Page 2 of 2 FirstFirst
1
2
  1. #21
    Quote Originally Posted by larix View Post
    Unless there is exact date set for release date of new component. If you are need of upgrade you upgrade, otherwise you will wait forever as there is always something new "right behind the corner"

    @op personally I would go for r9 390
    The Pascal based cards will be arriving Q1-Q2 next year. The reason why it's worth waiting is that both they and AMD's Arctic Island cards will be around a 60%~ speed increase over existing cards, with overall far better DX12 functionality

    Their pricing will be the same as Nvidia's existing ranges, with perhaps, at most, a $100 or $200 increase on the standard price tags to attribute for the amount of HBM in play.

    So while I know there's always something new on the horizion, next years cards should be the first /actual/ game-changers in some time. The 980 and 970 will be relegated to budget cards, most likely being rebranded and dropped in price, to restate their new positions. The first Pascal card that drops will almost definitely be the Titan equivalent, along the $1000~1500 price range, with GP chipset variants that will cover the high-mid range cards. Then later in the year, they'll release the 980Ti equivalent as well as a budget equivalent, just to fuck with AMD's launches. This is after all pretty much Nvidia's modus operandi.

  2. #22
    Deleted
    Quote Originally Posted by Shinzai View Post
    The Pascal based cards will be arriving Q1-Q2 next year. The reason why it's worth waiting is that both they and AMD's Arctic Island cards will be around a 60%~ speed increase over existing cards, with overall far better DX12 functionality

    Their pricing will be the same as Nvidia's existing ranges, with perhaps, at most, a $100 or $200 increase on the standard price tags to attribute for the amount of HBM in play.

    So while I know there's always something new on the horizion, next years cards should be the first /actual/ game-changers in some time. The 980 and 970 will be relegated to budget cards, most likely being rebranded and dropped in price, to restate their new positions. The first Pascal card that drops will almost definitely be the Titan equivalent, along the $1000~1500 price range, with GP chipset variants that will cover the high-mid range cards. Then later in the year, they'll release the 980Ti equivalent as well as a budget equivalent, just to fuck with AMD's launches. This is after all pretty much Nvidia's modus operandi.
    Still no word on TMSC ability to mass produce that manufacturing node though, we know they have the ground work for the tech, but lets see, sadly I am expecting another delay and another issue is the mass supply of HBM, AMD has hynix global supply and you bet they will have some of samsungs as well, so question is, where does Nvidia get that supply from.

    Saying that, seems AMD is using TMSC to make the chips and same issue with Nvidia, so far no details stating they can mass produce those chips, its been a bit silent at the moment which is strange.

    I am hoping to see both camps release these cards sooner then later however as I want one of the halo products, considering Nvidias driver distribution coming into force pretty soon, I am very much leaning towards AMD this round.

  3. #23
    Deleted
    Quote Originally Posted by Shinzai View Post
    The Pascal based cards will be arriving Q1-Q2 next year. The reason why it's worth waiting is that both they and AMD's Arctic Island cards will be around a 60%~ speed increase over existing cards, with overall far better DX12 functionality

    Their pricing will be the same as Nvidia's existing ranges, with perhaps, at most, a $100 or $200 increase on the standard price tags to attribute for the amount of HBM in play.

    So while I know there's always something new on the horizion, next years cards should be the first /actual/ game-changers in some time. The 980 and 970 will be relegated to budget cards, most likely being rebranded and dropped in price, to restate their new positions. The first Pascal card that drops will almost definitely be the Titan equivalent, along the $1000~1500 price range, with GP chipset variants that will cover the high-mid range cards. Then later in the year, they'll release the 980Ti equivalent as well as a budget equivalent, just to fuck with AMD's launches. This is after all pretty much Nvidia's modus operandi.
    1. Q1-Q2 next year for flagship likely means something like Q3 for high-mid range, which op is targeting makes it almost a year of wait
    2. There will not be 60% speed increase, I will not believe it until I see benchmark, 30% is best I would hope for tbh. Unless you have some proper source to back this up, other then some graph made 2 years ago and blind predictions, there is no point bringing this up.
    Besides let's be honest, buy decently good gpu now and you are pretty much set till next get of consoles - unless you want to always play on ultra that is.
    3. Given OP is in market for $300 bracket gpu "The same pricing + $150 on avg" makes equivalent cards 50% more expensive, so yeah that's alot of an increase.

    so yeah if you need upgrade, just upgrade.
    Next gen consoles won't out sooner then in 2 years(most likely 5 years) so in either case with gtx 970 or r9 390 one could last these 2 years. If the wait for next gen is longer those that care about high/ultra are likely to upgrade after 2-3 years anyways and by that time yet better gpus will be available.

  4. #24
    Quote Originally Posted by Thorianrage View Post
    Still no word on TMSC ability to mass produce that manufacturing node though, we know they have the ground work for the tech, but lets see, sadly I am expecting another delay and another issue is the mass supply of HBM, AMD has hynix global supply and you bet they will have some of samsungs as well, so question is, where does Nvidia get that supply from.

    Saying that, seems AMD is using TMSC to make the chips and same issue with Nvidia, so far no details stating they can mass produce those chips, its been a bit silent at the moment which is strange.

    I am hoping to see both camps release these cards sooner then later however as I want one of the halo products, considering Nvidias driver distribution coming into force pretty soon, I am very much leaning towards AMD this round.
    I'm not sure where you get your information from, it's the same nonsense you came out with back in summer. TSMC has been pumping out 16nmFF for a while now. They entered high risk production months ago and are going through high volume production this year's final quarter. They're already at a forecasted $0.6 billion sales for their 16nmFF parts this year. They're moving over to 10nm production next year.


  5. #25
    Deleted
    I am just seeing small chips being made from that process, nothing real about proper sized chips, we have already seen TMSC use 20 nm on other types of chips but never made viable chips for desktop GPUs, if thats your evidence then don't know what to say, this slide would of been better supported if 20 nm actually saw the light of day on desktop but never did, again I ask the question, where is the evidence they have been producing desktop viable chips.

    You have not provided any as you would of provided the same evidence for 20 nm on desktop, look I am searching websites and non exist.

    - - - Updated - - -

    Not to mention it also just mentions tape out, there is no evidence of large scale production on large chips.

  6. #26
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    $0.6 billion sales this year? Aside from source, if that's to be believed, that's... not great compared to their current revenue. $0.6b contributing to $22billion with two months still missing, 2% is not great.
    http://www.tsmc.com/tsmcdotcom/ExtIR...ion=listByYear

    The current high profile chip is Apple's A9, and currently that's 2/3rd Samsung 14nmFF share and 1/3rd TSMC 16nmFF.
    http://anandtech.com/show/9686/the-a...-plus-review/3

    And don't believe TSMC and time table... last time they said they'll have 16nmFF in Nov 2013, here we are 2 years later. At this point pretty much every silicon manufacturer are having harder and harder time going down node processes.
    Last edited by Remilia; 2015-11-29 at 09:19 PM.

  7. #27
    Quote Originally Posted by Thorianrage View Post
    I am just seeing small chips being made from that process, nothing real about proper sized chips, we have already seen TMSC use 20 nm on other types of chips but never made viable chips for desktop GPUs, if thats your evidence then don't know what to say, this slide would of been better supported if 20 nm actually saw the light of day on desktop but never did, again I ask the question, where is the evidence they have been producing desktop viable chips.

    You have not provided any as you would of provided the same evidence for 20 nm on desktop, look I am searching websites and non exist.
    Why are you even going on about 20nm?

    Quote Originally Posted by Remilia View Post
    $0.6 billion sales this year? Aside from source, if that's to be believed, that's... not great compared to their current revenue. $0.6b contributing to $22billion with two months still missing, 2% is not great.
    http://www.tsmc.com/tsmcdotcom/ExtIR...ion=listByYear
    And it's supposed to instantly take over all of their earning instantly some how? In what kind of business, especially one with such solidly rooted older designs is that ever going to happen? almost 80% of TSMC's revenue comes from 28nm and larger designs. That's not going to change over night.

    They've had roughly 3 months to turn a profit on 16nm in the real world and as you point out, it's already 2% of their revenue. That's good business, given that Apple's the only one around ordering genuinely large quantities of 16nm chips atm. As a point of comparison, Samsung's semiconductors revenue has only increased 2-3% this year, so far.

    FF isn't a game changer for these companies revenues right now, it's an investment.

    The current high profile chip is Apple's A9, and currently that's 2/3rd Samsung 14nmFF share and 1/3rd TSMC 16nmFF.
    http://anandtech.com/show/9686/the-a...-plus-review/3
    ...TSMC had delays till mid-year. Apple wanted them in Q1. Samsung was ready, TSMC was not ready until late Q2. TSMC still has a third of the share despite this fact. 1+1=2, it hasn't been news for almost 9 months.

    http://www.macrumors.com/2015/04/15/tsmc-a9-iphone-30/

    And to finish off:

    http://en.ctimes.com.tw/DispNews.asp...ABC822USAA00NM

  8. #28
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Is it supposed to take over in a small time? No, but you make it sound like it's already a significant portion.

    And why are you believing marketing talk... We'll see if their yields get better or not and whether performance is better or not. TSMC is not a company that has good track record of being able to keep their target. They're not the worst company, but it sure as hell isn't the best.

    By the way, revenue is not profit. Revenue is cash before expense.

    Also revenue means jack in terms of technological progress (granted this is slightly off topic, well this entire part is). IBM's foundry (sold to Global Foundry now) is the first to have a commercially viable working chip < 10nm, 7nm to be exact. Granted they're not using pure silicon but silicon-germanium which is probably going to make a lot of factory changes. IBM though, is not top sale silicon manufacturer but they have technical prowess to be able to get to that point. Just because someone has a lot of revenue doesn't mean they are able to produce better products.
    Last edited by Remilia; 2015-11-30 at 12:51 AM.

  9. #29
    Quote Originally Posted by Remilia View Post
    Is it supposed to take over in a small time? No, but you make it sound like it's already a significant portion.

    And why are you believing marketing talk... We'll see if their yields get better or not and whether performance is better or not. TSMC is not a company that has good track record of being able to keep their target. They're not the worst company, but it sure as hell isn't the best.

    By the way, revenue is not profit. Revenue is cash before expense.
    I used the word revenues because of that exact reason. Have you not seen how much time has invested into the production of these chips? Even if it makes billions by the end of the year, there's no guarantee of it showing an actual profit until next year. I know business, I know economics. The returns are significant thus far, given the time frame. That's not up for debate.

    The fact is I'm not believing any marketing hype from TSMC. I'm basing it on what market analysts and researchers say. What are you basing what you say on?

  10. #30
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Why bother linking the TSMC is superior then? It has nothing to back it up other than what TSMC said, what a market analyst says means nothing, actual stats/performance does. That's the marketing, not the macrumor thing which we already know.
    Last edited by Remilia; 2015-11-30 at 09:12 PM.

  11. #31
    Quote Originally Posted by Remilia View Post
    Why bother linking the TSMC is superior then? It has nothing to back it up other than what TSMC said, what a market analyst says means nothing, actual stats/performance does. That's the marketing, not the macrumor thing which we already know.
    What stats or performance? What are you drivelling on about? Their revenue is up in double figures year on year, other than that, what? Their 16nm chips are performing as well as Samsung's 14nm chips, if not better. What exactly are you fishing for?

    http://www.kitguru.net/components/gr...ss-technology/

    Do you seriously believe they're just pretending to mass-produce these chips? I don't even know what your argument is supposed to be any more, other than "I don't want to believe, because... let me try and tell you what revenue means."

    Fact of the matter is that TSMC's 16nm chips match the performance of Samsung's 14nm chips. TSMC is ramping up production ridiculously and already has been provenly shown to be doing so, with billions of dollars invested into their production. The iPadPro (A9X), for example is all TSMC 16nmFF. Or are you going to suggest that doesn't exist and they can't possibly be supplying the chips for it?

  12. #32
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    What's the point of linking the last one then? Yield goes up, usually at least, when a process matures.
    We don't know how well they perform to the competition, the only way is if you actually have benchmarks for them. We don't know what their yield rate is actually going to be, it never is going to be as high as they advertise it'll be.
    Never have I said once that their revenue is down or anything. Why you bring up revenue at the start, that is really pointless to their overall yield and production and think that they're going to get 10nmFF next year is really odd, when even intel has noted issues with 10nm and under with pure silicon.

    Granted it started with Thorium's note on TSMC. It is true that the only productions have been 150mm2 chips. GPUs reach 300+ easily and big chips are at 600mm2 and the bigger they are the less yield and higher complexity. It is why historically AMD has never bothered with big chips (relatively) until Fury.

  13. #33
    Quote Originally Posted by Remilia View Post
    What's the point of linking the last one then? Yield goes up, usually at least, when a process matures.
    We don't know how well they perform to the competition, the only way is if you actually have benchmarks for them.
    There are benchmarks. Lots of benchmarks. I assume you at least know about 'chipgate'. http://www.techgrapple.com/apple-a9-...ith-iphone-6s/

    We don't know what their yield rate is actually going to be, it never is going to be as high as they advertise it'll be.
    Based on?

    Never have I said once that their revenue is down or anything.
    You literally questioned the implications of 16nm on their revenue.

    Why you bring up revenue at the start, that is really pointless to their overall yield and production and think that they're going to get 10nmFF next year is really odd, when even intel has noted issues with 10nm and under with pure silicon.
    ...They'll start a 10nm production next year, aiming for yields in 2017. This was stated back in July and they're still on target for that. http://www.kitguru.net/components/an...rm-cortex-a57/

    You know what, forget it, this is a waste of my time.

  14. #34
    Did you guys even answer the OPs question?

  15. #35
    Fluffy Kitten Remilia's Avatar
    10+ Year Old Account
    Join Date
    Apr 2011
    Location
    Avatar: Momoco
    Posts
    15,160
    Quote Originally Posted by Shinzai View Post
    There are benchmarks. Lots of benchmarks. I assume you at least know about 'chipgate'. http://www.techgrapple.com/apple-a9-...ith-iphone-6s/
    and I can show this...
    http://www.tomshardware.com/news/iph...smc,30306.html
    Based on?
    Every single marketing point ever...
    Qualcomm for example advertises 20% battery life (think it was that) from 800 to 810, reality was about 12-15% iirc. Numbers are always embellished for marketing purposes.
    You literally questioned the implications of 16nm on their revenue.
    That what, it's an extremely small portion of their revenue and is still insignificant? You were the one that even brought up revenue to begin with.
    ...They'll start a 10nm production next year, aiming for yields in 2017. This was stated back in July and they're still on target for that. http://www.kitguru.net/components/an...rm-cortex-a57/
    And they said that 16nmFF would be at November 2013, here we are two years later...
    Quote Originally Posted by Shokfactor View Post
    Did you guys even answer the OPs question?
    Sort of? No idea at this point, granted probably shouldn't have bothered to begin with but meh...
    Last edited by Remilia; 2015-12-01 at 02:45 AM.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •