AMD Catalyst 15.15 (Official Radeon 300 Launch Driver)

Discussion in 'Videocards - AMD Radeon Drivers Section' started by d2kx, Jun 16, 2015.

  1. theoneofgod

    theoneofgod Ancient Guru

    Messages:
    4,677
    Likes Received:
    287
    GPU:
    RX 580 8GB
    Using this logic. 95% of people that buy GPU's don't have a clue they're rebrands.
     
    Last edited: Jun 27, 2015
  2. Krelianz

    Krelianz Guest

    Messages:
    112
    Likes Received:
    0
    GPU:
    Sapphire R9 290x
    he makes a number and asks others to prove him wrong when he doesn't provide any evidence to support his numbers, stupid guy has never heard of burden of proof, but what can you expect froma moron whose argument revolves about made up numbers.
     
  3. MaxBlade

    MaxBlade Master Guru

    Messages:
    907
    Likes Received:
    13
    GPU:
    980Ti/970/390/+ more
    "The remaining 95% of users just never upgrade anything during the lifetime of their PCs -"

    haha loved this.. I'll take "Wild Guess's" for a 100 Alex
     
  4. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    If you see the sales numbers since Holidays 2014-now, AMD had a huge plummet in GPU sales since NVIDIA presented their 900 series. Their total revenue went down almost 20% between Q3 and Q4 2014, and a further 20% in Q1 2015. What happened then was the 900 series. What that means is that they have a ton of unsold Hawaii. What do you do about your unsold Hawaii? You take your good memory partner Hynix who wants to switch to HBM and DDR4 and has a ****load of stock of GDDR5, and you make them an offer to get that memory stock off their hands in exchange for your HBM patents, at the same prices as you were getting the cheaper Elpida memory.
    Thus the 390x is born. 8GB of Hynix hi-speed VRAM and Hawaii with a different bios. There is NOTHING changed in the chip, literally nothing. People with Elpida memory 290x are running the 390x bios.

    No, you are wrong.
    The whole difference is in the driver, and then maybe on the extra speed on the memory.
    First of all, you are the one who pulls numbers out of his ass. I linked the Steam hardware survey that showed that 80% of the Steam user base (which is larger than 100 million since last year) has discreet GPUs.
    NVIDIA's whole business is selling discreet GPUs as upgrades. Look at the NVIDIA sales numbers from Q1 2015.
    NVIDIA's whole growth pattern is based on people who upgrade the same way that you say they don't. They expect a good percentage of 10 million users to make the jump in three months alone. Next time, please bring some actual numbers if you have them, and don't try to generalize the situation you see around you. The world is a big place.

    A multi-billion dollar company growing like crazy by doing exactly what you believe doesn't happen, invalidates all the arguments you are making.

    The only article I've seen was from Valve's Rich Geldreich about the state of the OpenGL drivers from all vendors, a couple of years ago. After that nothing really. If you have any links, please share. I haven't seen any kind of confession from AMD about the state of their driver anywhere.

    That they do, because otherwise they would have to be called rebadges, and that would hamper their sales, and we wouldn't be having this conversation, like we didn't for the 280x. The bios modification thread we have here proves that the famous marketing-speak for "power management block", is the tuned bios of the card and nothing more. Power consumption figures are spot on for overclocked Hawaii, performance the same. Why are we even arguing this? At least you don't change the subject like TheMac did, who took it from having Hawaii with a different name, to the amazing differences of producing the exact same thing in a slightly altered way, and how that (philosophically I guess), made it a new product.
     

  5. gamervivek

    gamervivek Guest

    Messages:
    32
    Likes Received:
    0
    GPU:
    HIS/4850/512MB
    You're using HardOCP's review to make a case when they are not using the hacked drivers, and digitalfoundry, who are using it, find >10% improvement in many cases with just a 10% core overclock on the 390X compared to 290X.
     
  6. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    HardOCP is testing with the same driver and frequencies, and found no difference. Same exactly with the Digital Foundry.
     
  7. fr33jack

    fr33jack Guest

    Messages:
    1,153
    Likes Received:
    4
    GPU:
    1050Ti @1.9/ 9.0GHz
    This should be sticked here on main AMD page...with title such as "The most important thing that you should know about 300 series"...written in red colored text with big font size. XD
     
  8. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    That's just speculation, but the facts of the performance and the deals we know, seems to back it up, and it looks like the most logical one.
     
  9. fr33jack

    fr33jack Guest

    Messages:
    1,153
    Likes Received:
    4
    GPU:
    1050Ti @1.9/ 9.0GHz
    As my favorite fictional character mr. Sherlock Holmes once said, "The world is full of obvious things which nobody by any chance ever observes" :)
     
  10. gamervivek

    gamervivek Guest

    Messages:
    32
    Likes Received:
    0
    GPU:
    HIS/4850/512MB
    Where do they say that? They mention this in their test setup,

    And if you look up 285's reviews, it performs the same as 280 at same clockspeeds in most games.
     

  11. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    You need to actually read the review, or at least the index before you tell me that I can't read it myself. There is a specific page in the review called "Clock vs Clock, Apples-to-Apples". There they specifically mention:

    I have quoted and linked that specific page 100-****ing times. No, the 285 doesn't perform like the 280 at all. In bandwidth hungry tests it is slower, and in tessellation heavy tests it is much faster.
     
  12. gamervivek

    gamervivek Guest

    Messages:
    32
    Likes Received:
    0
    GPU:
    HIS/4850/512MB
    I don't care if they think the architecture is the same, where do they say they are using the same driver for both 390X and 290X like digitalfoundry do with the hacked one?

    They don't. You do actually need to read the review, or at least the part I quoted and was also quoted on the previous page.
     
  13. sammarbella

    sammarbella Guest

    Messages:
    3,929
    Likes Received:
    178
    GPU:
    290X Lightning CFX (H2O)
    If you dare to post this quotes from HardOCP in the "390 and 390X aren't rebadges" thread the resident AMD fanatic fanboys will run to get a rope to hang you...

    ...After denying the validity of the source of course!

    LOL
     
  14. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    Wait. Let's take the case that they didn't use the same driver. So, even with the 390x supposedly using the newer drivers that everything has an improvement with and the 290x using the older ones, the cards perform identically at the same clocks. What does that tell you? :nerd:

    They also specifically mention (as any other serious website like Guru3D, Anand etc), that the chip is the same. But I guess that doesn't matter either. :nerd:

    Coincidentally that is exactly what the Digital Foundry guys say. And the same that the guys in the bios thread in the forum say. AMD themeselves are listing it as GCN 1.1 with the same details and guidelines as for Hawaii.

    But wait, there is more! The 390x works with the exact same open source driver as the 290x does! Devilish devils! How did they manage that sorcery! They even play Steam games by doing a standard library workaround because Steam on Linux doesn't recognize the name string of the card yet!
     
  15. gamervivek

    gamervivek Guest

    Messages:
    32
    Likes Received:
    0
    GPU:
    HIS/4850/512MB
    That you're wrong, quite wrong. And fwiw, it's not a case but what has happened.

    They also mention a GCN1.3 for Fury. And yes it doesn't matter what they claim, won't be the first time that they're wrong like with the recent macbook cape verde fiasco.

    DigitalFoundry are getting upto 20% performance improvement with a 10% core overclock using the same driver, though hacked. As for AMD, listing it as GCN1.1 when Dave Baumann doesn't use that nomenclature?

    You don't seem to understand, you think that my intentions are to prove that it isn't a rebranded Hawaii while it's something quite different.

    HardOCP are using the old driver on 290X in their review. So you're wrong when you at the same time claim driver improvements which are withheld from 290X while using the same review to claim architectural similarity/sameness between the two cards. You need something better than that, after reading the review of course.

    I hope that is perfectly clear now.
     

  16. sammarbella

    sammarbella Guest

    Messages:
    3,929
    Likes Received:
    178
    GPU:
    290X Lightning CFX (H2O)
    You should already be used to this, AMD is like Adidas:"Impossible is nothing!"

    It seemed impossible that a regedit values mod can magically add an "harware scaler" for old GPUs to get VSR.

    The use of old drivers by the AMD new 390(X) "hardware" doesn't seem so impossible or impressive.

    The people flashing 390X firmwares on their 290X are doing what is also considered impossible.

    AMD internal name for "Grenada" during development was "Hawaii refresh" so we can name the old 290(X) with new bios: "Hawaii reflash".

    LOL
     
  17. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    So wait. In the HardOCP article they perform within 1fps when clocked the same (although they don't give frametimes, which is usually the tell between the two forks of the AMD driver). The guys from the Digital Foundry say that they perform exactly the same with the same drivers at the same clocks.

    The 290x open source driver works out of the box with the 390x, and it plays Steam games. What is the logical conclusion of this mental trip we just had? :infinity:



    I don't know if the Fury is GCN 1.3, but it is at least GCN 1.2.1. It has extra L2, extra instructions, and a new video engine.



    They use the "Generation" nomenclature on their developer documentation.


    The title of this thread is very clear. What you want to say is not. If you believe that they indeed are rebranded Hawaii, then where is the difference from what I say? Unless you say that restricting the newer drivers is not the reason for the speed differences between the 300 series and the 200 series, while they clearly are as the hardware is identical.

    Fair enough. The Digital Foundry review is quite clear that the 290x performs much worse without the new driver. Everyone in here in the drivers forum has seen numbers and tested drivers that are from the branch that AMD is only releasing for the 300 series, and those drivers are more efficient, have higher framerates and consistently lower frame times.
    The scumm part is where they artificially limit them on the "new" 300 series (that even has Pitcairn in), so that their rebrands will appear much faster than what they actually are during the reviews period.
    I can't be more clear than that.
     
  18. kn00tcn

    kn00tcn Ancient Guru

    Messages:
    1,604
    Likes Received:
    3
    GPU:
    570m / MSI 660 Gaming OC
    was DF using a reference 290? that is, are they sure the gpu is fixed clocks?

    [​IMG]

    edit: this is regarding the screenshot on the modded 15.15 driver, NOT 15.5, where the 290x still didnt catch up to 390x

    edit2: i am also concerned about separate drivers, but this isnt new, almost every amd launch has had a separate branch until they merge a few weeks to months later, it's just getting messy now with win10 launch

    are the slower 3 series available or with reviews yet?
     
    Last edited: Jun 29, 2015
  19. PrMinisterGR

    PrMinisterGR Ancient Guru

    Messages:
    8,125
    Likes Received:
    969
    GPU:
    Inno3D RTX 3090
    I don't believe they say, but still they mention more than once that the performance of the cards was identical, so it sounds like they didn't have clock fluctuation in both.

    If they presented anything really different hardware-wise, the driver separation would be understood. But the drivers they launched only for the 300 series, are basically running on everything from the 5000 series up since January.

    As for the cards, at least the 380 is widely available as it looks like.
     
  20. gamervivek

    gamervivek Guest

    Messages:
    32
    Likes Received:
    0
    GPU:
    HIS/4850/512MB
    Again, HardOCP are not using the same drivers on the two cards, digitalfoundry are.

    They both say that the cards perform same at same clocks.

    Yet digitalfoundry claim performance improvement for 290X while in their own benchmarks it is beaten often by way more than the core clockspeed increase.

    Anyway, using both of them at the same time is wrong.

    Well you wouldn't be having this problem if you bothered to read. I just don't like spoon feeding.

    Then don't muddy the waters without reading something before using it as your infallible proof. I can't be more clear than that.
     

Share This Page