New Upcoming ATI/AMD GPU's Thread: Leaks, Hopes & Aftermarket GPU's

Discussion in 'Videocards - AMD Radeon' started by OnnA, Jul 9, 2016.

  1. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV
    AMD/ATI TIRAN GPU !

    [​IMG]
     
  2. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV
    Some Editorial an 'glimpse into the future' ;)

     
  3. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV
    AMD To Utilize PCIe 4.0 Standard With Next Generation Radeon Instinct Vega 20 GPUs

    According to latest Linux drivers, it looks like graphics engineers at the Radeon Technology Group are working to support PCIe 4.0 speeds for upcoming graphics cards.
    The new interconnect will be doubling the bandwidth over PCIe 3.0, offering a 16 GT/s bit rate from 8 GT/s, and total bandwidth of 64 GB/s (x16 interface) compared to 32 GB/s (x16 interface) on PCIe 3.0.

    This aligns with some of the leaked roadmap and slides we have seen in the past which revealed PCIe 4.0 xGMI (inter-chip global memory interconnect).
    The leak also mentioned that PCIe 4.0 link would be available with Vega 20 graphics cards that arrive in late 2018. AMD had also previously confirmed the addition of a new high speed I/O with the 7nm Vega parts to enable more general purpose HPC clusters. That was most definitely referring to the PCIe 4.0 interface.

    Based on the 7nm process node, the new Vega 20 GPUs will offer 2x the density since AMD is packing more stuff in a smaller die than Vega 10, it will also be twice as much efficient due to architectural and process maturation, and finally, AMD has planned out several products based on 7nm in the pipeline.

    [​IMG]

    [​IMG]

    Yup, our ATI 7970 was the first :p PCIe 3.0 GPU in the World.

    [​IMG]

    More -> https://wccftech.com/amd-vega-20-radeon-instinct-pcie-4-support/?tqid=hvTkb2M1BlcBeiED9SP.Zlva_96L_liI8RDQq4U$
     
    Last edited: Jun 30, 2018
  4. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV

  5. Maddness

    Maddness Ancient Guru

    Messages:
    2,440
    Likes Received:
    1,739
    GPU:
    3080 Aorus Xtreme
    7970 was a fantastic GPU. I'm hoping for more of the same when Navi launches.
     
  6. warlord

    warlord Guest

    Messages:
    2,760
    Likes Received:
    927
    GPU:
    Null
    ^^ We will never see again a fight like 7970 vs gtx 680 it was a lucky bet from AMD gcn plus pciex 3.0 and it went fine wine after a year or two. Even if vega refresh of 2019 goes pciex 4.0 it could never touch or beat nvidia's new opponent. They are always number #2.
     
  7. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV
    [Editorial]
    ^^ Yet we have 290X Vs 780 goes easy win for Hawaii then was refresh with 8GB and crushing victory (Fine wine still i presume ;))
    Then we have a good fight between Fiji vs Maxwell yet another good trade blow & Fine wine hits us again.
    And now we have Young Vega uArch and it's still pushing some good FPS in all Games (i do not count my XTX LiQuiD because it's an different Level of GPU)
    For ~389-420€ you can get Vega 56 AIB and Play your favorite Games easy....

    Is it really number two? IMO it's not, of course it is when one is interesting in Pushing 3Dmarks :D
    But when it come to Playing Games, it's good GPU and along with Free FreeSync Monitor it's an Easy Win for any Gamer....

    Also we can have some good Quality HDR/SDR FreeSync Monitors for budged ~250-498€
    I mean IPS with =>70Hz and 10Bit (Quality)
    And Yes FreeSync is Real Game changer for any gamer.

    =
    As for Best ATI/AMD GPU -> Vega 64 XTX LiQuiD
    We have an Real Beast of an New Tech & Power House in all good means.

    Here you have Vampyr in UE4 ;)
    Normal Vega64 is ~50-66
    My Vega Not even OC, TBF De-OC'ed ;) from 1668-1750 (Yes 1750 is an Vega XTX Default, AI Catalyst Boost it's my name for this tech) to 1720Mhz i have Chill On and Game is Hoovering at Chilled 63-70Hz/FPS
    When tested i have ~77FPS Average ! (With reshade etc.)

    And some Heavy hiter CryEngine Game (no shortcut in render here ;))

    [​IMG]

    [​IMG]

    There is Fun & PowaH in Vega for sure :D
    It's like always IMHO, im not done Testing my Beast Yet, but i need to do my Own benchmarking in alll my Games using as always FRAPS & FLACalculator :D
     
    Last edited: Jul 1, 2018
  8. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV
    AMD GPU Rumors: Polaris 30 12nm GPUs in Q4 2018, Navi GPU Based Mid-Range Cards Arrive Before High-End HBM2 Variants

    AMD has a strong roadmap outlined for their next-generation products but on the GPU side, and particularly the gaming segment, things aren’t that clear. If we look at the recent roadmaps, there have been various GPUs listed including their upcoming 7nm parts but they are being aimed at the server market first which leaves users wondering when we would see the first 7nm gaming GPUs.Now I want to point out that most of this information is based on rumors so don’t take it as fact.
    The details come from Chiphell user WJM47196 who has a history of posting AMD CPU and GPU rumors. The same user had posted details on Ryzen and Ryzen Threadripper CPUs, most of which turned out to be close to the real thing. Even still, we would advise you to treat this as a rumor for now until more details are confirmed.

    AMD Polaris 30 GPUs in Q4 2018

    So let’s start off with the rumors, there are three GPUs discussed in the post which start with AMD Polaris 30. The AMD Polaris 30 GPUs are completely unheard of and not even part of any roadmap we have seen so far.
    The Polaris 30 GPU is said to be the third iteration to the high-end Polaris line which initially shipped as Polaris 10 in 2016 followed by Polaris 20 in 2017. Polaris 30 based GPUs are said to use the latest 12nm FinFET process and deliver a 15% performance jump over Polaris 20 GPUs.

    Now I’ll come back to Polaris 30 not being in the roadmap at all, you know what else wasn’t in the AMD roadmap till launch? The Polaris 20 GPUs.
    AMD Polaris 20 was not mentioned in any roadmap and eyes were on the Radeon RX Vega launch in 2017. AMD Radeon RX Vega never came to the mainstream desktop segment and Polaris 20 was launched offering an intermediate solution under the RX Vega 56. With Polaris 20, AMD fine-tuned the architecture with better process optimizations and aggressive clocks with similar competitive pricing.

    RX Vega still hasn’t made an official entry in the discrete mobile segment and the only entry-level chips we know of are found in Raven Ridge APUs or other configurations that are custom designed for Apple iMac’s or Intel’s 8th generation Kaby Lake-G parts.
    If Vega is that hard to produce for the mainstream segment, then a new iteration of Polaris graphics cards for the mainstream segment may be arriving soon but that would be a very disappointing move by RTG for the gaming market.

    AMD 7nm Navi For Mainstream Users First, High-End HBM2 Variants Later

    Now coming to Navi, the next-gen we have all been waiting for. Succeeding Vega in 2018 as was planned, it looks like the gaming cards are actually much far away than we have imagined.
    Our exclusive report which discusses AMD’s semi-custom ventures, designing Navi primarily for the Sony PlayStation 5 sheds more light on this.

    The rumor mentions that the first Navi graphics cards for the desktop market will arrive between 2018-2019. The first cards would be aimed at the budget segment and will be based on the 7nm process.
    The high-end variants will arrive much later and would be using HBM2 memory. It’s really interesting that HBM2 is mentioned only for the high-end variant which makes me think that the mid-range cards can utilize GDDR6 memory which is in full production from Micron and Samsung right now.
    The mid-range cards are expected to replace Polaris once and for all and their launch this late might explain why AMD would do another stop gap Radeon RX Polaris lineup since Navi will be coming really late to desktop PCs.
    It also aligns with our report which mentions:

    • Navi 10 will be the first Navi part to arrive and will be landing sometime in 2H 2019 or early 2020, ludepending on a couple of factors. The performance level of this part will be equivalent to Vega and it will be a small GPU based on 7nm.
    • Navi 14 will follow Navi 10 soon after.
    • Navi 20 is going to be the true high-end GPU built on the 7nm node and as things stand right now, you are tentatively looking at it landing sometime around 2020 – 2021.
    We also reported that Navi GPUs will be using a traditional monolithic design over an MCM (Multi-Chip Module) design for gaming variants due to design complexity, cost factors and proper OS / Software based optimizations for multi-chip and multi-GPU support.

    AMD 7nm Vega 20 With 16 GB HBM2 in Q1 2019

    The last detail mentioned in the forum is regarding a particular 7nm Vega GPU for gamers. The product is said to be still in consideration and may or may not arrive in the gaming market. All of that depends on HBM2 availability and pricing.
    The card is said to use all of the 7nm advantages that Vega 20 will include such as faster clocks, much efficient operation, and a 4096-bit interface. This card will be featuring up to 16 GB of HBM2 instead of 32 GB like the server variant.
    The reason again is the high price of HBM2 and 16 GB would be more than enough for gamers.

    It is stated that the card seems very unlikely to happen since this card would have to be priced way too high otherwise AMD would be shooting themselves in the foot if they try making it competitive against NVIDIA’s offerings at that time.
    The cost of producing such a card would be so high that it will eventually give little or no return. But then again, AMD has promised 7nm Radeon RX to gamers for next year so whether that would be Vega or Navi based remains to be seen.
    I’d still vote for the latter since Navi goes with AMD’s official plans.
     
    warlord likes this.
  9. warlord

    warlord Guest

    Messages:
    2,760
    Likes Received:
    927
    GPU:
    Null
    Interesting rumor for a better mid range like polaris 30. I am curious about its price. Perhaps it is this for me or gtx 1160? because i have only 2560x1080@72hz monitor. More power is useless for me never gonna be used unless I buy new monitor. Either way my 390x has insane power consumption to be honest for what it offers. I hope polaris 30 will be a lil' more eco with consumption/temps.
     
  10. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV

  11. Komachi

    Komachi Member

    Messages:
    29
    Likes Received:
    10
    GPU:
    Radeon Pro Duo Fiji
    Last edited: Jul 3, 2018
    user1 and OnnA like this.
  12. Komachi

    Komachi Member

    Messages:
    29
    Likes Received:
    10
    GPU:
    Radeon Pro Duo Fiji
    This is My opinion. but It is almost matches.
    GFX804 = Polaris. GFX900/901 = Vega 10/Vega 10+HBCC.
    GFX902 = RR iGPU. GFX903 = Vega 11 or RR+HBCC?.
    GFX904 = Vega 12. GFX905 = Vega 11 or Vega 12+HBCC.
    GFX906/907 =Vega 20/Vega 20+HBCC. GFX1000/1001 = Navi 10/Navi10+HBCC?
     
  13. user1

    user1 Ancient Guru

    Messages:
    2,782
    Likes Received:
    1,305
    GPU:
    Mi25/IGP
    Good to see some new stuff landing in the driver.

    From what I have gathered with the gfx900 and gfx901 , Hbcc works on both, but the 900 variant pulls more power (power management bugs perhaps?)
    I don't think gfx900 was supposed to ship on commercial cards , I would guess they are engineering samples effectively, only very early vega cards have the gfx900 chip version
     
  14. OnnA

    OnnA Ancient Guru

    Messages:
    17,963
    Likes Received:
    6,824
    GPU:
    TiTan RTX Ampere UV
    Hmm, can you elaborate on the mater?
    What is this name sheme: gfx900,902,904 etc.
    Size of an chip?
     
  15. Komachi

    Komachi Member

    Messages:
    29
    Likes Received:
    10
    GPU:
    Radeon Pro Duo Fiji
    I think GFX900 is HBCC off ver (in Driver). if you active HBCC. Vega 10 has two ID (GFX900/901)

    URL1:https://compubench.com/device.jsp?benchmark=compu20d&os=Windows&api=cl&D=AMD+Radeon(TM)+RX+Vega&testgroup=info
    URL2:https://gfxbench.com/device.jsp?benchmark=gfx40&os=OS X&api=metal&D=AMD+Radeon+Vega+Frontier+Edition&testgroup=info

    PS:I`m sorry that my english is so bad.
     
    Last edited: Jul 4, 2018

  16. Komachi

    Komachi Member

    Messages:
    29
    Likes Received:
    10
    GPU:
    Radeon Pro Duo Fiji
    GFX9 is Vega Family. GFX90X is based Vega.
    Naming? AMD Changed Islands codename to GFX Codename.
    GFX90X:X is probably design start time.
    Same thing. Vega "10"/Vega "11"/Vega "12" etc.

    PS:I`m sorry that my english is so bad.
     
    OnnA likes this.
  17. user1

    user1 Ancient Guru

    Messages:
    2,782
    Likes Received:
    1,305
    GPU:
    Mi25/IGP
    Komachi likes this.
  18. Fox2232

    Fox2232 Guest

    Messages:
    11,808
    Likes Received:
    3,371
    GPU:
    6900XT+AW@240Hz
    GCN was not lucky bet. They had VLIW4 and GCN on table and VLIW4 has been picked. Can you imagine difference if they did invest directly into GCN?
     
    warlord likes this.
  19. warlord

    warlord Guest

    Messages:
    2,760
    Likes Received:
    927
    GPU:
    Null
    Hm, interesting I didn't know that detail. So they could fare much better and have more time with GCN in hands if they stopped pushing VLIW. But anyhow one way or another they would get into a point GCN gets EOL. I hope they have ready something great like this again.
     
  20. -Tj-

    -Tj- Ancient Guru

    Messages:
    18,103
    Likes Received:
    2,606
    GPU:
    3080TI iChill Black
    No need to drop gcn, it's as advanced as it can be, they started with gcn just in the right time.

    Now they can focus on its power efficiency more and add extra ROPs and more cores, its mostly this holding its raw power back. If Vega64 had 96 ROPs it would crush even titanV, but power and thermals didn't allow it.
     
    warlord likes this.

Share This Page