1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

Radeon RX 5700 series review leaks out at Polish website

Discussion in 'Frontpage news' started by Hilbert Hagedoorn, Jul 5, 2019.

  1. slicer

    slicer Member Guru

    Messages:
    106
    Likes Received:
    32
    GPU:
    Gigabyte VEGA 64
    HWgeek and Undying like this.
  2. Loophole35

    Loophole35 Ancient Guru

    Messages:
    9,356
    Likes Received:
    835
    GPU:
    EVGA 1080ti SC
    You realize that if AMD were able to sell the Navi cards at the original MSRP they would have. Please stop acting like these companies care about the consumer. The fact they officially announced the the cards at $379 and $449 then perform one of their now famous price cuts before even launching the product is seriously only something AMD can get away with (at least from their blind faithful).

    Most said the Navi cards were a bad value when announced, The RTX's are still a high price but they have hardware based RT as an excuse (and seriously look at the die size of the RTX GPU's). Navi 10 is half the die size of TU104 and has 3 billion less transistors. AMD is still gonna make a killing with the current prices.
     
    olymind1 and airbud7 like this.
  3. pharma

    pharma Maha Guru

    Messages:
    1,085
    Likes Received:
    117
    GPU:
    Asus Strix GTX 1080
  4. OnnA

    OnnA Ancient Guru

    Messages:
    9,463
    Likes Received:
    1,927
    GPU:
    Vega 64 XTX LiQuiD
    cryohellinc likes this.

  5. Fox2232

    Fox2232 Ancient Guru

    Messages:
    9,738
    Likes Received:
    2,199
    GPU:
    5700XT+AW@240Hz
    Now discus it with those who came to conclusion that RT cores are actually cheap in terms of transistor budget.
    As for 1080p. Are you one of those 60fps average & 40fps dips guys? Mentioned cards do barely 130fps average on 1080p and have dips under 90fps. And that's without DX-R enabled.
    They are definitely not DX-R 1440p cards that can play without sacrifices of details at good fps. (And what is going from 1440p to 1080p else than sacrifice of details?)
     
  6. Fediuld

    Fediuld Member Guru

    Messages:
    106
    Likes Received:
    48
    GPU:
    MSI 1080 Armor OC /8GB
  7. Loophole35

    Loophole35 Ancient Guru

    Messages:
    9,356
    Likes Received:
    835
    GPU:
    EVGA 1080ti SC
    That link has been posted like 3 times in this thread OnnA.

    Honestly it's the prices they should have announced at originally. But now everyone thinks they are doing consumers a favor. Id be willing to bet AMD will make more on the sale of one Navi10 based GPU than Nvidia will make on 3 TU104 based 2070S's.
     
    airbud7 and OnnA like this.
  8. Fediuld

    Fediuld Member Guru

    Messages:
    106
    Likes Received:
    48
    GPU:
    MSI 1080 Armor OC /8GB
    Yet AIB 2070S are coming out at exuberant $600 prices!!!!!
     
  9. Loophole35

    Loophole35 Ancient Guru

    Messages:
    9,356
    Likes Received:
    835
    GPU:
    EVGA 1080ti SC
    Link? If that is true Nvidia will continue to not sell RTX cards.
     
    pharma and airbud7 like this.
  10. Kaarme

    Kaarme Ancient Guru

    Messages:
    1,622
    Likes Received:
    459
    GPU:
    Sapphire 390
    In my opinion AMD and Nvidia do care about their customers decently. As much as you could imagine a corporation do. In the end a business enterprise's first priority needs to be profit or they will go out of business. If you look back at Nvidia's GPU generations for quite a while, you will see every generation has generously beaten the previous one (performance, efficiency). 2000 series was a bit different since Nvidia decided to risk it all and try to change the games with the simulated RT. AMD has also done its best (for a company that was near bankrupt) to offer better products, and pulled it off well with Ryzen. AMD GPUs have been less than stellar of late, especially with the rebrands, but all in all every (real) generation of AMD GPUs has also offered more power to the buyers.

    Then you can look at the "eternally four cores" Intel and see what a corporation that doesn't give a crap about customers looks like. Because in the PC industry caring about customers means keeping the development rolling and offering more for the buyers with each generation. Not offering exactly the same, like Intel did until AMD kicked their ass with Ryzen.

    For me personally the RTX cards were too expensive, but that's my own problem. Lots of people are happy with them. It's not anything like the American nouveau drug companies that suddenly hike the price of a certain old drug up by 5000%.
     
    Aura89 likes this.

  11. Aura89

    Aura89 Ancient Guru

    Messages:
    7,637
    Likes Received:
    898
    GPU:
    -
    You mean you install that?

    I haven't touched geforce experience since it came out, not sure why someone would be sick of nvidia because of its geforce experience since....you don't have to install it....

    In my opinion, 5700 should be $249 and 5700 XT should be $299, maybe $349.

    It's clear these aren't top-end cards, otherwise they'd name them as such. So unless AMD has completely changed how they are naming their cards (which they could, for all i know), these are labeled as mid-ranged GPUs yet have low-ish high end prices.

    And it's not like their die is very large. RX 580 has a die size of 232 mm2 and 2304 cores, RX 5700 XT has a die size of 251 mm2 and 2560 cores. I mean at least when it comes to crazy prices on nvidia, they have large dies to explain at least some of the price, but i don't see what is driving the prices up here. It's a new architecture, which is great, i get that, but it's not like they are even bringing anything new, such as ray tracing, to the table.

    RX 580 8GB GDDR5 - 232 mm2 - $229
    RX 5700 XT 8GB GDDR6 - 251 mm2 - $449
    RTX 2070 8GB GDDR6 - 445 mm2 - $499
    RTX 2070 Super 8GB GDDR6 - 545 mm2 - $499

    Now, i'm not saying either companies pricings are right here, don't get me wrong. I'm simply stating that for $50 more, one company is able to bring in a chip that is near twice as the size, or more then twice the size of its competitors, with dedicated hardware for ray tracing.......i love AMD by RX 5000 series so far is overpriced and there doesn't seem to be any way to actually explain the reason behind this unlike nvidias gigantic sized GPUs and R&D of new technologies such as ray tracing.....

    I'm not trying to harp on AMDs GPU division here either, i'm simply stating even before the Super series, these GPU prices didn't make sense, other then to say "I can have similar performance to your GPUs, price similarly, and make more money from it"

    If the price drop stated earlier in the thread actually happens, that's great, but to me it doesn't seem enough, as i said, the prices i listed above seem like what they should be. But competition is competition and hopefully in the next couple of years there will be top end competition from both companies.
     
    Last edited: Jul 5, 2019
  12. Jawnys

    Jawnys Member Guru

    Messages:
    125
    Likes Received:
    21
    GPU:
    zotac amp extreme 1080ti
    well if amd comes out with stronger card than the 5700xt, i might have a worth it upgrade for my 1080ti without having to pay the rtx tax, fun times ahead, a full amd system in the near future, cant recall the last time i had that since my athlon 64
     
    Loophole35 likes this.
  13. Fox2232

    Fox2232 Ancient Guru

    Messages:
    9,738
    Likes Received:
    2,199
    GPU:
    5700XT+AW@240Hz
    @Aura89 : What does die size on 12/14nm vs 7nm has to do with adequate price? There were big GPUs on 28nm too, but that argument is perversion of logic anyway.
    If you want to make die size to price argument, state how many times more AMD has to pay for 7nm wafer than nVidia for 12nm. Then do normalization. And then you can start including something like GDDR5 vs GDDR6 pricing, since you included that too.

    But if you can't write manufacturing price of those cards and say that there is 1.XXx premium, entire argument is pointless.

    If they brought all of those mentioned cards for $50 less, your $299 and $349 would be lower. That's how human mind works. Would you expect nVidia to release 2060S for $299 and 2070S for $349 too? Would that lower price persuade you to buy AMD's card?
    If not, then AMD would be wasting their efforts reducing price for you.
     
  14. schmidtbag

    schmidtbag Ancient Guru

    Messages:
    4,427
    Likes Received:
    1,359
    GPU:
    HIS R9 290
    Not looking too promising so far, but I'm interested how this scales in 4K, and, how it'll perform after driver optimizations. So no matter what, it doesn't seem I'll be buying these when they're first released.

    People seriously need to stop thinking node size dictates the performance level of a product. Just because the transistors are half the size, that doesn't mean you get double the performance and double the efficiency. Die shrinks used to make a difference back when they were shaving off ~30nm at a time, but even then, the differences (though noticeable) weren't so black and white.
    The only thing that actually matters is the architecture. A smaller node size basically just allows them to add complexity to the architecture (in other words, more transistors), while offering a minimal improvement in performance and efficiency. More transistors doesn't mean better gaming performance. It doesn't even imply that.

    I'm getting so sick of people talking about 7nm like it's the holy grail of processor tech.
     
    Last edited: Jul 5, 2019
    HandR and Kaarme like this.
  15. Aura89

    Aura89 Ancient Guru

    Messages:
    7,637
    Likes Received:
    898
    GPU:
    -
    I'm sorry but no matter what is stated, die size costs money. I understand that 7nm wafers should cost more then 28nm or etc. wafers, especially since it's newer, but it doesn't change this fact that larger die sizes cost more overall due to, one, size, and two, lower yields.

    Unless YOU want to actually find out the information you're asking for, then the only thing we can do is make presumptions based off die sizes and historical information.

    And it's not like this hasn't always been the case, so unless 7nm is so far more expensive then any other transition we've made (Zen 2 prices would disagree with this idea) then i'm going to continue with my, logic based off the previous 30+ years, methodology.

    8800 GTX for instance on 90nm and a 484mm2 die for $600, went down to the relatively unchanged 9800 GTX on 65nm and a 324mm2 die for $349, and even further to the 9800 GTX+ on 55nm and a 260mm2 die for $229.

    And i'm not going to reply to anything you state after this as it's been clear that you don't want to listen to reason, replying would serve neither of us any good, because unless you can prove that 7nm is so, so, so expensive and that explains why such as small GPU has such a high price on it, then there's nothing to talk about. We only have past history to explain these situations if we don't have the niddy gritty details these companies typically don't give up. And considering the fact that Zen2 on the same 7nm technology is able to give us so much more CPU for the money, i don't see any evidence that 7nm is "expensive" to the point that makes these GPUs way over priced for their die size.
     

  16. zcubed

    zcubed New Member

    Messages:
    9
    Likes Received:
    0
    GPU:
    Geforce 1080 Ti
    If these results are accurate, I'm a little disappointed. I was hoping AMD could at least match my $699 1080 Ti from 2017... They have a long way to go if they are still playing catch up with cards that old. I know the price is 'only' 399 but seriously Nvidia just has to release minor tweaked cards to blow away AMD consistently. and don't get me started on RTX pricing, the 2080 Ti is an abomination because of that.
     
  17. kings

    kings Active Member

    Messages:
    99
    Likes Received:
    72
    GPU:
    GTX 980Ti / RX 580
    TSMC disagrees with you.

    https://www.tsmc.com/english/dedicatedFoundry/technology/7nm.htm

    Of course, the architecture matters in the general efficiency that can be achieved, no one said otherwise, nor was this what was being discussed.
     
  18. schmidtbag

    schmidtbag Ancient Guru

    Messages:
    4,427
    Likes Received:
    1,359
    GPU:
    HIS R9 290
    Yeah... obviously... pretty much their entire business revolves around transistor tech. They need to convince people that it's better than it really is. AMD has no reason not to play along with this.
    This is hardly any different than cigarette companies during the 1940s that got "endorsements" from doctors. The only difference is TSMC isn't doing something morally wrong here, but the point remains: they're making their product out to be a lot better than it actually is, and yet people like yourself are surprised when things don't turn out as good as claimed.
    C'mon everyone knows you take the numbers straight from the manufacturer with a grain of salt. What they're showing there is a best-case scenario, but if we are to trust the leaks shown so far, we're obviously not seeing that.

    I quoted you the first time around because you were using transistor size as a basis behind your comments. You were basically implying it should be more efficient. Well, it probably is in comparison to Vega, but you were comparing to Turing. Oxidized, meanwhile, was expecting better performance, using transistor size as the highlight. So yes, this is actually what's being discussed because (at least based on your phrasing) both of you are treating the transistor size as more important than the architecture. Otherwise, why bring it up as the main focus to your point?
     
  19. kings

    kings Active Member

    Messages:
    99
    Likes Received:
    72
    GPU:
    GTX 980Ti / RX 580
    Radeon VII have the exact same architecture as RX Vega, but on average has a performance/watt ratio 30% to 35% higher.

    Where did these efficiency improvements come from? They were made possible by the reduction of the manufacturing process.

    However, I understand what you mean, not only that counts for the total efficiency and I agree, but to say that the process does almost nothing, does not correspond to reality. It helps, and sometimes a lot.
     
  20. schmidtbag

    schmidtbag Ancient Guru

    Messages:
    4,427
    Likes Received:
    1,359
    GPU:
    HIS R9 290
    To give your argument more leverage, let's for a moment ignore the fact that the VII had double the memory, which also supplied more bandwidth (and yes, this makes a difference).
    Performance:watt ratio makes the results seem better than they really are. Don't get me wrong, it's very important. A 30-35% improvement is great, but still worse than what TSMC claims. On average, the VII is only 15% faster and about 11% more efficient than Vega 64. Both of those numbers are worse than TSMC's claims. Then, when you account for the improvement in VRAM (and probably drivers too), the improvements from the die shrink really starts to become underwhelming. Still important, but nothing to really pay much attention to.

    EDIT:
    If we are to compare to Vega 56 (which has a better performance:watt ratio than the 64), the margins to VII are even more narrow.
     

Share This Page