So right, you are right that the SATA and USB connections do not use PCI Express channels, but they usually come at their expense.

  24. Thanks for the detailed responses.

    Quote of smalul

    A. First of all - it depends on the design of the motherboard. Even if the processor gives you 100 PCI Express channels, it will not help if the board moves only 16 channels to the video slot expansion slot.

    I suppose extensions in general and X99 in particular are designed so that the lines in the panel are able to provide the processor with the information from the expansion slots, are there exceptions?

    Quote of smalul

    The problem is that the chipset for them, the X99, only provides 8 PCI Express channels, so if out of a total of 48 PCI Express channels, 32 channels will be allocated to the graphics cards, 16 will leave channels to be shared among all other motherboard components (and especially the M. connections). 2 and card expansion slots).

    So what I understand from you is that if there are 2 video cards (32 lines) + 6 lines for SATA + 6 lines for USB, we basically only stayed with 4 lines for one NVME direct processor?

    But is it possible for two NVMEs that the other will go through the chipset?

    (Unless you, of course, give up the 2X16 for video cards).

    If Z170 provides additional 20, then how do they divide? Again 12 to I \ O? And another 8 for more expansion cards?

  25. Ummm smalul

    A. Not every SATA or USB sits on the channel…

    On USB, usually 2-4 sit on a controller that is connected to one channel and several other to Zipst.

    In SATA usually 6 come from Zipst and do not take any PCI-E connection except through DMI / QPI on Intel.

    When you say that the processor has 40 channels it is only For extensions like M.2 (and not always, sometimes it's via Southern Controller) and PCI-E cards.

    More to the point, today's southern controllers have so many other PCI-E connections that not even a few PCI-E X1 connections to expansion cards are connected to the southern bridge and not to the processor.

    When you say 6700, let's say 16 has pci-e only for expansion cards and usually video cards in this case.

    Everything else (20) includes M.2 coming on the southern bridge:

    f372-1a.gif

    The X99 has the same 40 lines that were previously referred to as expansion cards but because of the few PCI-E connections from the Southern Bridge PCI-E X1 are sitting on them and not on the southern bridge.

    X99-Block-Diagram_w_600.png

    In the AMD that the guy cried out earlier that there are only 16 lines, there are actually 44 PCI-E2.0 lines, which is ancient Zipst

    990fx-diagram.png

    When 32 is assigned to a main expansion card by video cards in 99% of the time.

    Bottom line if in zen it will remain the same only if PCI-E3.0 is more than Z170, does anyone here think that Z170 does not provide enough connections to the home market?

  26. Quote of elikh

    I suppose extensions in general and X99 in particular are designed so that the lines in the panel are able to provide the processor with the information from the expansion slots, are there exceptions?

    I also guess that's what actually happens. In fact, there is no real incentive to do this with other chip sets, including the standard home chip sets. If I remember correctly ASRock has one (at least) board for the 1150 bracket that takes the 16 PCI Express channels provided by the processor and splits them into two groups - 8 channels will be used for the graphics card and from the remaining 8 channels 4 is assigned to the M.2 connection, so even in such a panel There will be an M.2 connection that has 4 PCI Express channels (this was not common at all in 1150 - until then the M.2 connections usually only received 2 PCI Express channels). You can also go in a different direction and add a kind of "channel multiplier" that makes The 16 channels provided by the processor to 32 channels, but boards that do this are usually very expensive.

    Quote of elikh

    So what I understand from you, if there is 2 graphic cards (32 lines) + 6 lines to SATA + 6 lines to USB, basically we only stayed with 4 lines to NVME one direct to the processor?

    But is it possible for two NVMEs that the other will go through the chipset?

    I don't really know what is going on electronically on the motherboard, but I don't think there is any motive to create a motherboard where one expansion slot will receive 16 PCI Express channels from the processor and a second expansion slot that will receive 16 channels PCI Express chipset. In fact, CrossFire arrays that are not based on the ZXX family of chipset work just like this - 16 channels arrive at the first graphics card from the processor and 4 additional channels arrive at the chipset graphics card. Of course in this case the video card array works on X4 and not on X16, but I don't think there is any reason to give the other expansion slot to the video card more channels that result from the chipset.

    Similarly, I do not think there is a reason why M.2 connections should be directly connected to the processor.

    Quote of Idan P.

    Not all SATA or USB Sitting on the channel…

    On USB, usually 2-4 sit on a controller that is connected to one channel and several other to Zipst.

    In SATA usually 6 come from Zipst and do not take any PCI-E connection except through DMI / QPI on Intel.

    I explained it - it's not exactly a channel, it's a port and it's not an equal thing.

Back to top button
Close
Close