AMD's GPU market share drops again, even after the release of Fury X

[Killer7xx]

In the desktop market, AMD should just focus on performance APUs and high end cards and then release mid range cards which are rebrands of last gen's high end or cutdown versions of their current high end cards. That would save them a lot in R&D costs.

Also, I'd like to see more focus on the mobile GPU market from them. It's an area which could benefit hugely from HBM and APUs, it's currently an Intel/Nvidia stronghold but that could easily change in the future. DX12/Vulkan may also bring other players into the mix such as PowerVR or Qualcomm since GPU performance in games will be less driver dependent.

 

[CAT-THE-FIFTH]

It does seem to be as innovation and progress goes down, (meaning less demand) that prices are going up to offset so they can maintain profit margins.

Bit boring for the consumer, I think things will get more exciting next year. With finally get a die shrink and a real jump in tech with HBM 2.0 etc. Hopefully it will be like a reset in the DGPU space, as the long road on 28nm has led to massive stagnation.

Yeah,but the main issue is whether AMD or Nvidia will seek to give a big performance bump or use this to increase profit margins.

Remember when 28NM came along,there was complaining that the cost per wafer was more now,and I remember some Nvidia charts saying future nodes were even more expensive.

So looking at the decreasing dGPU sales,I expect there will be a move to make sure the GPUs which are sold make more money - hence I think prices will still be high,with one eye on just doing enough to make sure Intel does not get a bigger foothold at the lower end.

I expect there will be more milking going on,as there is no guarantee that 14NM/16NM will be problem free and even the lack of noise on what is replacing them is worrying.

 

[CAT-THE-FIFTH]

bookmarked for when the Q3 and Q4 results come out

I think we need to bear in mind that both nvidia and AMD had product launches late in Q2, so people "not buying" in Q2 is very likely to be because people were waiting to see what happened.

If CAT is right then we shouldn't see any kind of uplift in the Q3 and Q4 results.

So using your logic,that means more discrete card sales will be sold in Q3 2015 and Q4 2015 than Q3 2014 and Q4 2014??

So please tell me how that will happen,when the GTX970 and GTX980 were launched at the same time last year??

You are another "enthusiast" in denial of what is happening.

Its happening consistently now for years. Discrete card sales are down.

Edit!!

Another one from JPR:

http://jonpeddie.com/publications/market_watch

Discrete GPUs continue to slip as embedded GPUs in the CPU and APU bring better performance for free. The erosion of the low end of the market by tablets seems to have subsided, and even tablet sales were off for the second quarter in a row.

It seems andybird123 knows better than JPR. Time for you to take over!! Earn the mega-bucks!!

Oh,another one which might be easier for you to digest.

Right,the GTX980TI and Fury are going to rescue the discrete market!!

 

[Boomstick777]

Yeah,but the main issue is whether AMD or Nvidia will seek to give a big performance bump or use this to increase profit margins.

Remember when 28NM came along,there was complaining that the cost per wafer was more now,and I remember some Nvidia charts saying future nodes were even more expensive.

So looking at the decreasing dGPU sales,I expect there will be a move to make sure the GPUs which are sold make more money - hence I think prices will still be high,with one eye on just doing enough to make sure Intel does not get a bigger foothold at the lower end.

I expect there will be more milking going on,as there is no guarantee that 14NM/16NM will be problem free and even the lack of noise on what is replacing them is worrying.

Yeah in that case, expect higher prices again. Leading to less overall sales and more rising costs per, to offset and keep their profit margins.

I'm sure there will still be a decent midrange option, the higher you go up the less bang for buck you see.

Such a long time on same node is unique though. We're right at the end of massive stagnation, so bound to see market tapering off now.

The flipside off this for AMD is that once Zen APU is off the ground, with decent gaming capable graphics integrated, AMD could see profits coming from that market over DGPU. Meaning although things aren't great now, they could be poised for a better future.

 

[Mauller]

I expect there will be more milking going on,as there is no guarantee that 14NM/16NM will be problem free and even the lack of noise on what is replacing them is worrying.

The main reasons for a lack of noise on 10nm processes is because we have started to hit the wall when it comes to silicon node shrinks.

The main thing that will see a massive increase in performance is alternate materials. But then it depends on who will build it first and invest. Since the majority of manufacturers don't want to build new production lines unless the tech is worth it. So unless they can implement a new material into the current cmos process then I think we will be on 14/16nm for a while.

 

[andybird123]

I expect there will be more milking going on,as there is no guarantee that 14NM/16NM will be problem free and even the lack of noise on what is replacing them is worrying.

no noise at all...

http://www.kitguru.net/components/a...lidation-chips-with-quad-core-arm-cortex-a57/
http://www.androidauthority.com/tsmc-and-samsung-10nm-finfet-623749/

and congrats on showing graphs of APU sales in a discrete card conversation
people are keeping cards for longer but that doesn't mean the GPU market is dying or collapsing or whatever you want to call it

 

[CAT-THE-FIFTH]

no noise at all...

http://www.kitguru.net/components/a...lidation-chips-with-quad-core-arm-cortex-a57/
http://www.androidauthority.com/tsmc-and-samsung-10nm-finfet-623749/

So when is the first 10NM chip making meant to be shipping? 2016? 2017? 2018?? 2019??

Test 14NM/16NM chips have been out for a while and only now have they entered any level of large scale volume production - the first being a small phone SOC made by Samsung which is 78MM2.

Plus Samsung only produces chips for itself - they licensed the tech/colloborated with GlobalFoundries??

Heard,anything about them actually doing much,apart from delay after delay??

The first large chips,ie, GPUs will be lucky to actually enter large scale production and availability by summer,ie, a year after the first phone SOC on 14NM made by Samsung.

Plus 20NM was cancelled by TSMC and so was 32NM. Every other node they have done for the last few years has hit problems.

Intel despite having issues with 14NM,they are producing 170MM2+ chips running at 3GHZ+ NOW.

and congrats on showing graphs of APU sales in a discrete card conversation
people are keeping cards for longer but that doesn't mean the GPU market is dying or collapsing or whatever you want to call it

Well,JPR and every other analyst firm does not agree with you and they even state the same,and as usual another enthusiast is in denial.

But then you seem to confuse YoY metrics with QoQ,so just to clarify do you think there will be a massive uplift in graphics card sales due in Q3 2015 and Q4 2015 over the same time periods in 2014 and do you expect 2015 graphics card sales to be more than 2014??

Yes/No?

 

[andybird123]

wait wut? 20nm wasn't cancelled lmao, Nvidia and AMD decided it was better to skip and go straight to 16, but it wasn't cancelled

TSMC and Intel have said they are expecting to start producing 10nm volume in 2017 - TSMC Q1 which I'll take with a pinch of salt yes.

 

[D.P.]

Imagine how much the high end nvidia cards would cost without a competitor. I'd have to move to console gaming!

Have a look at the current prices to get a good idea. Seriously, Nvidia wont increase price much if AMD disappear, they already largely ignore AMD because AMD is simply not competing at the same price-performance level (http://techreport.com/r.x/radeon-r9-fury/value-99th.gif).


Without AMD Nvidia will continue to do what it has always done, try and maximize profits. That doesn't mean increasing sale prices, that means finding the optimum balance between price and sales based on supply and demand, something they already do. For Nvidia to generate increased sales they will still have to innovate and produce new GPUs with significant performance increases that people are willing to pay for.

What you would see though is the end of paper launches and wood screws. Release drivers will be more optimal. There would be time to stockpile cards ready for a hyped up launch. GPUs that didn't hit expected performance may go for a re-spin rather than getting rushed out the door with higher clocks and a giant cooler (5800 Ultra, 480). Paper launches and rushed delivery is a result of competition, without competition Nvidia could release products when they are ready. So in that sense progress may slow down a little but not significantly.


Remember, Nvidia also sees competition form Intel, infact intel is a much bigger concern to Nvidia than AMD is. Intel already dominate the low end market through IGP, and that is now eating sales of low end discrete cards. Intel also competes against Nvidia for HPC with the Intel Phi. That alone will see Nvidia push GPU tech forwards

 

[D.P.]

wait wut? 20nm wasn't cancelled lmao, Nvidia and AMD decided it was better to skip and go straight to 16, but it wasn't cancelled

TSMC and Intel have said they are expecting to start producing 10nm volume in 2017 - TSMC Q1 which I'll take with a pinch of salt yes.

yeah, 10nm is going to be OK but in general node shrinks are getting harder, which is not surprising.

It is the sub 10nm that the experts are worried about.

 

[Boomstick777]

Nvidia also sees competition form Intel, infact intel is a much bigger concern to Nvidia than AMD is. Intel already dominate the low end market through IGP, and that is now eating sales of low end discrete cards.

^^ Intel's own IRIS graphics line is improving extremely fast, the latest Broadwell i7 5775C and the i5 IGPU lay waste to AMD's best IGPU graphics but obviously the chip costs more, but is worth is as CPU performance is much much higher than AMD's as well.

With everything shrinking down, IGPU might be the future. Maybe AMD has more potential there than in DGPU moving forward imho, not just PC but mobile / console etc..

The traditional bulky desktop PC will eventually fade away for consumers (Not workstations) imho. PC will evolve and always be around though in one for or another.

 

[Mauller]

With Intel using eDram and future AMD APUs using HBM. The low end discrete market will more than likely vanish in its current form.

The kaveri parts have very capable Gpus, they are purely held back by memory bandwidth. And intels newest parts are more capable than ever.

The down turn in discrete is more than likely the effects of the above. Mid to high end parts sell in far lower volume than lower - low mid parts.

 

[Boomstick777]

With Intel using eDram and future AMD APUs using HBM. The low end discrete market will more than likely vanish in its current form.

The kaveri parts have very capable Gpus, they are purely held back by memory bandwidth. And intels newest parts are more capable than ever.

The down turn in discrete is more than likely the effects of the above. Mid to high end parts sell in far lower volume than lower - low mid parts.

Agreed,

Kaveri is ok as a budget PC for general use / web gaming. But falls flat on it's face with actual demanding PC games.

Like you say HBM, removing that memory bottleneck and die shrinks allowing for more shaders could allow the next wave of APU's based on Zen to remove the need for many people to even have a DGPU.

Could see future APU's being extremely popular in laptops, future consoles. Gaming PC's etc. As DGPU market shrinks IGPU market grows and AMD seems poised to take advantage of that going forward.

However next year DGPU could buck the trend and generate more demand if the products are true advancements, some exclusive PC games would be a incentive for the need for GPU as well.. Either way AMD will be ok, just right now the market is tapering off after a long boring period on 28nm lol. Market needs a shakeup for sure.

 

[CAT-THE-FIFTH]

wait wut? 20nm wasn't cancelled lmao, Nvidia and AMD decided it was better to skip and go straight to 16, but it wasn't cancelled

TSMC and Intel have said they are expecting to start producing 10nm volume in 2017 - TSMC Q1 which I'll take with a pinch of salt yes.

GF bulk 20NM was cancelled(which AMD was supposedly going to use) and TSMC 20NM had problems it appears with larger chips. The only TSMC 20NM chips made were tiny chips.

Plus again do you think Q3 2015 and Q4 2015 graphics cards sales are going to be more than Q3 2014 and Q4 2014 sales??

Plus do you think 2015 graphics sales will out do 2014 ones??

AMD is also in a more precarious situation overall due to their exposure to Intel over many fronts IMHO,so if Zen does not fire on all cylinders,AMD will be finished!

 

[queamin]

Trouble Nvidia needs Intel more than Intel needs Nvidia.

 

[Boomstick777]

Trouble Nvidia needs Intel more than Intel needs Nvidia.

Right now Intel and Nvidia have a Symbiotic ecosystem. Nvidia offer best graphics, Intel offer best CPU. Over time though with Intel's rise in IGPU negating the need for many people to bother with DGPU maybe that ecosystem could be altered. Although Nvidia have stated they are not worried about IGPU as they said they would simply raise the bar on lower end DGPU to compete.



Would love to see an Intel / Nvidia hybrid all in one chip. Would buy that in heartbeat.

Maybe in the future Nvidia will look at offering CPU/GPU solutions who knows?

 

[queamin]

Nvidia needs Intel cpu/apu to run their cards, intel don't really need the cards and I would think gaming is quite small in Intel eyes.

 

[andybird123]

Well,JPR and every other analyst firm does not agree with you and they even state the same,and as usual another enthusiast is in denial.

But then you seem to confuse YoY metrics with QoQ,so just to clarify do you think there will be a massive uplift in graphics card sales due in Q3 2015 and Q4 2015 over the same time periods in 2014 and do you expect 2015 graphics card sales to be more than 2014??

Yes/No?

they haven't said the dGPU market is "collapsing"
it has shrunk yes, but that isn't the same as collapsing
a small reduction over time isn't collapsing, as long as at least one company is making millions of dollars a year serving that market there will continue to be a market

Nvidia's turnover and profit from FY 2011 to 2015

2011 2012 2013 2014 2015
3.54B 4B 4.28B 4.13B 4.68B
253.15M 581.09M 562.54M 439.99M 630.59M

Yeah, collapsing

 

[queamin]

If more people used cards than IGPU then yes but I would think the IGPU usage out weight cards by a lot.

 

[queamin]

What happened in 4th quarter 08 and 1st quarter of 09 to make them dip so bad?.