New Upcoming ATI/AMD GPU's Thread: Leaks, Hopes & Aftermarket GPU's Part 2

Discussion in 'Videocards - AMD Radeon' started by WhiteLightning, Sep 28, 2018.

  1. Agonist

    Agonist Ancient Guru

    Messages:
    2,857
    Likes Received:
    232
    GPU:
    XFX 5700XT Raw II
    You running triples? I sim race on a single 35 inch 3440x1440 100hz panel and my 5700xt does very well. I went from triples to a single monitor.
    Got tired of the problems with eyefinity. It used to be pretty streamlined and worked well. Now its hot trash and a pain in the ass.
     
  2. Chastity

    Chastity Ancient Guru

    Messages:
    2,544
    Likes Received:
    814
    GPU:
    Nitro 5700 XT
    I've been analyzing NV's play with Ampere, and how they are literally fscking over their customers. They come out with a baseline model, and a Titan replacement, both are nerfed in some form. (3080 with memory, 3090 with poor rasterization bump from 3080) Both units are not full die units, so the chips aren't at full capacity. They did this intentionally so that they can have a play after RDNA2 releases, and bet on the fact they can really make a full Ampere chip into a hotrod.

    So their early adopters get jebaited. They did this with Turing, and the Super releases. They jacked up the pricing more than necessary to deal with cost increases. The performance increases are always about 25% from previous models. Their business model is not about innovation and improving PC gaming; it's about delivering a product and making you reliant upon them. Pushers do this.

    What a miserable, evil, and greedy company. I am so glad that they will never see my money ever again. I could care less about who's the better value or performer. The utter contempt for their customer base is downright appalling.

    I honestly feel sorry for whomever uses their products, because their only purpose is to get milked, and to accept what is offered, with Marketing making sure you feel good about it and want it, and nobody else can replace them..
     
    Last edited: Sep 25, 2020
  3. Morax

    Morax Member Guru

    Messages:
    142
    Likes Received:
    40
    GPU:
    Asus RX Vega 64
    Not to defend Nvidia, or blame AMD, but I fully expect AMD to release a higher clocked HBM-equipped big Navi at a later date, like early next year, as a response to the 3080 Ti/Super/whatever.

    But for now I'm really curious about what AMD is getting on the table end of October. And "where exactly" Navi 21 will turn out to be at "somewhere between the 3070 and 3080".
     
  4. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    16,979
    Likes Received:
    2,516
    GPU:
    AMD S. 5700XT Pulse
    https://www.phoronix.com/scan.php?page=news_item&px=AMD-Hiring-Radeon-Rust

    From:
    https://www.reddit.com/r/Amd/comments/iz8ege/amd_is_hiring_to_work_on_new_radeon_driver/

    Which also had this post explaining it.

     
    OnnA, lukas_1987_dion and Undying like this.

  5. lukas_1987_dion

    lukas_1987_dion Master Guru

    Messages:
    461
    Likes Received:
    40
    GPU:
    RTX 2080 Ti AMP! OC
    I hope RX 6900 XT (or 6800 XT?) will have HBM2 VRAM, just GDDR6 with 256 or 192 bit bus is a little worrying when it comes to memory bandwidth, or if it's really GDDR6 let it be at least with 384 bit, AMD states that the card will be perfect GPU for 4K gaming, so it would make sense I guess.
    Either way I'm very excited and I'm curious what final specs will be.

    Damn, why is time moving so slowly?! Still more than a month of waiting :p
     
    Last edited: Sep 25, 2020
  6. JonasBeckman

    JonasBeckman Ancient Guru

    Messages:
    16,979
    Likes Received:
    2,516
    GPU:
    AMD S. 5700XT Pulse
    Could also be varying info to different sources and such to prevent leaks of the real details or track what gets leaked and from where.
    HBM shows up as 256 bit too I believe so there's that but maybe for a future RDNA GPU like RDNA3 next year but it could be used.

    Isn't that how they called the Fury 4096-bit and Vega 2048-bit since the Fury was 4x 256-bit HBM chips (1024-bit) and then four of those total for 4096-bit in full whereas Vega only had two stacks so it got 2048 bits.
    Big bus width but the total bandwidth and speed was improved on Vega and Navi improves on it further over Vega even with GDDR6 although it's back to regular 256 bit bus width now. :D

    EDIT: What's next anyway here, 5nm I think it was no it's 6nm but it seems to be more about overall density here if I got that right from what I was reading yesterday about TSMC, Zen4 and RDNA3 early rumors.

    Well about a bit over a week until the Zen3 details from AMD and then the end of October for Navi2 which will hopefully clear a few things up about these.
     
  7. Ryu5uzaku

    Ryu5uzaku Ancient Guru

    Messages:
    6,899
    Likes Received:
    161
    GPU:
    5700 XT UV 1950~
    Well tbh full 102 ampere is max 2SM more then 3090 so eh not gonna make a difference. While they could release 100 ampere I doubt it as it's a bit different.
     
    lukas_1987_dion likes this.
  8. PrEzi

    PrEzi Master Guru

    Messages:
    373
    Likes Received:
    250
    GPU:
    Sapphire Vega 64 LC
    Some nice additional information about RDNA2 :
     
  9. Valken

    Valken Ancient Guru

    Messages:
    1,561
    Likes Received:
    125
    GPU:
    Forsa 1060 3GB Temp GPU
    I am waiting and the only benchmark I really care about is 4K RT performance. If AMD has near 60 FPS sustained, with very little difference to min frames, I would give up 100 FPS of peak raster performance.
     
    Maddness likes this.
  10. pharma

    pharma Maha Guru

    Messages:
    1,438
    Likes Received:
    316
    GPU:
    Asus Strix GTX 1080
    Thanks for posting! I missed this one.
     

  11. PrEzi

    PrEzi Master Guru

    Messages:
    373
    Likes Received:
    250
    GPU:
    Sapphire Vega 64 LC
    If we are talking about GN then you definitely need to see this. I laughed my @ss off :D
     
    lukas_1987_dion and fantaskarsef like this.
  12. Saabjock

    Saabjock Master Guru

    Messages:
    314
    Likes Received:
    42
    GPU:
    PNY GTX1070XLR8
    I'm running a single 1080p G-sync/ FreeSync compatible LG 144Hz IPS panel along with a Lenovo Explorer WMR headset. I was running an MSI GTX1080Ti Duke and it was smooth in every racing sim...even with WMR.
     
  13. OnnA

    OnnA Ancient Guru

    Messages:
    11,790
    Likes Received:
    2,720
    GPU:
    Vega XTX LiQuiD
    pharma likes this.
  14. OnnA

    OnnA Ancient Guru

    Messages:
    11,790
    Likes Received:
    2,720
    GPU:
    Vega XTX LiQuiD
    PS5 IO System to Be ‘Supercharged’ by Oodle Texture, Bandwidth Goes Up to 17.38GB/s
    You might remember that the official specifications listed the console's IO bandwidth at 5.5GB/s with uncompressed data and 8-9GB/s with compressed data.
    However, it looks like the partnership with Oodle will allow the PS5 to go much further when it comes to bandwidth peaks with compressed data. We already knew that the PlayStation 5 would support the Kraken hardware decoder, but the folks at Oodle have now unveiled that Sony also licensed Oodle Texture technology, which is optimized for compressing GPU textures.

    Oodle Texture is a software library that game developers use at content creation time to compile their source art into GPU-ready BC1-7 formats. All games use GPU texture encoders, but previous encoders did not optimize the compiled textures for compression like Oodle Texture does. Not all games at launch of PS5 will be using Oodle Texture as it's a very new technology, but we expect it to be in the majority of PS5 games in the future. Because of this we expect the average compression ratio and therefore the effective IO speed to be even better than previously estimated.

    Kraken is a powerful generic compressor that can find good compression on data with repeated patterns or structure.
    Some types of data are scrambled in such a way that the compressability is hard for Kraken to find unless that data is prepared in the right way to put it in a usable form.
    An important case of this for games is in GPU textures.

    [​IMG]

    Oodle Kraken offers even bigger advantages for games when combined with Oodle Texture. Often the majority of game content is in BC1-BC7 textures. BC1-7 textures are a lossy format for GPU that encodes 4x4 blocks of pixels into 8 or 16 byte blocks.
    Oodle Krak

    en is designed to model patterns in this kind of granularity, but with previous BC1-BC7 texture encoders, there simply wasn't any pattern there to find, they were nearly incompressible with both Zip and Kraken.
    Oodle Texture creates BC1-7 textures in a way that has patterns in the data that Kraken can find to improve compression, but that are not visible to the human eye.
    Kraken can see that certain structures in the data repeat, the lengths of matches and offsets and space between matches, and code them in fewer bits. This is done without expensive operations like context coding or arithmetic coding.

    According to Oodle, the previous 8-9GB/s IO bandwidth figure was provided by Sony by multiplying the 5.5GB/s peak bandwidth of the PS5 SSD for the standard compression ratio of 1.45/1 or 1.64/1.
    However, when factoring in both Oodle Kraken and Oodle Texture, the developers of the technology have seen an impressive compression ratio of 3.16/1 for a texture set in a recent game;
    this would translate into an IO bandwidth peak of 17.38GB/s, as mentioned in the headline.

    Of course, that may not be typical, and not all PS5 games will necessarily take advantage of this technology to the fullest.
    That said, it does sound like the PlayStation 5 is equipped to remain the leading gaming architecture for quite some time yet when it comes to IO.
     
  15. OnnA

    OnnA Ancient Guru

    Messages:
    11,790
    Likes Received:
    2,720
    GPU:
    Vega XTX LiQuiD
    AMD Navi 21 is Sienna Cichlid

    The speculation is over, Sienna Cichlid is indeed Navi 21 and Navy Flounder is Navi 22. This is all confirmed by AMD through ROCm Developer Tools source code.
    Whether these codenames refer to the specific products (such as Radeon Pro), we don’t know yet.
    We do at least know for sure that these codenames are used internally by AMD open-source software developers in regard to Navi 2X series of GPUs.

    [​IMG]

    AMD ROCm Developer Tools (gdb/doc/gdb.texinfo)

    As explained by the developers themselves, those codenames have no meaning, they are generated from a random selection of word lists.
    We believe this is a technique to derail the interest of everyone looking for RDNA2 GPUs information. After all, it took nearly 3 months to finally see a single line of code connecting the codenames to GPUs.

    The nature of open-source software allows everything to track the progress of next-generation Radeon graphics.
    Historically they were always added weeks ahead of launches, so whenever something happening in the code, we knew new GPUs were coming.

    Just three days ago we reported on a third codename had appeared in AMD OpenGL driver (Mesa) code.
    AMD Dimgrey Cavefish is more than likely Navi 23, a third GPU in the GFX103X family, this, however, was not confirmed in the code (like Navi21 and Navi22), but rather deduced from available data.

    Note:
    So we have 3 SKUs for RDNA2 (GFX 1030, 1031 & 1032)
    6900 or Arcturus
    6800/6700
    6600/6500

    -> https://videocardz.com/newz/amd-sienna-cichlid-confirmed-as-navi-21-navy-flounder-is-navi-22

    [​IMG]
     
    lukas_1987_dion and pharma like this.

  16. Saabjock

    Saabjock Master Guru

    Messages:
    314
    Likes Received:
    42
    GPU:
    PNY GTX1070XLR8
    AMD has another 'golden' opportunity given the 'cock up' by some Nvidia AIB partners.
    I have to admit I was a bit concerned for AMD' sale numbers...given the long lead Nvidia and their partners had to market
    The reported 'drop to desktop' issue is suspected to be electrical noise at elevated boost frequency.
    It is suspected to be as a result of power filtering of lesser quality than the OEM board.
    Given this information and the fact that some of these were top end manufacturer's...in some cases charging much more...
    How likely will you be to buy AIB partner offerings on the AMD side versus the reference cards?
     
  17. Goiur

    Goiur Master Guru

    Messages:
    903
    Likes Received:
    268
    GPU:
    RTX 2070 by G3D
    Just wait 1-2 weeks before buying, let people beta test the launch batch.

    On nvidia side some aib maybe under spec their cards, or the spec was right and its boost ¿2.0? at fault pushing too high.
     
    OnnA likes this.
  18. Undying

    Undying Ancient Guru

    Messages:
    13,819
    Likes Received:
    3,058
    GPU:
    Aorus RX580 XTR 8GB
    Yeah cant wait to see how AMD screw up their launch too.
     
  19. Maddness

    Maddness Maha Guru

    Messages:
    1,376
    Likes Received:
    579
    GPU:
    3080 Aorus Xtreme
    Hopefully not. They actually may have a great chance here, but they have to take advantage of it.
     
    Saabjock, OnnA and Undying like this.
  20. OnnA

    OnnA Ancient Guru

    Messages:
    11,790
    Likes Received:
    2,720
    GPU:
    Vega XTX LiQuiD
    We really need HBM2 Part to be Real.
    It will be ATI selling point for me, i don't wanna downgrade... to mere GDDR ;)
     
    lukas_1987_dion likes this.

Share This Page