Announcement

Collapse
No announcement yet.

AMD Radeon RX 6600 XT Launching For 1080p RDNA2 Gaming At ~$379 USD

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
    bridgman
    AMD Linux

  • bridgman
    replied
    Realistically anything we did with Polaris at this point would be aimed at miners, not gamers. If the goal was increasing gaming GPU supply (leaving gamers still competing with miners for the supply) then starting with newer products would make more sense, and moving to newer fab processes than 7nm would probably work better.

    Once you start changing more than a few transistors on a chip you add two major chunks of work - redoing all of the simulation/emulation testing in the affected areas (which is significant) and redoing all of the physical layout for the changed blocks and any larger blocks which contained them. In practice that means redoing pretty much the entire chip for the kind of changes discussed here (FP16, AV1).

    One other thing I should mention early is that all this discussion assumes there is capacity at the AIB partner level, in terms of both their own production capacity and availability of all the sub-components that go into making a dGPU board. Many of those parts are also seeing shortages and extremely high prices at the moment.

    Probably the easiest way to think about it is levels of effort:

    * Low effort:

    Restart/increase RX 590 production (perhaps with more mining-friendly clocks) with current physical design, which I believe is the 14nm layout fabbed on 12LP. I'm not sure how much the 14nm design was updated to take advantage of 12nm - the leading rumor is that it's basically 14nm masks on 12LP fab to get better speed/power.

    I say restart/increase because I don't know if we are still placing orders for full size Polaris dies or only for the smaller 550/560 dies.

    * Small effort:

    Move RX 590 to 12LP+ with no change other than clocks/voltages, but without a shrink (or maybe with a linear shrink if feasible, which I doubt).

    * Medium effort:

    Redo RX 590 physical layout for 12LP+ taking full advantage of reduced feature size

    Redo newer chips (RDNA2 or 5500XT) for newer fab process

    * Large effort:

    Any logic design change (FP16, AV1 were examples) to RX 590 plus move to 12LP+

    These are all approximate but I believe they are generally correct.

    Anyways, key point is that making a Polaris-based product at this point would have to be aimed at miners because the current design works well in that role. My impression is that it is pretty well balanced between compute and memory throughput in its current form.

    Bringing Polaris up to date for gaming ends up requiring as much or more engineering work than starting with a newer product generation.
    bridgman
    AMD Linux
    Last edited by bridgman; 01 August 2021, 05:51 PM.

    Leave a comment:

  • drakonas777
    Senior Member

  • drakonas777
    replied
    By "forking GPU architectures for a different node" I meant backporting to non-compatible ones, basically like 4/5->6/7->8/10->12/14. This involves considerable amount of R&D and general production pipeline time. Mere shrink to compatible one sounds reasonable to me. I would certainly take something like RX690 on 12N+ or whatever assuming sane pricing. On the other hand, it wouldn't change the performance that much, so even standard RX480/580 would work for me just fine, especially if standard one would save months of waiting for shrank one.

    Leave a comment:

  • drakonas777
    Senior Member

  • drakonas777
    replied
    Forking GPU architectures for a different node is unrealistic/borderline unfeasible way to counter market changes. It's too expensive and too much of "inertia" to dynamically follow the demand. The most realistic thing AMD could have done is to extend lifetime of Polaris cards without any product changes. And even then it's not like you can shuffle around 12N resources like you wish - reservations of GloFo wafers exist most likely and all this assuming some general material shortages do not affect older nodes (which they did at least at some point AFAIK).

    Another way could be a small and cheap 7N GPU with 4G of VRAM to make it unattractive for miners, but powerful enough for 1080p ~medium for those, who are building new systems (mid range especially) and can't get a GPU at all, at the sane price that is. It would be a temporary solution to wait until market comes down. 5300XT was something like this, but it got nerfed due 7N shortage in general I assume. NAVI24 could be something like this at the and of the year.

    I do't think there is an easy solution to this problem. Either way you would have to pay some premium for some inferior product compared to "normal" conditions.
    drakonas777
    Senior Member
    Last edited by drakonas777; 01 August 2021, 07:14 AM.

    Leave a comment:

  • drakonas777
    Senior Member

  • drakonas777
    replied
    Oh, I see birdie has already memed around 6600XT price. It's funny how mr. value guy was totally silent during 3080Ti launch: +70% MSRP for +15% performance was a good value for him I guess, since it's a "quality product"...

    MSRP of 6600XT is not that bad considering current competition and semiconductor economy context. Problem is scalpers.

    Leave a comment:

  • bridgman
    AMD Linux

  • bridgman
    replied
    Originally posted by Teggs View Post
    I think I understand the point about price/performance/node, and that you are speaking about things which could be done now, but the situation on the low end is going bad over time. AMD is going to need a strategy to deal with this, it does not seem to have one currently.
    It's not just the situation going bad, it's that the market is gradually going away - with APUs taking over the new system sales - and vendors are reacting to the change. Low end parts for new systems used to be the largest contributor to GPU sales (not just by units, but by dollars if you go back a decade or more) and that has pretty much gone to zero thanks to iGPUs.

    The market that remains is a lot smaller and doesn't go down as far as it used to - basically entry-level gaming systems aiming for more performance than can be had from a current APU, plus upgrades to those systems. There is also an even smaller market made up of people upgrading an entry-level system that was built with a lowest-cost-possible GPU.

    The strategy is pretty clear and AFAIK is the same for NVidia - for each generation (or most generations eg we skipped Vega) position the lowest end product at the perceived sweet spot for entry level gaming. We put the 5500XT there, while NVidia straddled it with 1650 and 1660. Neither vendor made anything lower, relying on previous generations (Polaris & Pascal) instead.

    For upgrades to entry level non-gaming systems we both generally rely on previous generations. This usually works because when the previous generations launched the sweet spot for entry level gaming was lower since the new games at the time were less demanding.

    Originally posted by Teggs View Post
    Like you, I wished even at the end of last year that AMD had been foresighted and ported Polaris to 12LP+, but while that was possible and would have been helpful it doesn't address the issue in the future.
    It would have taken more than foresight at that point, unfortunately, it would have taken a lucky WAG. At the time our biggest supply problems came from mining demand followed by consoles competing for wafers, but back then it was expected that ETH would move to POS in 2021 and cut away one of the biggest supply challenges long before we could have put such a product into production. Making a mining-oriented product at the end of last year would have seemed like a terrible idea.

    Originally posted by Teggs View Post
    In fact a user used to get current tech at every point on the performance and price scales, but now not only are they getting a poorer experience because they bought a cheap product, they get old and sooner to lose driver support tech on top of that. When Polaris came out, AMD did not tell low-end customers, 'Buy our five year old shit!'
    So at the risk of outraging the internet, yes we did and so did everyone else. Until a couple of years ago the entry level products were mostly SI-based, since the last really small GPU we made was Oland with up to 6 CUs:

    https://www.amd.com/en/products/graphics/radeon-520

    We don't sell the 520 any more - the lowest end product offered is the 540 which I believe is a cut-down Lexa (550):

    https://www.amd.com/en/products/graphics/radeon-rx-540

    We only cut off driver support for older parts every 3-4 years IIRC, so Polaris should continue to be actively supported for a while.
    bridgman
    AMD Linux
    Last edited by bridgman; 01 August 2021, 01:59 AM.

    Leave a comment:

  • Teggs
    Senior Member

  • Teggs
    replied
    Originally posted by bridgman View Post
    ...If moving to a new fab process no longer automatically provides an automatic increase in price/performance then there's a strong argument for continuing to ship existing products (RX 550, RX 560, RX 5500 XT, GTX 1030, GTX 16xx) rather than creating a new card which would not offer much more in the way of price performance, and would possibly offer less once you included the (increasingly high) cost of designing, taping out and delivering a new product.

    I don't expect the lower priced cards to go away, but I do expect that new ones might come out ever 2nd or 3rd generation rather than every generation like the higher end cards. This is not an AMD position, just my own opinion from looking at the market and thinking about options.

    Guessing this is the JPR article ?

    https://www.jonpeddie.com/store/worl...-report-series

    If so then this link includes some extra comments from JPR on top of what the original article contains:

    https://www.pcgamer.com/gpu-supply-p...ments-from-jpr
    That is the article. Thank you for your response.

    I would point out that although RDNA 1 cards didn't provide much in the way of price/performance, they were at least newer tech. Newer decoding blocks, newer standards, etc and (presumably) longer driver support because they are the newer architecture. Someone who bought an RX 460 ~2016 has gotten years of current driver support. But what would I tell someone now? Buy an RX 550? I could have told them that when the 5500XT came out, but now it's another generation, the 550 has fallen even further behind in performance for their chosen task, and end of driver support is ever closer. It would be like telling them to buy R5 Terascale parts when the R9 series was current.

    I think I understand the point about price/performance/node, and that you are speaking about things which could be done now, but the situation on the low end is going bad over time. AMD is going to need a strategy to deal with this, it does not seem to have one currently. Like you, I wished even at the end of last year that AMD had been foresighted and ported Polaris to 12LP+, but while that was possible and would have been helpful it doesn't address the issue in the future. In fact a user used to get current tech at every point on the performance and price scales, but now not only are they getting a poorer experience because they bought a cheap product, they get old and sooner to lose driver support tech on top of that. When Polaris came out, AMD did not tell low-end customers, 'Buy our five year old shit!'

    I suppose there is another possible solution to this problem. AMD has droned on for years about low-end gaming being APU territory. If some of the leaks are correct and mainstream desktop Zen 4 parts are all effectively APUs, then maybe the problems with that argument can eventually disappear. It still needs customers to be served well in the meantime.

    Leave a comment:

  • bridgman
    AMD Linux

  • bridgman
    replied
    One thing I missed in previous post... a 512-bit interface makes die, package, board and memory expensive. Hawaii was one of the most expensive consumer products we ever made - IIRC only Vega20 and Navi21 boards cost more to make.

    I probably don't need to say this, but infinity cache doesn't slow down hashing, it still increases the hash rate relative to not having it. It's the lower memory bandwidth that limits hash rates on RDNA2 cards. There are a couple of different ways to bypass IC - per page or per shader instruction IIRC.

    The problem with all of these options is that we're talking about enough changes to require nearly a full multi-year design cycle, ie if we started now then by the time it came out we would have moved a lot of our products past 7nm anyways.
    bridgman
    AMD Linux
    Last edited by bridgman; 31 July 2021, 08:22 PM.

    Leave a comment:

  • bridgman
    AMD Linux

  • bridgman
    replied
    Originally posted by Qaridarium
    why do i think this will not work??? very simple if we talk about GPU mining we talk about Ethereum and Ethereum will soon switch to proof of stake instead of proof of power. also the mining hype is over because the price of the crypto coins is low...
    As soon as the mining boom is over our supply problems are significantly reduced - we should be past Peak Console by then as well - and this whole discussion becomes irrelevant. A gap-filler product has to deal with the scenario where the mining boom has not gone away by the time we are able to get it to market.

    If we are going to get gaming cards into gamers hands in that situation then we need something with good hash/$ (and maybe hash/W for the miners who pay for electricity) that is easy to build in volume that can divert the mining demand. Polaris is still one of the top mining cards by most metrics, even when bought used at well over MSRP.

    If I was just trying to make more consumer gaming cards I would go the other way, moving aggressively to a finer fab process.
    bridgman
    AMD Linux
    Last edited by bridgman; 31 July 2021, 07:17 PM.

    Leave a comment:

  • onlyLinuxLuvUBack
    Senior Member

  • onlyLinuxLuvUBack
    replied
    Originally posted by Jabberwocky View Post

    Same I bought a RX 480 the day that the AIB cards came out. I'm using it with 1080p 144Hz though. Still very happy with it. It's ironic but in most games that I play my 2700x is still the bottleneck.

    I won't not mind for a RDNA2 upgrade, but the prices are just to damn high at the moment.
    Yes in stock newegg 6700xt for 1000 bucks... Only double MSRP... open box... shut up and take my money.

    Leave a comment:

  • Jabberwocky
    Senior Member

  • Jabberwocky
    replied
    Originally posted by Melcar View Post

    Hahaha.

    Here I'm still on an old RX 480 8GB gaming on a 1440p 60Hz panel. Have a 6700XT on several wish lists, but those prices man. To be fair, I had a Pitcairn card before this (HD7850) that I also kept as a main card for 3+ years.
    Same I bought a RX 480 the day that the AIB cards came out. I'm using it with 1080p 144Hz though. Still very happy with it. It's ironic but in most games that I play my 2700x is still the bottleneck.

    I won't not mind for a RDNA2 upgrade, but the prices are just to damn high at the moment.

    Leave a comment:

Working...
X