Some bad news

Discussion in 'AMD Flavor' started by _mockingbird, Oct 7, 2017.

Thread Status:
Not open for further replies.
  1. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    Not sure why you think I said all AMD cards look like crap, I stated all Asus Turbo cards (that specific line of Asus cards) look like crap.....

    Now the rest of it, nope Cagey. The price hike or package sku prices, has nothing to do with demand. I can get a Vega 56 or 64 easily now, demand isn't there. Do you remember how long it took the 1070 to show up on steam? 1 month. Still no Vega. And demand for the 1070's was much much higher than it is for Vega. And in that month 2 weeks were cut off from launch to actual retail availability.

    Ask anyone at your local mom and pop store, or Best buy, ask them if their is so much demand they can't keep them on selves.

    Unlike the 1070's, and 1080's which miners are buying by the handful, Vega is fairly easy to get. I can't special order over 5 cards for those two nV cards right now from my Best Buy, cause they just don't have allocation in their warehouses country wide. Vega on the other hand same thing. And its not a miners card. Its purely supply constrained. You don't see miners buying Vega's by the pallet. Those small time miners that have a few rigs and don't understand perf/$ are the only people that are buying up Vega for mining. The rest, know better.

    Shit the 1080's get 12 bucks a card on the some alt coins (outside of Eth). Vega can't even dream of that.

    PS also the demand excuses was also used for Intel's latest CPU, demand for it made them sell out, BS, we know Intel is supply constrained.
     
    Factum likes this.
  2. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    Production of refreshed RyZen and Vega has been announced by AMD, 1st quarter production 2018 is the jest. RTG may have enough Vega 10's chips currently to carry through to then. Anyways this maybe a whole new AMD we are now seeing - not waiting on the competition to beat their pants off. A Ryzen refresh in a year is coming, Vega looks like it will be too but less than a year. So the news looks to be more good than bad.
     
    Last edited: Oct 9, 2017
  3. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    Even with the refresh they will get killed by current nV cards, 12nm process they are using is LPP, low power, and is only supposed to give 10% addition performance at the same wattage. Pretty much what AMD stated about it. Its more about cost savings than anything else just like 12nm from TSMC, GF 12nm also employs less layers.
     
  4. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    Hopefully a little design improvements, targeted areas of the previous design is tweaked. For Ryzen getting the clock speed up would be most welcome. Well for Vega it could use more improvement to say the least - I guess we just have to wait and see.

    As for production stopping on a chip - of course that will happen with Vega 10, maybe to allow Vega 11 production to take off while Vega 10+ is made ready. Who knows besides RTG/AMD and a few others.
     
  5. Aireoth

    Aireoth Gawd

    Messages:
    512
    Joined:
    Oct 12, 2005
    I do not understand these threads (especially in the amd section) its like a shitty neighbour throwing gas on a burning house.

    Those that know amd is in trouble already know that, its been on the wall for a long time. Those that don't are not going to change their mind because some forum posters.

    The rest are just fans and don't want to see their product die, and no one here can really do anything to help.
     
  6. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    10% is with design changes lol, I don't think they had time to do any design changes anyways. This was a fast turn around time, 12nm wasn't even on GF's map so.... 12nm is essentially 14nm. Just like 12nm tsmc is the same as 16nm.

    You have to understand AMD only has one CPU design team and one GPU design team, that is what it looks like and their resources that's all they got. If you look at Intel, they have many more just for CPU's, and nV we know of 3 design teams just for GPU's. That is a huge resource barrier for AMD.

    They can't stagger development the way nV and Intel do. Intel's capability of bringing out Coffee Lake on 14nm wasn't a last minute effort, they already tapped out Coffee Lake on 14nm close to year ago, even though it was planned for 10nm. You can't expect AMD to compete with companies like this when they have so much flexibility and resources, doesn't matter what AMD does, these companies will have responses ready to go within 6 months, cause they have multiple plans, where AMD has one.
     
    Last edited: Oct 9, 2017
  7. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    AMD has no choice now with Intel CFL and Nvidia current and future GPUs. I am sure design wise most of Ryzen+ has been in the works for over a year (GF 12nm is more or less GF 14nm++ and is not unexpected, I am sure GF wants buyers of the node to have the best chance of being ready so they as in GF can make as much money as possible). As for Vega II, it may have gotten in the recent period of time some pumping blood as in money to speed up the project. We just don't know enough in how much different is the node and how much is needed to make it work with current designs - it really doesn't sound like much. While Nvidia maybe delayed with DDR6, AMD may not have that issue with HBM2 in 2018 and a big faster Vega II will come out. I am more interested in Vega 11 and if HBM2 for real - that will be very interesting if the case.
     
  8. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    I don't think AMD is in trouble, in the black out of the hole for now. AMD will need to pick up the pace and continue with a much faster blistering rate than their competition. AMD needs to have Navi out next year for HPC and kick Volta's ass in the process - I doubt Navi gaming cards will be out next year. Now if GF can move up their schedule a little would be nice.
     
    cybereality likes this.
  9. Aireoth

    Aireoth Gawd

    Messages:
    512
    Joined:
    Oct 12, 2005
    How can they not be in trouble given their financial track record and the markets they compete against (intel and nvidia seperately)?

    Edit:
    I don't want to throw onto the fire, but thats a tall order when you have a fraction of a budget. I don't want amd gone, i remember the stagnation of the before times.
     
    razor1 likes this.
  10. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    I guess we have to wait 10 days to find out AMD's quarterly earnings. Ryzen and TR so far has been very successful. This quarter if Intel can step up the availability of CFL may start to slow them down some. It is not always about the money spent but more to deal with who and the team you are dealing with that can win the day. Same as with movies, some rather big money spent movies flopped big time, same with any company R&D.
     
  11. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    On the CPU side consumer market they are safe for the near feature (q1 of next year, once lower end motherboards for CL come out back to the drawing board), but server market, nada, and that will only get worse once skylake server products are released. There is no apparent reasons for large server companies to switch over, AMD's server products don't offer anything over Intel server products outside of niche reasons.

    HBM 2 on Vega 11 not going to happen, too expensive for their large chips, let alone midrange...

    If margins don't improve for AMD 10 days from now another words, 37 to 40%, that will tell us A LOT, that means server markets haven't shifted. Server products have margins almost double that of consumer, right now AMD is at 33% margins, if I remember correctly, and that is not much higher than before which they were at 31%. The first quarter of Ryzen's release they went up to 35% and then dropped back down to 33%. So that was consideration of both DIY Ryzen products and OEM systems. OEM systems dropped margins for AMD while DIY increased margins. Bottom line they are still in the red, so if they manage to get into the black, they will have increased sales figures, how much will tell us if they are really taking away from Intel's market share or its just the course of increased margins. If we see just hitting into the black, that means they haven't gained. They need to show 50 million + possible 100 million + per quarter to really make dent into Intel's market share. Remember server markets 1% will net 1 billion per year, that's 250 million a quarter. That's why I highly doubt servers are going to play much of a role next quarter results.

    AMD is not out of its hole yet, Coffee lake pretty much rendered Ryzen as a less competitive part, still better than before with BD, but still not that great. We are back to Phenom vs. Nehalem type situation. Any chip other than the 1700 and 1700x in Ryzen's line up are pretty much useless now (well still have to wait for low end motherboards for CL of course which won't happen till early next year, 1Q)

    Lets not even go to Navi vs. Volta, cause by the time Navi comes out, Volta's replacement will be on the wings. To even reach Volta is pretty much impossible right now, if the compute performance we have seen with V100 holds up, Volta is going to be crazy. Even TSMC's 12nm process, doesn't give much over the 16nm process, they said higher clocks at same power consumption, something 20% nothing spectacular, but this isn't a low power node unless 12nm from GF.

    The best AMD can hope for with Navi is compete with midrange and possibly the performance segments, just like now. But even the performance segments I think they will have a though time. As I said 200% increase in perf/watt to match Volta, I can see them getting 100% which will take them and put them in the same situation they are in right now with Vega vs Pascal and also a year later.
     
    Last edited: Oct 9, 2017
  12. Imhotep

    Imhotep Limp Gawd

    Messages:
    398
    Joined:
    Feb 12, 2014
    The whole arch of Vega and its upcoming smaller versions are designed around HBM. Navi and beyond will also be evolutions of HBM. Get your 411 correct.
     
  13. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    I highly doubt that, unless they will go to HBM 3 (low cost), just too much money for the lower end cards man.

    Now if you guys are citing WTF tech as you source, I wouldn't even believe what they say .....
     
  14. Imhotep

    Imhotep Limp Gawd

    Messages:
    398
    Joined:
    Feb 12, 2014
    You can believe whatever you want, but I think you should review all of the pre release Vega slides.
     
  15. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005
    There was no mention of Vega 11 in Vega pre release slides not consumer slides, the first time it came was packaging deals shifted from one company to another ;) Packaging isn't just for HBM, HBM just needs more complex packaging. that rumor was inclusive of Vega 20 which will use HBM of course. So Vega 11 with HBM probably not, just a deal to make packaging for both Vega 11 and 20 under the same roof so to speak.

    https://videocardz.com/65521/amd-vega-10-and-vega-20-slides-revealed

    You mean this one, where it doesn't talk about memory?
     
    Last edited: Oct 9, 2017
  16. Imhotep

    Imhotep Limp Gawd

    Messages:
    398
    Joined:
    Feb 12, 2014
  17. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005
    They never talked about Vega 11, they just stated Vega which at that point could have been just Vega 10.

    I don't see why they would not do the same thing as nV did with their Tesla cards vs their gaming cards, have two different memory bus for designated markets.

    It makes no sense to have HBM come down to lower end parts when they are cost prohibitive.

    AMD still hasn't even used GDDR5x yet, which would be even better option if Vega 11 has the performance to saturate that bandwidth.

    Now updated PS4 (PS4 pro), uses Vega as its core and it doesn't use HBM...

    Don't tell me Vega can't use other memory types other then HBM, it can and will if the market doesn't allow it from a cost/benefit ratio.
     
    Last edited: Oct 9, 2017
  18. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    Well to get HBM costs down you need to up the quantity produced. If RTG is going the HBM route they will have to support the infrastructure beyond very high priced compute devices and top end small volume gaming cards. It is a gamble either way - go with DDR5(x) and DDR6 and get your ass whipped by Nvidia, go a different route, create a niche no one else has and maybe still get their ass handed to them. Their whole shift in memory architecture also goes along with their packaging processors with memory and maybe other stuff like SSD's. If they had a very compelling product they can make money not only on the processor but memory and storage as well. I would think Apple would be all over this if they had it down. The mobile market is ready for something like this, AMD just has to tap into that huge market which they have lost over the years to both Nvidia and Intel.

    Ryzen is a system on a chip type processor with USB ports, PCIe for M.2, Sata controller -> it design speaks of a package option from AMD probably most likely mobile orientated. AMD just can't do everything at once but the design is there. Adding in the GPU as in APU should help.

    Unfortunately Vega 10 just does not bring that much excitement once performance is viewed, it competes with a 1080 which was great performance for last year, now with the 1080Ti it looks a little anemic. Vega 11 maybe different (I did say maybe). With one stack of 4gb of ram and of course the blaring HBCC with marketing speak "you will never ever ever need more with that jest", of course you can get 8gb of HBM2 on one stack but I would think that is when the costs would skyrocket. RTG really needs the 1070 and down market, as in dominate to win out. Beating Nvidia at the 1080 Ti level would be nice but beating Nvidia 1070 and down including Volta versions would be the big turn around they need for the PC gaming market.

    HPC is another area that is ripe for making money if RTG can start to compete their, Nvidia making a specialized compute heavy chip and then a separate one for their gaming cards and lower end professional cards has paid off big. With Vega power requirements I just don't see a win here either, while Volta V100 is somewhere around 132% faster than P100, Volta just has more of everything else that is needed, software, hardware support, APIs, developers with years of experience etc. Navi could upset Nvidia a little if ready a year earlier than Nvidia next compute chip - big if.

    Will Vega 11 have HBM2??? That is not clear but I would not rule it out at this stage.
     
  19. Anarchist4000

    Anarchist4000 [H]ard|Gawd

    Messages:
    1,600
    Joined:
    Jun 10, 2001
    Allowing for some spin, it's not exactly unexpected even without a source. Standalone cards are nearly impossible to find. Bundles are widely sold to hedge against miners. We have no insight as to how many Vegas are being sold directly into some HPC markets. There are some state actors, one with a vested interest in AMD, instituting a state sponsored crypto currency. Partners have said well binned chips are rare, but that doesn't necessarily mean they don't exist. Just that AMD could be pushing all of them into higher margin Instinct, SSG, or dedicated mining cards. At least in the financial analyst presentations GPUs are still seeing strong demand for AMD.

    HBM really doesn't cost all that much more than any other DRAM. Some higher packaging costs, but that comes with other benefits that may very well offset those costs, if not make the final product cheaper. HBM is ideally suited for the mainstream markets considering how AMD implemented Ryzen MCMs. Ryzen uses Infinity to connect to other Ryzen cores (Threadripper). Vega uses Infinity to connect the graphics core to the memory controller. Connecting a CPU to a HBM2 memory controller is a logical step. More power efficient, less complexity on the motherboard, and could alleviate the need for any DIMMs. For a NUC that's huge (or small)! I'm unsure the memory makers are ready to ramp production of HBM2 quite that much. Comparisons to DDR5/6 aren't even valid as they can't be made small enough and costs possibly not all that different. At least I haven't seen prices on motherboard without DIMM slots. Embedded may make more sense than a MCM approach, but with multiple chips sharing a HBM stack. HBM, or stacked technologies, are the future, there's not really any way around that.

    Nvidia, other than Shield, really isn't in that market. Limited revenue and nothing in comparison to Intel's 70%+ graphics marketshare thanks to integrated graphics. That's still allowing discrete graphics to pick up some share.

    Competes on some current titles, but the new generation of engines designed around DX12/Vulkan are only starting to hit and heavy on compute. Even Doom, AOTS (~30% compute), etc are poor examples of the technology, despite being well optimized for what they are doing. They were also engineered with fallbacks to DX11/OGL in mind.

    Consider that 80%+ async compute quote for a moment. Compute doesn't use ROPs, TMUs, compression, or triangles. Performance in those cases falls to FLOPs and memory bandwidth. Cases where Vega10 is ahead of 1080Ti in FP32, and double with RPM (comparing FP16 to FP32 as it's the best case).

    Most of HPC isn't GPUs, so while deep learning is significant, the broader market may be some sort of APU or MCM design. Not because of the GPU, but the memory and bandwidth that comes with it. HBM2 + NVDIMMs shouldn't be underestimated here, although those designs are still being explored. The broader market comparison would be AMD vs Intel CPUs, but with HBM2 bandwidth thrown into the equation as a fast cache. Even a smallish GPU on the side could be used for some workloads, but massive arrays of discrete GPUs are rather limited in their uses. Certain fields sure, but that's only a segment of a larger market. In the case of a national lab, a generic cluster is often used over a specialized one as the workloads can vary significantly.
     
  20. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    Density of the design, getting wide high bandwidth communications paths between the chips are key elements in getting better performance. HBM falls under that perfectly with a small footprint, close proximity to everything else and very wide communication path and bandwidth. For example AMD comes up with a package solution of an EPYC cpu, GPU and HBM memory that fits into an EYPC socket. Small dense high performance solutions would win out in the mobile market, embedded devices. Just the next huge shift that maybe coming. Unless something big happens, 7nm is about the end of the line for process nodes, 5nm may take a decade if ever. Packing more and more transistors on larger and larger chips won't work at a given point - hence separate smaller chips but then you have to communicate effectively between them. Anyways I just don't see RTG giving up on HBM with Vega 11 but I really don't know.
     
    N4CR likes this.
  21. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005
    Cryptocurrancy is not why AMD hasn't been able to gain marketshare, they would actually gain marketshare if they start producing more chips/cards. The state of Ethereum's Byzantium blockchain patch will increase mining rates by around 35%, Eth will not go to POS till end of 2018 now. So anyone mining with GPUs or thinking of it, its still strong. Now I would still stay away from any AMD products as they use much more power.

    its considerably more costly around 75% more. Once volume increases it will go down though. So just from that alone the rest of your comment......

    By then a new generation of cards will be out, so what? Same old thing we have seen from nV vs AMD, nV focuses on the current. And when the future comes they end up over AMD's.

    You just generalized why too much I'll wait and post later on this, see if you can find your fault.

    HPC is a growing market, but it needs both CPU's and GPU's, APU and MCM designs are nice but won't cut it we have already seen that.....

    HBM2 is being used in HPC and DL by all vendors, NVDIMMS are nice but not a necessity, even in virtualization not really need. Smaller systems like for small businesses or home use yeah what you are talking about come in handy, less upfront cost, outside of that, not needed.
     
  22. Pieter3dnow

    Pieter3dnow [H]ardness Supreme

    Messages:
    4,969
    Joined:
    Jul 29, 2009
     
  23. Jedibeeftrix

    Jedibeeftrix n00bie

    Messages:
    37
    Joined:
    Dec 1, 2016
    If they don't start shipping HMB2 in volume the price will never decrease to commodity pricing.
    To do that it needs to ship en-masse in millions of mainstream cards, not thousands of high end cards.
     
    razor1 likes this.
  24. Romale23

    Romale23 Gawd

    Messages:
    880
    Joined:
    Dec 12, 2006
    you know, this thread almost has me convinced to short amd
     
  25. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    Yeah for Vega 10 its not possible because its too damn power hungry, you think Vega 11 a mid range chip will use the same power?
     
  26. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    Don't short it, not yet, this is why AMD hasn't been giving out forcasts for their up coming quarters, cause there are way too many unknowns.
     
  27. Romale23

    Romale23 Gawd

    Messages:
    880
    Joined:
    Dec 12, 2006
    I get what your saying but we all know its going to drop back down, the outlook timeline companies use is just to short for AMD to really fix itself. I've always bought AMD really low and sold at over 100% profit except for this last time (i told too early). I totally think its utterly overvalued right now, should be somewhere around 6.
     
    razor1 likes this.
  28. Algrim

    Algrim Gawd

    Messages:
    856
    Joined:
    Jun 1, 2016
    I think that the Ryzen/Threadripper/Epyc trifecta will go a long way to help AMD's bottom line. Unfortunately for AMD/RTG, it needed the same kind of win in the graphics space and, to date, doesn't seem to have pulled it off.
     
  29. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    it is overvalued but speculation driven markets are always like that. So realistically 6 is where AMD should be based on their current value as a company, the potential is driving it much higher, without guidance from AMD themselves, which they aren't giving we really don't know the future. We can guess and that is where the speculation comes it. I would use a put and not a short, just to be on the safe side because of this.

    Having said that I don't do options much, unless I know exactly what is going to happen lol. Just too risky with a volatile market.
     
  30. Communism

    Communism n00bie

    Messages:
    55
    Joined:
    Feb 24, 2013
    If we were going by fundamentals AMD stock would have been delisted decades ago, since they never make a profit in a "legacy," declining, industry segment.

    Never doubt the "irrational exuberance" that having infinite AMD shills everywhere on the internet produces.

    "Fundamentals analysis" is 100% pointless in a company that has almost never had any fundamentals ever in it's history.

    The entire stock is based on fantasy and always has been, it can go anywhere depending on how the fantasy is spun in the media and the mindshare.


    The time to short was just before Ryzen launch and just before Vega launch, since AMD shill bubble always inflates the reality of the situation and the reviewers will eventually tell closer to the truth once the AMD bribes run out (in an attempt to salvage a portion of their credibility).
     
    Last edited: Oct 12, 2017
    trandoanhung1991 and razor1 like this.
  31. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005
    hmm good point but I don't think it can be delisted for not making into the black. Its more about equity of stock the company holds and stock price if not mistaken when companies get delisted.

    Nasdaq is a bit different than the DOW.

    Ryzen is a decent step up from what they had before, will it make a significant enough impact, I don't think it will as I stated when it came out, its nothing over what Intel already has, it ruffled Intel's feathers a little bit here and there, but that's all it did.
     
  32. ManofGod

    ManofGod [H]ardness Supreme

    Messages:
    7,958
    Joined:
    Oct 4, 2007
    AMD is not really going anywhere and this whole thread is based on someone's opinion represented as fact. In fact, there is a Good News thread that actually is fact and I can see lots of folks not appreciated that, at all. :D This thread, among other things, is why I will not install Nvidia in my own personal machines. (That and the washed out desktop colors that Nvidia gives, which could not be fixed.)

    As too the person above, if all it did was slightly ruffle Intel's feathers, then why did they paper launch a processor or two, eh?
     
    noko likes this.
  33. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    Retail sells, AMD is a smash hit, OEM wise (bigger sells) not so much - OEMs AMD will have to claw back. Upcoming APU should help there. Mobile market??? Raven Ridge with low power Vega (;))? We have to see how AMD does there, CFL without lower cost chipsets is not going to keep Intel in the lead, while AMD has more cores they are still at or lower power levels as well which is very interesting.

    AMD should be in the Black and have Momentum, a little bit deflated by Vega 10 but Vega 11 will be more important (Vega 11 maybe is step to Navi, we have to see). I see AMD CPU sells continuing due to lack of Intel availability of CFL which just killed their Kaby Lake line :LOL: while not delivering on Coffee Lake - Intel better have some nice discounts to get rid of all the obsolete stock.

    AMD needs to deliver a sound response to CFL with Ryzen +, Intel misstep won't last forever and I am sure Intel will turn up the volume, have lower end chipsets etc. AMD is about to own the HEDT market as it is.

    So AMD has strong retail sells, OEMs are picking up but that takes time and can be slow, almost non-existent in mobile volume wise which is a huge opportunity if they can pull it off, HPC? Servers? I see AMD only going up and not down here overall.
     
    cybereality likes this.
  34. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    Retail as in DIY? yeah but that is a very small % of processor sales.

    CFL will take time to hit but OEM"s aren't going to change their contracts for just a month or two.

    Retail sales are so small they don't do anything. Look at AMD back with Athlon, Retail DIY they had 50% of the market and growing. But over all they were what still in the 30% marketshare range over all? Granted Intel locked the OEMs down illegally but still that is how important OEM's are in processor sales.


    its a slight ruffle, Intel they could easily go into a price war and shut AMD down for good, they could have done it in any of the past 10 years after Ivy Bridge's launch.

    That is what you don't understand, AMD right now, if Intel or nV wants to screw them over all they have to do is drop their processor and GPU prices by 10%, both companies have margins over 60% in their respective products. For them its nothing. For AMD to go down 10% will kill them. If Intel or nV felt threaten, that is what they will do, it will be a last resort. Not releasing a 6 core part that lays waste AMD's 8 core parts in almost all tests. Just imagine if Intel didn't have problems with 10nm what would have happened? Ryzen instead of looking like Phenom, would look like BD again. Pretty much the only time AMD could get anything done right was when their competition screwed up.

    AMD Althon and 64 were great chips but only because Pentium 4 sucked that much. P4 sucked so bad Intel went back to Pentium II architecture to make the core line ups!

    ATi, r300 was a great chip but only because FX series sucked so much. FX was barely faster than GF4 ti cards in DX8 games!

    Now lets look at it from the other side when Intel and nV did good and AMD failed.

    CPU wise, AMD failed with BD, PD, Phenom (was the start),3 gens in a row. These were going up against, Nehalam and then Ivy Bridge. If they didn't realize what was coming after Nehalam, we saw where Intel was going. Was there another gen before Nehalam, Penryn right?

    All this time Intel failed once with P4, but AMD just kept on going one after the other.

    now lets look at graphics, we see a similar trend, outside of 1 time when ATi took the performance crown and was able to hold on to for more then one generation was with the r300.

    See the pattern, AMD every now and then gets a leg up but they can't keep it for long nor do their competitors stay down for very long.

    In reality AMD in graphics has not held on the performance crown for more then 6 months, ever.

    Now back to Ryzen vs CFL, that is exactly why Intel brought it out early, cause they aren't going to give AMD any room to move. If they truly wanted to outright crush them they can do that easily, 10% drop in prices is all they got to do. AMD won't be making any where near the black if that happens. Put this into perspective 2% margins for AMD is around 150 million bucks per quarter. So a 10% drop in margins, figure out that as 750 million per quarter drop. Game over.

    Just to put this into financial perspective, Intel last quarter beat the streets estimates by 4 cents which is pretty much that's like 7.5% difference, that was because the street thought AMD was going to do better than they actually did. Intel is also estimating to remain flat from Q2 to Q3, that means they are not expecting AMD to gain any ground from their Zen line up from server to desktop...... So in their eyes AMD will only gain from their increased margins not by taking anything away from Intel.

    This is what happens with a company over hypes their shit, once reality hits all the good fortune of stock prices going up, will come crashing down.

    This is also why its so damn hard to come back when you have less then 20% market share in a two horse race in the tech sector. Cause you are going to be at the mercy of the opposing companies. Your margins drop as market share drops cause you don't have the influence with B2B negotiations anymore, while the competitors have all the luxuries with these types and grossly increase their margins.

    Did you ever hear 60% + margins in a silicon company? I think that is as high as Apple...... Its crazy isn't it?

    Edit made a mistake, Intel would need to price their chips 10% lower than AMD's current offerings, so that would mean a 20% drop from retail prices for Intel would cause a 10% drop in margins for AMD, pretty much forcing AMD to drop their retail prices by 10%.
     
    Last edited: Oct 12, 2017
    Factum likes this.
  35. cybereality

    cybereality 2[H]4U

    Messages:
    2,364
    Joined:
    Mar 22, 2008
  36. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005
    Factum and cybereality like this.
  37. noko

    noko 2[H]4U

    Messages:
    2,958
    Joined:
    Apr 14, 2010
    I don't see CFL destroying RyZen, competing effectively yes somewhat. Everything $300 and down looks better and better for AMD from my view.

    Had my Ryzens since March, in March Intel had nothing that competed with it for the price. Now the 8700K does do that and glad to see Intel giving us a real upgrade option from last generation (if one can find one that is). The one thing, at least AMD does well is they do have a loyal following and there is probably a few good reasons that is the case. While CFL 6 core beats (not always) my 1700x (All 8 cores if clocked the same perform virtually the same) I also will have an option to upgrade next year with a Ryzen + if I want to or Ryzen II in 2019. No one can say that with Intel, Kaby Lake Owners have no major upgrade then what was already available - that is pissed poor. CFL who knows what upgrade path those owners will have with the CPU. Intel Icecake will probably once again require a whole new motherboard, I maybe on my third CPU by then on the same motherboard :LOL:. Cost wise Intel is still very much bloated and should cut their prices more.

    Do agree OEM sells really is the cake which AMD has been growing but so far it looks sluggish. Mobile is another area that AMD needs strong growth, here I think AMD has a good shot, Ryzen is very efficient and gets almost unbelievably efficient at lower clock speeds and voltages, AMD should work with Nvidia, bite the bullet if Vega does not perform well there and have AMD/Nvidia run laptops. A 6 core Ryzen with a 1050Ti or above would be a very potent combination. Of course AMD could use an Rx 560 and above and dual graphics when APU's are around but the Nvidia combination would probably be better overall and would help sell more AMD processors.
     
  38. razor1

    razor1 JustReason is my Lover

    Messages:
    9,460
    Joined:
    Jul 14, 2005

    CFL 8700 non K completely destroys the 1700 (non X, 1600x and 1600 (non X), its priced so that it competes with both those, right in the middle so pay a little more and get a lot more performance or pay a bit less and still get more but not too much more, the I5's were and still are Intel's money makers in both business and retail they are going to be a tough fight for the 1600 and 1500 . And forget about the lower end Ryzen's cause 4 core 8 thread or 4 thread Intels, that like taking a Skylake or Kaby lake and putting it up against a 1300. The main problem with these chips are there are no cheap motherboards yet, but for OEM's (large ones, dell, hp, lenovo) they make their own boards, so no problems there.

    They don't need to cut prices if they have a line up that does this well. AMD might be forced to cut prices to stay relevant in the market though.

    Intel has had AMD by the balls all this time, even with Ryzen, all Intel had to do is release something a little earlier which they planned for a 14nm CFL at least a year prior, actually 2 years prior cause validation alone takes 8 months to year added to that when it was taped out and designed..... gotta be more than that. If Ryzen didn't come so close, we would see a 10nm 6 core CFL coming out mid next year instead of an updated CFL 10nm 8 core.

    Its only been 7 months since Zen launched, that's not enough time and Intel just knee capped AMD's possible growth.

    Zen + can't come soon enough, and they have two things to fix, IPC which needs a 30% improvement and core clocks which they need to get to 5ghz when overclocking 25% improvement. Not easy to do. The time line they must do this in is by Mid next year cause that is when 8 core CFL is coming or there about. Intel doesn't need to do much at all architecture wise from the current CFL if they choose to and still remain competitive cause they have the node advantage 10nm vs 12nm (12nm is just a modified 14nm and its LLP)
     
    Last edited: Oct 13, 2017 at 12:20 AM
    Factum likes this.
  39. _mockingbird

    _mockingbird Limp Gawd

    Messages:
    314
    Joined:
    Feb 20, 2017
    Navi may be the last GCN architecture as RTG is looking forward to a non-GCN future.

    Work on Navi has been scaled back as RTG plans a brand new architecture or at least drastically different enough to be considered non-GCN.

    A major goal is to significantly reduce power consumption.

    Of cause, if this new non-GCN is not released on schedule, we may be looking at some stopgap GCN architectures.
     
  40. Gideon

    Gideon [H]ard|Gawd

    Messages:
    1,042
    Joined:
    Apr 13, 2006
    Yet paper does not work well in the cpu socket and Ryzen is still selling quite well on Amazon. Course every time you say AMD is screwed the stock surges, so I am good with that.
     
    JustReason and cybereality like this.
Thread Status:
Not open for further replies.