NVIDIA Likely Unveils GeForce RTX SUPER Lineup on July 2nd

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Jun 26, 2019.

  1. gerardfraser

    gerardfraser Guest

    Messages:
    3,343
    Likes Received:
    764
    GPU:
    R9 290 Crossfire
    I would say no RTX 2070 super is not as fast a s a GTX 1080Ti.
    My testing GTX 1080Ti is faster than RTX 2080 when overclocked.I would say RTX 2070 is not faster than either card when overclocked.
     
  2. Aura89

    Aura89 Ancient Guru

    Messages:
    8,413
    Likes Received:
    1,483
    GPU:
    -
    I'm confused why some of these things are not considered rebrands. I mean, i guess you could call them a refresh, but AMDs lineup of GPUs with GCN got very confusing. I mean cmon there were 4 different generations of architectures on the 200 series of GPUs. Terascale 2, GCN 1, GCN 2 and GCN 3

    But, many of these, for instance what was listed above as a 7970 and stated were not rebrands, were rebrands in the 200 series. There were maybe slight improvements on frequencies to make them not a "direct" rebrand, but they were still rebrands.

    7970, GCN 1, core config: 2048:128:32, size: 352mm2, memory: GDDR5 384-bit

    8970, GCN 1, core config: 2048:128:32, size: 352mm2, memory: GDDR5 384-bit

    280X, GCN 1, core config: 2048:128:32, size: 352mm2, memory: GDDR5 384-bit

    Now this is just one card of many that have been rebranded multiple times in the last...what, 8 years? The fact they have so many generations of architectures in a "new" generation of video cards is why there are so many "rebrands", sometimes it's in the high end, sometimes it's in the low end.

    HD 6000 series: 2 Generations of architectures

    HD 7000 series: 3 Generations of architectures

    HD 8000 series: 3 Generations of architectures

    200 series: 4 Generations of architectures

    300 series: 3 Generations of architectures

    400 series: 3 Generations of architectures

    500 series: 3 Generations of architectures

    Every single generation of GPU in the last 9 years (did not check more) appear to have some rebrands, sometimes many, sometimes small, sometimes in their high-end and low-end, sometimes only in their low-end.

    Compared to nvidia in the last 9 years

    400 series: 2 Generations of architectures

    500 series: 1 Generation of architecture

    600 series: 2 Generations of architectures

    700 series: 3 Generations of architectures

    900 series: 1 Generation of architecture

    10 series: 1 Generation of architecture

    16/20 series: 1 Generation of architecture

    Like others have stated, 700 series was nvidias "worst" series in regards to having rebrands, and it was confusing. It started out as a refresh of the 600 series, then fermi cards appeared in the lowerish end, and then some maxwell cards from the 900 series showed up to. It was a mess.





    Now, the reason i'm stating all this is that i feel it should be understandable when people look at AMDs past history of their GPUs and not being happy that they appear to be rebranding their cards a lot, as...they do, it's not always with the same tier of cards, but they rebrand, a ton, especially compared to nvidia.
     
    Last edited: Jun 28, 2019
    Loobyluggs likes this.
  3. Fox2232

    Fox2232 Guest

    Messages:
    11,808
    Likes Received:
    3,371
    GPU:
    6900XT+AW@240Hz
    Have you heard about boosting ability? Not back portable to HD7970.
    But technically 280X was rebrand by definition which I wrote. In other words, your argument is based on same thing I wrote while you try to disagree...)
    In this case you are making things up for sake of disagreeing.

    @Aura89 : Rebrand is when particular GPU is used in multiple named generations. It is not rebrand if multiple GPU generations end up with same naming scheme.
    This makes fundamental difference. GPU naming is not made by architecture version, but by release dates, so people do not think that just released GPU is actually something that's many years on market.

    Therefore your list should be made other way around like:
    Kepler: 600, 700, 800
    Maxwell: 700, 800, 900
    Pascal: 1000
    But that's still highly inaccurate.

    Instead real rebrands are:
    Tahiti: 7970 => 280X
    Pitcairn: 7870 => 270
    Bonaire: 7790 => 260X
    Hawaii: 290X => 390X (Grenada)
    Tonga: 285 => 380 (Antigua)
    Fiji: Fury X (No rebrands)
    Polaris 10: RX 480 => RX 580 (Polaris 20)
    Vega 10: Vega 64 (No rebrands)
    Vega 20: Radeon VII (No rebrands)

    Your way of not looking makes things appear much worse than they are. If you applied same method to nVidia, it would look much better too.
    And if you were into reality of rebranding, then Polaris 20 is not Rebrand of Polaris 10 because it is new tapeout with different properties of chip. (+Polaris 30)
    Or Grenada/Antigua. Even if chip is exactly same, different electrical properties resulting from new tapeout result in different potential.
    New tapeout is no different from die shrink (would you call that rebranding?). Chip may have exactly same architecture composition down to last transistor, but it will behave differently and is not rebrand as result.
     
    carnivore likes this.
  4. Picolete

    Picolete Master Guru

    Messages:
    494
    Likes Received:
    261
    GPU:
    Sapphire Pulse 6800

  5. gerardfraser

    gerardfraser Guest

    Messages:
    3,343
    Likes Received:
    764
    GPU:
    R9 290 Crossfire
  6. Aura89

    Aura89 Ancient Guru

    Messages:
    8,413
    Likes Received:
    1,483
    GPU:
    -
    I just want to point out i never said that the changes within the same generation (for instance 7970 and 7970 ghz edition or whatever) were rebrands.

    My point of view is any time there is a SINGLE change, it's no longer a rebrand. be that node size, memory bandwidth etc. When it comes to frequencies, that gets a little....tricky, as in some places i feel they are rebrands if they are too close to performance, AKA there's not enough of a performance gap, but if a "new" GPU in a "next-generation" naming has exactly the same architecture, node size, memory bandwidth, etc. but has 25% more performance due to a 25% higher frequency by default, i wouldn't really call that a rebrand.

    And no matter what view you take at it, AMD has rebranded way more products, in all tiers of their performances, then nvidia has, at least in the last 9-10 years. And the way i displayed my information is purposeful and not in my opinion worse then they are, it shows how much cross-architecture play is being done within AMDs namings vs Nvidia, which directly affects how much rebranding is going on within them. You can't have a generation of graphics cards with rebranding in it if there are only 1, new, architecture. But a generation that has 4 of them like the 200 series, or even 2 of them, shows there's likely rebranding.

    I mean cmon, i don't care that this is low-end, this is messed up:

    Radeon HD 5450, Terascale 1, 59 mm2, 80:8:4 core config, 40nm, released in 2010
    Radeon R5 210, Terascale 1, 59 mm2, 80:8:4 core config, 40nm, released in 2013(maybe 2014?)

    And no i don't care that the 210 is for "oem"

    My personal wish from BOTH companies is that they would never do this. Each generation, in my opinion, should never have anything leach over from the previous generation, unless it's a refresh. I get the point of refreshes even if i don't like them. But if i buy a card out of AMDs 200 series, i should not have to worry that it might actually be a terascale 2 GPU from Radeon 5000 series multiple generations from before, no matter if it's low end, or high end, it doesn't matter, i shouldn't have to worry about that. That's realistically my one and any standpoint, is that none of us should have to dig majorly into a GPU to make certain that you're not buying a "brand new GPU" that is actually 1 or more generations old. And fact is, AMD does this far more then nvidia does.
     
    Last edited: Jun 28, 2019
  7. Fox2232

    Fox2232 Guest

    Messages:
    11,808
    Likes Received:
    3,371
    GPU:
    6900XT+AW@240Hz
    Those lower end cards would not pay for themselves. Imagine 4~5 tapeouts per generation to cover entire scale. That would put AMD into red numbers much more than they were. And they would still be in red numbers. Sometimes using older generation and selling it at lower price makes more sense than trying to replace it with new generation which reaches marginally higher clock and has changes which are showing benefits only in some games/workloads.

    Basically replacing those GPUs with around 1792~2048SP with each GCN version made no sense, since till Polaris GCN was not really able to clock very high. And Polaris could have as well get there via 14nm.
    (AMD managed to put out one, maybe 2 GPUs per GCN generation. And their financial situation did not look good.)
     
  8. Aura89

    Aura89 Ancient Guru

    Messages:
    8,413
    Likes Received:
    1,483
    GPU:
    -
    I'm not saying they should make new low-end cards every generation, i'm saying they shouldn't release "new" cards that are old cards ever, regardless if it's low-end or not. Look at nvidia, they don't release low-end cards every generation. GTX 900 series only went lowest to 950 (not talking mobile), and i highly doubt there will be a 16/20 series low-end GPU. Neither of them NEED to have a new "named" low end GPU every generation.

    Yet AMD makes "new" low end GPUs every single generation it seems.

    Radeon HD 2400 (Terascale 1)
    Radeon HD 3400 (Terascale 1)
    Radeon HD 4350 (Terascale 1)
    Radeon HD 5450 (Terascale 1)
    Radeon HD 6350 (Terascale 2)
    Radeon HD 7350 (Terascale 2)
    Radeon HD 8350 (Terascale 2)
    Radeon R5 210 (Terascale 2)
    Radeon R5 330 (GCN1)
    Radeon R5 430 (GCN1)
    Radeon 520 (GCN1)

    That's 12 years of "low-end GPUs" that only have 3 generations of architectures in them.

    If they don't have something to replace a low end card next generation, they should continue to make the same low-end card with the same naming until they do.

    I mean imagine with these new RX 5000 series, such as the RX 5700, which is navi, the whole thing people have been interested in, and they make an RX 5300, and it's a GCN 1 rebrand...again(Radeon 520 is GCN 1), and has nothing to do with navi. I'm not saying that'll be what happens, but again given their track record, i would say it's likely.

    Yes, i know, i'm focusing on low-end here, but my point is, it's not always low end, but again, the whole point is: AMD rebrands a lot, this is fact, and people being upset about it should be understandable. I'm not happy either company rebrands, but if i'm going to have an issue with any company about rebranding, it's going to be AMD. They either rebrand, or tell us a brand new generation is "new" while using old generation parts in it way too often.

    To me, it seems like AMD puts out a smokescreen, makes it look like they are releasing new products after new products after new products, when in reality every generation they release very little "new" products.

    And this is with me: An AMD fan.
     
    Last edited: Jun 28, 2019
    0blivious likes this.
  9. 0blivious

    0blivious Ancient Guru

    Messages:
    3,301
    Likes Received:
    824
    GPU:
    7800 XT / 5700 XT
    $499 for the 2060 Super?

    Hard pass.
     
  10. Loophole35

    Loophole35 Guest

    Messages:
    9,797
    Likes Received:
    1,161
    GPU:
    EVGA 1080ti SC
    I doubt NVIDIA is that dumb. That’s the same MSRP as the 2070. This “leak” screams fake.
     
    airbud7 likes this.

  11. cowie

    cowie Ancient Guru

    Messages:
    13,276
    Likes Received:
    357
    GPU:
    GTX
  12. alanm

    alanm Ancient Guru

    Messages:
    12,233
    Likes Received:
    4,435
    GPU:
    RTX 4080
    Doubt it. Thats higher than current 2070 market pricing (roughly $470).

    Fwiw....
     
  13. Slammy

    Slammy Member Guru

    Messages:
    116
    Likes Received:
    31
    GPU:
    ASUS 7800GTX
    When talking about the RTX cards...the price!

    cause it sure isnt the performance. The RTX cards just became RTX 2080 Super Expensive editions.
    RTX is the most useless overpriced and unwanted "feature" ive ever seen slapped on a line of cards.
    I get it that they had to get the technology out but they should have ONLY had it on the titan. That is the only card that is powerful enough to sort of run it and its already stupidly expensive but atleast then it would have a reason for its price and would make is truly different.

    As things are now, Nvidia is out of their damn minds. $800-900 for the Ti version of the newest card, THAT IS IT for myself and most people. They are going to push alot of people to consoles with prices like this.
    RTX ON...FPS OFF....no thanks!
     
  14. Fox2232

    Fox2232 Guest

    Messages:
    11,808
    Likes Received:
    3,371
    GPU:
    6900XT+AW@240Hz
    This gain shows how flawed your approach is:
    2400 while terascale1 is RV610 w/ DE 2.0 and UVD 1.0
    4350 still terascale1, but RV710 w/ DE 3.2 and UVD 2.2
    5450 is terascale2, not 1.
    6350/7350/8350 while rebrands, were OEMs and therefore went to some office PCs and those who made them should have been capable enough to know what they are getting at large quantities. (Btw 2CUs, what kind of performance improvement one expects? It is to display excel sheet and play video. Those 80SP are not going to play games.)
    And last ones... are they any different from above? Information hints that they were OEM only and 520 was actually mobile only.

    I know what you are complaining about. But that mostly does not exist. Issue with rebranding is that it can be deceitful for uninformed buyer. But OEM is not that. And IIRC, case of HD 8000, those were made on OEM request.
    Than again, being of same CU architecture and having same number of TMU/ROP/SPs does not necessarily means they are rebrands if they have new chip codename or changes outside CU.

    That small list I gave you (few posts back) is only thing that matters. And that's when uninformed kid goes to build their 1st PC and get older tech under newer name.
    (There is no objective reason to try to make AMD more guilty of something than they actually are. Or is there?)
     
  15. alanm

    alanm Ancient Guru

    Messages:
    12,233
    Likes Received:
    4,435
    GPU:
    RTX 4080
    Sure, no argument there. But.... your Asus 7800 GTX in 2005 pricing ($560?) was just as much a rip-off as the RTX cards today, taking inflation, tech complexity and other variables into account.
     
    airbud7 likes this.

  16. Eini

    Eini Guest

    Messages:
    62
    Likes Received:
    6
    GPU:
    Nvidia 9600M GT
    I bought my RTX 2060 (Asus Dual) at a discount for 315 EUR, including VAT and one game. I mostly run it in cool (downclocked) mode, but can overclock it to around +145 MHz.
     
  17. Undying

    Undying Ancient Guru

    Messages:
    25,330
    Likes Received:
    12,743
    GPU:
    XFX RX6800XT 16GB
    Even though i would recommend waiting for super and navi reviews that was a good deal for 2060.
     
  18. Eini

    Eini Guest

    Messages:
    62
    Likes Received:
    6
    GPU:
    Nvidia 9600M GT
    Yes, I bought it one month ago. A Vega 56 would not have been much more expensive, though, and now you get them for 255 EUR and a Vega 64 for 312 EUR.

    A RTX 2070 would have cost about 100 EUR more, but I didn't think that 10-20% higher peak performance would be worth that money. That was before I started using Topaz AI plugins, which already run reasonably fast on the (overclocked) 2060, though.
     
    Last edited: Jun 29, 2019
  19. Noisiv

    Noisiv Ancient Guru

    Messages:
    8,230
    Likes Received:
    1,494
    GPU:
    2070 Super
  20. icedman

    icedman Maha Guru

    Messages:
    1,296
    Likes Received:
    268
    GPU:
    MSI MECH RX 6750XT
    The RTX 2080 needed this boost the most the gap between the vanilla 2080 and the Ti version felt like it was far large and prices I feel like they're trying to keep make the crypto prices the norm.
     

Share This Page