[VC] XFX R9 3X0 Leaks, Early April Delivery Indicated

Page 4 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
Not very optimistic about 390X in light of Titan X's release. NV seems pretty confident about what's to come from AMD camp. As an aside, I have a hunch that TSMC is playing favoritism as well. Reviews do not mention it (not because they write in bad faith but because they do not have information) but majority of Maxwell's efficiency gain probably has direct correlation with improved TSMC's 28nm. Qualcomm benefited hugely from that as well on mobile front last year.

One can dream of a 2nd coming of HD 5870. That card debuted @$370, shattering all performance records at the time by huge margins. Oh and the first DX10 compliant. A card like that from either camp might rekindle my enthusiasm on PC gaming.

Imo, even coming in at 95% of Titan X's performance at $699 would already make the 390X a great achievement. Think about it, AMD is basically rolling over GCN since Dec 2011 by enhancing the core fundamentals of the same architecture. In comparison, NV has separate teams that design new architectures every 3-4 years and they launch them every 2 years. I think NV really missed the mark with the Titan X. The price is wayyy too high and the cooler isn't good enough for those who overclock. That means the real comparison will be 390X vs. consumer AIB GM200 6GB.

Think about it another way:

If R9 390 non-X has 85-87% of Titan X's performance at $499-549, I would much rather get 2 of those than a single Titan. If AMD actually matches or beats the Titan, that's actually an embarassment for NV. Why? Because AMD isn't primarily in the business of making graphics and doesn't have the same financial or engineering resources. Right now price aside, the Titan X looks good against a 295X2 because if CF doesn't work, Titan X is 45-50%+ faster. However, if one is faced with a situation where if CF doesn't work, AMD's $500 card is 85-87% as fast, but when CF works, that setup is 40-50% faster than Titan X, I would personally find it impossible to consider a $1000 Titan X.

Don't forget that 5870 was slower than a 480, and much slower when overclocked/in tessellated game scenarios. That's why I really want to see GM200 6GB and 390 non-X/X parts. If 390 non-X can give us 85-87% of Titan X's performance at $499, who cares about the Titan for double the price!
 
Last edited:

beginner99

Diamond Member
Jun 2, 2009
5,210
1,580
136
I thought it was settled they're made at GloFo.

Which is another reason that rebrands are unlikely. If you port from TSMC to GF you can just as well improve the chips. It will be worthwhile for AMD simply due to the WSA. IMHO WSA is also the reason why large die GPU isn't a big deal for AMD.
 

5150Joker

Diamond Member
Feb 6, 2002
5,559
0
71
www.techinferno.com
Why? Because AMD isn't primarily in the business of making graphics.

Gonna add that to my signature, that has to be the quote of the decade here on AT.
roflmao.gif


Please refrain from trolling. If you want to dispute the facts, then do it.
-subyman
 
Last edited by a moderator:
Feb 19, 2009
10,457
10
76
Gonna add that to my signature, that has to be the quote of the decade here on AT.
roflmao.gif

Actually it's true.

AMD has always primarily been a CPU & Server company. When they got ATI they ruined it by going for value rather than raw market leader in performance. Their graphics division have been a small share of overall revenue for many years until recent times when their CPU & Server business tanked.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
Actually it's true. AMD has always primarily been a CPU & Server company.

AMD was a successful server CPU manufacturer between 2003 and 2006 mid when they reached 26% of the server market. After that they started losing market share gradually and after Intel introduced Nehalem the business started to shrink rapidly. Once Bulldozer launched the deterioration was rapid and by 2013 AMD was having a server market share of <3% . so basically negligible. For the past 1 year AMD is primarily being driven by their gaming business - game consoles, professional graphics (which is now at an all time high of 25% market share). so people should stop referring to the past. AMD might be able to revive its server business if Zen and K12 are competitive high performance CPU cores but thats still a huge question ?

When they got ATI they ruined it by going for value rather than raw market leader in performance. Their graphics division have been a small share of overall revenue for many years until recent times when their CPU & Server business tanked.

When AMD bought ATI they were in the midst of serious execution challenges with HD 2900XT. But they quickly turned it around with HD 4870 and HD 5870. In fact by Q3 2010 AMD was the largest GPU vendor (desktop + notebook combined)

http://www.xbitlabs.com/news/graphi..._on_Discrete_GPU_Market_Mercury_Research.html

But Nvidia quickly gained back the lost market share with GTX 500 series. Then with Kepler and Maxwell Nvidia dominated the notebook market as AMD's notebook products were just not competitive. :thumbsup:
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
I think his point is that AMD always goes for the CPU, in case of limited money to distribute in the R&D budgets. Something that seems to have burried their GPU division for good.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
I think his point is that AMD always goes for the CPU, in case of limited money to distribute in the R&D budgets. Something that seems to have burried their GPU division for good.

I don't agree. He does not have any proof to back his statement and to the contrary AMD is today surviving because of its game console chips. The businesses which are doing well are pro graphics, embedded. CPU business is the worst performing - Both CPUs and APUs combined. The GPU business was hit by Maxwell but the R9 3xx series will recover the lost market share and stabilize the GPU business. For the CPU business there is no hope till Zen/K12 launches.
 

ShintaiDK

Lifer
Apr 22, 2012
20,378
145
106
By the time the 300 series launches. AMD may only have 20% or less marketshare left. Not only do they need to stabilize. They also need to increase their marketshare by a factor of 2 or more. Something unheard of in their history. And you expect this to be realistic?
 

AtenRa

Lifer
Feb 2, 2009
14,001
3,357
136
Actually it's true.

AMD has always primarily been a CPU & Server company. When they got ATI they ruined it by going for value rather than raw market leader in performance. Their graphics division have been a small share of overall revenue for many years until recent times when their CPU & Server business tanked.

It was ATI that made that decision after the HD2900 fiasco, way before they were sold to AMD. They returned with smaller dies and higher efficiency with HD3870 and HD4870 and they continued that trend up until the HD7970.
 

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
By the time the 300 series launches. AMD may only have 20% or less marketshare left. Not only do they need to stabilize. They also need to increase their marketshare by a factor of 2 or more. Something unheard of in their history. And you expect this to be realistic?

If 300 series is fast, well priced, AND power efficient at least in the high volume SKUs, they could easily go up 10-20% total market share quickly. That's what drove down their marketshare afterall, those same characteristics delivered by Nvidia. I just think it's a bit foolish if they are deliberately, no technical reasons involved, waiting until June. But if they don't launch sometime in April it will most likely be the rumored Computex launch.
 
Last edited:

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
By the time the 300 series launches. AMD may only have 20% or less marketshare left. Not only do they need to stabilize. They also need to increase their marketshare by a factor of 2 or more. Something unheard of in their history. And you expect this to be realistic?

Nvidia will not be able to retain the seasonal gain of market share. Once AMD has the R9 3xx product stack out in Q2, AMD will reverse the losses and get back to its traditional 35% market share. amd could do even better if they take the GPU crown because the halo effect of the GPU crown improves the overall brand and boosts sales across all price segments. thats why Nvidia has been so successful as they held the GPU crown most of the time.
 

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
I don't agree. He does not have any proof to back his statement and to the contrary AMD is today surviving because of its game console chips. The businesses which are doing well are pro graphics, embedded. CPU business is the worst performing - Both CPUs and APUs combined. The GPU business was hit by Maxwell but the R9 3xx series will recover the lost market share and stabilize the GPU business. For the CPU business there is no hope till Zen/K12 launches.

Sure there is proof. Without going into in-depth financial technical analysis, it is impossible to conclude that AMD primarily survives and intends to survive off pure graphics, like NV does:

http://www.trefis.com/stock/amd/mod...09c261e1bb9b92be289e7935994576189&from=search

AMD does not derive most of its free-cash flow/EBITDA from Graphics. You are clumping semi-custom designs as graphics but that's not right. According to that logic Qualcomm is primarily a graphics firm because they manufacture millions of SOCs (APUs). When I am talking about pure graphics, it means selling graphics cards/graphics sub-systems to consumers in desktop/notebook/embedded form. The minute you talk about APU, you are already discussing CPU+GPU or SoC like Qualcomm.

By the end of 2015, 50% of AMD's business will come from non-PC sources. This change of direction does not come suddenly but it took years of re-aligning where to divert resources. If you have a limited number of engineers and R&D, you cannot allocate most of that towards graphics either. In % terms, there is no doubt that NV allocates more of its resources towards graphics than AMD does.

Think about it, AMD wouldn't even be wasting time on Zen, Carrizo, ARM CPU/SoC if it was primarily a GPU firm. That is because each of those investments is its own market. NV's business model is a lot less complex. They manufacture graphics chips and repurpose them to GeForce, Quadro, Tesla, with Tegra/Automotive a new growth arm.

In that sense AMD makes products that are a lot less inter-related. AMD is like a smaller Intel+NV combined -- a server/CPU/APU/graphics/custom design firm. Because of that, AMD cannot just throw 80% of its engineers and R&D/expenses on graphics only. That's why on paper, NV should win every single generation (and they have since HD2900XT). I never found this surprising and I am amazed some people do. That's why 390X matching or beating Titan X is an embarassment for NV. Not only does NV have more human capital, cash flow and other resources, but they cana afford to make 550-600mm2 die because they amortize those costs across so many product lines. We talked about this years ago when Fermi launched.

Look at how NV invented CUDA, their push for graphics compute, graphics in research/medicine/finance/deep learning/automotive. AMD is a lot more narrow with their graphics arm since they have way too many other investments like Zen. That is why for NV there is much more incentive to invest into graphics than there is for AMD. Even if AMD is good at graphics, that was never and still isn't the primary focus for the firm. It's certainly one of their key revenue streams, but not the most important one. For example, AMD might not even have won the Sony/XB1 business if it couldn't provide the Jaguar x86 APU. The Jaguar CPU design didn't just come from thin air; and NV has no such product. That already tells you that AMD and NV are different in terms of their strategy and how the 2 firms are structured. NV is not really an Intel competitor, but AMD is both an Intel and NV competitor. I think you forgot this because Bulldozer failed but it doesn't change the facts about who and what AMD was/is.

Let's look at it from another point of view. Imagine if the world had 10 consoles based on x86 APUs, not 2 main ones. AMD could easily become a company that makes custom embedded APUs for consoles. They wouldn't even need to sell PC or notebook graphics. NV can't just do that. Even how NV defines itself as a firm is all about visual graphics. AMD in no way defines itself as a graphics firm. In description about who AMD is, there isn't even a word graphics:

http://www.amd.com/en-gb/who-we-are/corporate-information

Vs.

http://www.nvidia.com/object/about-nvidia.html

NV = visual computing + graphics

While AMD and NV happen to be competitors, their long term visions are not the same. Because of that, how NV allocates its finite resources is different to AMD. That's why for me the fact that AMD has even kept up with NV since Phenom failure before Bulldozer, is nothin short of remarkable.

If we still had standalone ATI, AMD and NV, I have 0 doubt in my mind ATI would have been run completely differently than how AMD is run. Why? Because like NV, ATI was primarily a visuals/graphics firm. That in a nutshell is one of the reasons AMD is struggling so much --- their don't have enough resources to compete with both Intel in CPUs/APUs/servers and at the same time compete with NV in visual computing/graphics. If both firms have 8000-10000 engineers, how in the world could AMD just throw 80% of them towards graphics? Not happening! That's why both Intel and NV are winning their respective industry segments. AMD is trying to fight a 2-front war, while smaller than EITHER of its competitors!
 
Last edited:

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
Sure there is proof. Without going into in-depth financial technical analysis, it is impossible to conclude that AMD primarily survives and intends to survive off pure graphics, like NV does:

I never said that. I said the current reality is such that AMD is surviving because of their game console chips.

AMD does not derive most of its free-cash flow/EBITDA from Graphics. You are clumping semi-custom designs as graphics but that's not right. According to that logic Qualcomm is primarily a graphics firm because they manufacture millions of SOCs (APUs). When I am talking about pure graphics, it means selling graphics cards/graphics sub-systems to consumers in desktop/notebook/embedded form. The minute you talk about APU, you are already discussing CPU+GPU or SoC like Qualcomm.

AMD won the console contracts because they delivered a good gaming APU with lower cost than discrete CPU and GPU. The CPU core is important but only next in importance to the GPU architecture.

By the end of 2015, 50% of AMD's business will come from non-PC sources. This change of direction does not come suddenly but it took years of re-aligning where to divert resources. If you have a limited number of engineers and R&D, you cannot allocate most of that towards graphics either. In % terms, there is no doubt that NV allocates more of its resources towards graphics than AMD does.

There is no doubt about that.

Think about it, AMD wouldn't even be wasting time on Zen, Carrizo, ARM CPU/SoC if it was primarily a GPU firm. That is because each of those investments is its own market. NV's business model is a lot less complex. They manufacture graphics chips and repurpose them to GeForce, Quadro, Tesla, with Tegra/Automotive a new growth arm.

I disagree. Nvidia designs GPUs (for desktop, notebook, HPC) and mobile SoCs. Nvidia even designed a custom CPU core called Denver. Thats not cheap. Not many companies do custom CPU cores. Intel, AMD, Qualcomm, Apple are the others.

In that sense AMD makes products that are a lot less inter-related. AMD is like a smaller Intel+NV combined -- a server/CPU/APU/graphics/custom design firm.

You have got it all wrong. AMD is in the business of developing CPU and GPU IP and designing them into products for desktop, notebook, tablet, embedded, semi-custom and server market. AMD give equal importance to developing world class CPU and GPU cores.

If we still had standalone ATI, AMD and NV, I have 0 doubt in my mind ATI would have been run completely differently than how AMD is run. Why? Because like NV, ATI was primarily a visuals/graphics firm. That in a nutshell is one of the reasons AMD is struggling so much --- their don't have enough resources to compete with both Intel in CPUs/APUs/servers and at the same time compete with NV in visual computing/graphics. If both firms have 8000-10000 engineers, how in the world could AMD just throw 80% of them towards graphics? Not happening! That's why both Intel and NV are winning their respective industry segments. AMD is trying to fight a 2-front war, while smaller than EITHER of its competitors!

AMD's biggest problem is its last CPU architecture Bulldozer has destroyed the company's marketshare in all segments and made it very weak financially. I have no doubt that a strong CPU core from AMD will lead a long term revival and that in the long term the APU with HBM will definitely make the discrete GPU market a niche. :thumbsup:
 
Last edited:

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
You have got it all wrong. AMD is in the business of developing CPU and GPU IP and designing them into products for desktop, notebook, tablet, embedded, semi-custom and server market. AMD give equal importance to developing world class CPU and GPU cores.

No, I do not. What defines Apple? It's not iPod or iPad, it's the iPhone. Why? Because it by far generates the largest percentage of its revenues and cash flows. If we look at Nvidia:

1. Discrete Notebook GPU Revenues - $1.68B
2. Discrete Desktop GPU Revenues - $1.17B
3. ProGraphics Card Revenues - $1.06B
4. Tegra and Other Revenue - $0.752B

That means NV derives nearly $4B primarily from graphics and just $0.752B from other sources. Tegra actually is a negative cash flow revenue stream, which means NV is nearly 95% a Pure Graphics Card Firm.
http://www.trefis.com/stock/nvda/mo...IDER_7bca48f08d5a591a64217420106fa4e0a637fe94

vs. AMD

By end of 2015, 50% of AMD's revenues are projected to come from non-PC gaming sectors -- meaning graphics will comprise WAY less than 50% of AMD as a firm.

By end of 2014, Computing and Graphics division accounted for just 57% of AMD's revenue. You keep wanting to clump Enterprise, Semi-Custom and Embedded as "Graphics" but it's not.
http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-reportsannual

You do not need to design desktop/laptop/workstation PC graphics cards to design Semi-custom and embedded graphics processors/APUs. As I already said companies like Samsung and Qualcomm are already doing that and no one in their right mind would call those "primarily GPU driven firms."

No one disputes that graphics are important for AMD, but it definitely isn't the predominant/primary focus of the firm. For Nvidia, it most definitely is today. I don't know why you get offended that AMD is not primarily a graphics card company because in the eyes of investors, shareholders and analysts, it most definitely isn't either. Heck, NV doesn't even compete with Intel CPUs in any way shape or form - AMD does! Another way to look at it is AMD could stop making ALL PC graphics for laptops/notebooks/workstations and still survive as long as it gets design wins for XB2/PS5 and Zen is successful. If NV stops making graphics, the firm is bankrupt right away. That's proof right there that NV is primarily a GPU business while AMD is not. You were right that graphics occupied a far larger portion of AMD's revenues in the past but it's no longer true. AMD is surviving today not because of graphics, but despite of having a graphics division that barely makes any $$. Obviously, AMD will continue to design PC gaming graphics simply because that's a very large market but if Zen is successful, their ventures with MediaTek and so on take off, AMD doesn't absolutely require desktop/laptop PC graphics to survive, but NV does. Without desktop/notebook PC graphics card sales, NV would never be able to afford to subsidize the R&D required to produce only Tesla and Quadro GPUs. JHH said so himself when he launched Fermi and discussed how large monolithic GPU die ties into the firm's entire strategy for GeForce/Tesla/Quadro. It's actually one of the primary reasons why NV can afford to make 550mm2+ die gen after gen because it can amortize the costs across at least 3 different product lines.

Again, I think you entirely missed the other parts of my post where I touched upon AMD having a requirement to allocate a large amount of financial and human capital/engineering hours resources to huge projects like Zen, Carrizo, etc. NV doesn't have non-GPU projects of that scale. What the hell is Denver? Project Shield? Please. These are just side projects for JHH. Zen is not some side project but has the potential to sell millions of products when it launches; and if it fails would have dramatic implications for AMD's future. If Tegra fails or Project Shield fails, NV will just brush it aside as another risk they took. No big deal. Any time NV has strayed away from GPUs, it mostly fails, but if AMD fails with Zen, it could be catastrophic! How many people care about K1 and X1 SoCs they made? If AMD actually succeeds with making a good x86 CPU, it has the potential to double or triple its revenues. NV has no such opportunities with graphics in the short term. That's why AMD has to continue risking resources to invest into CPUs/APUs because it's not a company solely focused on visual computing/graphics. And for that reason it's remarkable to me that AMD even manages to compete with NV when NV re-purposes the same GeForce chips for 3 different product lines that generate 95% of its cash flows...
 
Last edited:

Head1985

Golden Member
Jul 8, 2014
1,864
689
136
By the time the 300 series launches. AMD may only have 20% or less marketshare left. Not only do they need to stabilize. They also need to increase their marketshare by a factor of 2 or more. Something unheard of in their history. And you expect this to be realistic?

I must agree with this.IN q4 2014 they have only 24%.Now they are i think at 18% and they still need 2.5 months more to release maxwell competition.Thats
9 months without competition vs maxwell.Thats longest in GPU history.Maxwell GTX970/980 was released 19.9 2014!!!!!!!!
By time q4 2014 Nv released GTX960 and now TITAN.

I think there is only one way to save AMD.
R300 series must be faster than NV counterparts and a LOT cheaper
I mean 500 for 390x -Titanx
400 390-cutdown GM200
350 380x-GTX980
300 380-GTX970

If 300series is overpriced like HELL just like NV cards AMD never gain market share and is DOOMED.
 
Last edited:

Vesku

Diamond Member
Aug 25, 2005
3,743
28
86
If 390X beats TitanX it doesn't need to be $500 to sell well, $650-800 would be very attractive to the kind of people who buy the fastest single GPU available and aren't brand locked to Nvidia.

Generally agree with the rest except 390 could be $450-550 if it's only very mildly cut down like 7950 and 290.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
AMD

By end of 2015, 50% of AMD's revenues are projected to come from non-PC gaming sectors -- meaning graphics will comprise WAY less than 50% of AMD as a firm.

By end of 2014, Computing and Graphics division accounted for just 57% of AMD's revenue. You keep wanting to clump Enterprise, Semi-Custom and Embedded as "Graphics" but it's not.
http://ir.amd.com/phoenix.zhtml?c=74093&p=irol-reportsannual


You do not need to design desktop/laptop/workstation PC graphics cards to design Semi-custom and embedded graphics processors/APUs. As I already said companies like Samsung and Qualcomm are already doing that and no one in their right mind would call those "primarily GPU driven firms."

No one disputes that graphics are important for AMD, but it definitely isn't the predominant/primary focus of the firm. For Nvidia, it most definitely is today. I don't know why you get offended that AMD is not primarily a graphics card company because in the eyes of investors, shareholders and analysts, it most definitely isn't either. Heck, NV doesn't even compete with Intel CPUs in any way shape or form - AMD does! Another way to look at it is AMD could stop making ALL PC graphics for laptops/notebooks/workstations and still survive as long as it gets design wins for XB2/PS5 and Zen is successful. If NV stops making graphics, the firm is bankrupt right away. That's proof right there that NV is primarily a GPU business while AMD is not. You were right that graphics occupied a far larger portion of AMD's revenues in the past but it's no longer true. AMD is surviving today not because of graphics, but despite of having a graphics division that barely makes any $$. Obviously, AMD will continue to design PC gaming graphics simply because that's a very large market but if Zen is successful, their ventures with MediaTek and so on take off, AMD doesn't absolutely require desktop/laptop PC graphics to survive, but NV does.

Your problem is seeing AMD as a CPU or GPU manufacturer. the dedicated CPU as we know is more or less dead. AMD's future lies with APUs or SoC in all segments - desktop, notebook, tablets, embedded, semi-custom and even HPC servers. The concept of dedicated CPUs will only be relevant for certain market segments in server market and maybe ultra high end gaming PCs. For everything else it will be the APU :thumbsup:

AMD's 25x20 vision is all about heterogeneous computing. Its not long before we will see Zen based APUs cross the 1 TFlop barrier and maybe even crash past the 2 TFlops mark with HBM. The future is all about multi TFLOP SoCs.

https://translate.google.com/transl...articles/2015/03/04/amd_pccluster/&edit-text=

http://news.mynavi.jp/photo/articles/2015/03/04/amd_pccluster/images/005l.jpg

http://news.mynavi.jp/photo/articles/2015/03/04/amd_pccluster/images/005l.jpg

Going forward even in the HPC market AMD will bring HPC APUs with HBM as is evidenced by AMD's vision roadmap. So stop thinking of AMD as a CPU company. You are living in the past. :)
 
Last edited:

RussianSensation

Elite Member
Sep 5, 2003
19,458
765
126
I think there is only one way to save AMD.
R300 series must be faster than NV counterparts and a LOT cheaper
I mean 500 for 390x -Titanx
400 390-cutdown GM200
350 380x-GTX980
300 380-GTX970

This is something I've disagreed with Silverforce11 on. AMD doesn't even need to sell a single R9 390/390X card to gain market share or be profitable. They need a very healthy low-end and mid-range desktop line-up and top-to-bottom amazing laptop GPUs. Conversely, if R9 390/390X are a slam dunk, but AMD has a horrible mobile dGPU R9 300M product stack/launch strategy and the low-end and mid-range R9 300 desktop cards are worse than 750/750Ti/960/970/980, then it makes no difference if R9 390X is 50% faster than the Titan X for $600.

I hope Lisa Su also understands by now that if you want to return AMD's image to the glory days of ATI's brand image, you do not price a card as fast as the Titan X for $499 just to gain market share. AMD can use 390X as a statement for $649-699 and price 390 non-X with 85-87% of Titan X's performance for $499. This way, they kill 2 birds with 1 stone: the price/performance gamer is happy and AMD starts to slowly re-establish the Radeon image as a premium brand.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
I think there is only one way to save AMD. R300 series must be faster than NV counterparts and a LOT cheaper

I mean 500 for 390x -Titanx
400 390-cutdown GM200
350 380x-GTX980
300 380-GTX970

If 300series is overpriced like HELL just like NV cards AMD never gain market share and is DOOMED.

oh another AMD doom prediction. no need to sell LOT cheaper. AMD will do very well if R9 390X beats Titan by 10% and launches at USD 800 with 8 GB HBM. same for R9 380X beating GTX 980 by 10% and launching at USD 400. same for R9 370X launching at USD 180 - 200 and beating GTX 960 by 5-10%.

Remember Titan-X is a fully enabled GM200 while the original Titan was not. The Titan-X is also clocked much higher and draws more power than 780 Ti. Nvidia's wiggle room is much lesser this time around. If R9 390X beats Titan-X by 10% then R9 390 will be competing against Titan-X. If R9 390 is at USD 700 and trading blows with Titan-X at USD 1000 the brand agnostic gamer will pick AMD. There are few who will buy Nvidia no matter what. So no point bothering about those people. :thumbsup:
 

DownTheSky

Senior member
Apr 7, 2013
787
156
106
oh another AMD doom prediction. no need to sell LOT cheaper. AMD will do very well if R9 390X beats Titan by 10% and launches at USD 800 with 8 GB HBM. same for R9 380X beating GTX 980 by 10% and launching at USD 400. same for R9 370X launching at USD 180 - 200 and beating GTX 960 by 5-10%.

Remember Titan-X is a fully enabled GM200 while the original Titan was not. The Titan-X is also clocked much higher and draws more power than 780 Ti. Nvidia's wiggle room is much lesser this time around. If R9 390X beats Titan-X by 10% then R9 390 will be competing against Titan-X. If R9 390 is at USD 700 and trading blows with Titan-X at USD 1000 the brand agnostic gamer will pick AMD. There are few who will buy Nvidia no matter what. So no point bothering about those people. :thumbsup:

I fully agree. If the performance is sufficient, there'd be no point for AMD to price 390 so low they'd make no money on it.
 

Qwertilot

Golden Member
Nov 28, 2013
1,604
257
126
If nothing else, the last few months have surely shown that severely cutting prices on high end GPUs to try and compete a the mid end just doesn't work :) They need proper mid end stuff. Especially for notebooks, but desktops too.

NV well placed to respond of course as they've had chunks of the market to themselves for a while now (6 months for the 970/80 now don't forget) but much better to make them.

I am a bit worried about AMDs timing vs the 14/16nm die shrink. They surely don't want to leave Pascal the same sort of scope Maxwell has had to pummel them with, so the 3xx series might not last terribly long.
 

raghu78

Diamond Member
Aug 23, 2012
4,093
1,475
136
If nothing else, the last few months have surely shown that severely cutting prices on high end GPUs to try and compete a the mid end just doesn't work :) They need proper mid end stuff. Especially for notebooks, but desktops too.

NV well placed to respond of course as they've had chunks of the market to themselves for a while now (6 months for the 970/80 now don't forget) but much better to make them.

I am a bit worried about AMDs timing vs the 14/16nm die shrink. They surely don't want to leave Pascal the same sort of scope Maxwell has had to pummel them with, so the 3xx series might not last terribly long.

AMD will have a full year before the first 16/14nm GPUs hit the market. Nvidia will come out with 16 nm FinFet GPUs in Q2 2016 and AMD will have their first 14nm FinFet GPUs in the same time with HBM2. The 2015 FINFET wafer supply would barely meet the iPhone and iPad volume requirements. btw I am talking about both TSMC and Samsung put together. Samsung will be manufacturing their 14nm Exynos chips for their own needs too. Then Qualcomm is waiting in queue for 16/14nm wafer allocation for their next gen Snapdragon 820. Nvidia will get FINFET wafer allocation only in Q1 2016. same for AMD at GF 14nm. Nvidia's lead with Maxwell could quickly turn into a disadvantage when AMD launches R9 390 and R9 380 series with HBM. The first mid range Pascal is not likely to ship before late 2016. The first FINFET chips from Nvidia are Maxwel dumb shrinks. So Nvidia could have a tough road ahead for the next 18 months.

The last time AMD got first to a new memory standard with HD 4870 and GDDR5 they hammered Nvidia in GPU market share and it took Nvidia close to 21 months to get GDDR5 on their flagship GPU (GTX 480). AMD's HD 5000 series was very successful and AMD led the overall GPU market share (desktop + notebook combined) in Q3 2010

http://www.xbitlabs.com/news/graphi..._on_Discrete_GPU_Market_Mercury_Research.html

AMD has a real opportunity gain a lot of market share especially in notebooks where they are pretty much non existent now at the high end.
 

el etro

Golden Member
Jul 21, 2013
1,581
14
81
If AMD 390x outperform TitanX at $700, Fiji will outsell GM200 automatically.
 

Qwertilot

Golden Member
Nov 28, 2013
1,604
257
126
Not sure it'll be quite that positive in terms of lifespan. Yes, likely for direct replacements but you can see NV getting a '1050(ti)' out early next year. Not huge volume of course.

Once that happens it'll get rather harder to sell any current gen stuff. We'll see :)

Notebooks should be more positive for them all round with HBM coming - surely they'll be able to produce an APU with a good enough iGPU to get some decent traction in the gaming notebook market?!!?
 

Stuka87

Diamond Member
Dec 10, 2010
6,240
2,559
136
I must agree with this.IN q4 2014 they have only 24%.Now they are i think at 18% and they still need 2.5 months more to release maxwell competition.Thats
9 months without competition vs maxwell.Thats longest in GPU history.Maxwell GTX970/980 was released 19.9 2014!!!!!!!!
By time q4 2014 Nv released GTX960 and now TITAN.

I think there is only one way to save AMD.
R300 series must be faster than NV counterparts and a LOT cheaper
I mean 500 for 390x -Titanx
400 390-cutdown GM200
350 380x-GTX980
300 380-GTX970

If 300series is overpriced like HELL just like NV cards AMD never gain market share and is DOOMED.

The 970/980 are a very small percentage of GPU sales. The high end cards always have the smallest sales percentages. Thinking those two cards are going to drop AMD's market share any noticeable amount is narrow sighted. The vast majority of GPU sales are low-mid range OEM sales.

But even at that, AMD's cards don't have to be cheaper and faster. But it would get them more sales sooner.