• Happy holidays, folks! Thanks to each and every one of you for being part of the Tom's Hardware community!

AMD CEO Lisa Su Interview: Confident In Next Graphics Launch, Zen's Success Is Key

Page 4 - Seeking answers? Join the Tom's Hardware community: where nearly two million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
Status
Not open for further replies.
Yeah, who knows what they's done with marketing expenditures. Fortunately, we won't have to wait very long to see what this next announcement puts out. As always, time will tell.
 
I am very glad that AMD apparently realized the dire need to call Marketing to heel. Exaggeration simply doesn't work in the tech world, where there are actual objective measurements that call BS.
 
I guess all we can do is to hope for some new innovations and a good start of the line (at least the really new cards with HBM) and their new processor lines. If AMD will be gone some day (and it really doesn't look quite bright at the moment)PC-Gaming will become extremely expensive and there won't be much more innovations to come.,
 


We'd all be sad to see such a significant chip company with almost 40 years of history go (I know I would). But I'm not sure innovation would stagnate upon AMD's dissolution. Nvidia and Intel would have simply won today's R&D arms race. But Intel, AMD, and Nvidia have already defined an industry whose rewards stem from progress and advancement. Opportunities for AMD's predecessor to emerge would arise if a surviving company grows lazy and leaves neglected demand on the table. Or someone might finally replace silicone as a semiconductor and create a new market alltogether. Who knows? What is certain, though, is that without that kind of carrot-on-a-string mentality, there will be no tech market to sustain any of these companies. Consumers want what they don't have, and that makes selling them things they already have very difficult.
 
AMD needs to fix its CPU lineup. You can't build an armada without a flagship. They need real quad core cpus competitively priced, then they need them at low heat, then at low power. People want cheaper notebooks. They don't exist, so people buy smartphones. People pay $700+ for smartphones because they make phone calls and are computers. Notebooks are too expensive for their performance and function.
 


Is this why we seeing 390x (aka 290x) rumors pic that have same cooler as fury (Fiji) (water cooled)?..
interesting...
will have to wait to see either 3xx straight re-brand or they have some improvement someway..

For Zen. I'm agreed, AMD should not try to beat Intel tops, just give us some decent performance for the money,
clearly last time is failure..
(when later, the new revision and the price drop, it getting better, but it's kinda to late..)

edit: they need to survive first before thinking beating Intel..
unless they have some super secret tech in zen that can beat Intel,
I highly doubt that in their current state..
 


All valid points, which is why I said compelling, not irresistible. Sure the cost is higher, but it could also be argued in a couple of ways: 1) That (aside from normal manufacturing defect rates) with Intel you are going to get a quality product without integration problems. My FX-4100 and FX-6100's had nothing but problems (sure it was sorted on later releases, but that's not usually a problem with Intel) and... 2) The problem with both of them in the past is that one brand or the other had more "horsepower" (either CPU or GPU) but not both at the same time or at least not enough to completely eliminate the need for a discrete solution.

So as far as I see it, Intel is stepping up their graphics game (lol) and possibly hinting at where they're going down the road. AMD needs to really make a vast leap in manufacturing tech and/or design to compete with Intel's CPU side of the equation... all the while making the next generation considerably more power efficient. It seems like quite a tall order to me. Am I wrong?

Don't misunderstand me though. I do want AMD to do exactly that. I want them to blow this competition wide open again, mostly because it's good for the segment, but also to give them another try this time around. If they can do all those things and still keep the prices competitive or better, they could have a winner.
 
I have high hopes for Zen, but after being burned by a launch purchase of the Pile Driver, I have to say I'm skeptical. I wonder though, will Zen still reflect the "affordability" of the past AMD CPU lineups with regards to Intel's line-up. I have no favorites, but it would be nice to see other options on the table for desktop gaming CPU's, and more affordable options would be even better.
 
AMD u are slowly withering away. your HBM will not save you. Nvidia are better and there timing of releasing new products are better and the products are not re branded HD7xxx. im switching teams. Your failure to release information and benchmarks means you can not match Nvidia.
 


That solves it, then. AMD will read this and officially close :lol:
 
Each time someone calls one entire company better than another company (and not a concrete aspect of a concrete product) i'm out, sorry..
 


Not true at all. First off, remember that AMD makes the graphics cards in Xbox One, PS4, and Wii U, and makes the CPUs in Xbox One and PS4 (Wii U's is made by IBM). Right there is one thing that keeps AMD rolling with the cash. The console makers tend to favor AMD to Intel. Secondly, I don't see how you can blatantly take AMD's new and innovative memory technology, HBM, and say, "it won't save you". The first question is whether they need to be saved. With 3-4 year old graphics card architecture that still competes with Nvidia's newest and still performs better frequently (280x > 960) it is most certainly not a failing company.

Rebranding is very smart of AMD. First off, they did not rebrand everything, as the Fury X runs on a completely new architecture. Also, if AMD's current cards still perform on-par with Nvidia's or even better with price-to-performance, the rebrands' 10-15% improvement or so will still benefit them. They're not behind at all, so they don't need to release something that has a 40% improvement.
 


The console makers went with AMD out of costs more than anything. Intel would have cost too much and trying to keep with custom chips from IBM was too expensive.

AMD is still losing cash which is not good.

HBM was proposed by AMD and Hynix but I am willing to bet Hynix did all the work, AMD is just adopting it first. AMD doesn't have the cash for the R&D required to develop something completely new. As well HBM is a specialized application of Samsungs Wide I/O memory tech.

Fiji XT is not a completely new uArch. It is still GCN, 1.2 to be exact, so it is very similar to Hawaii XT (290) and Tahiti XT (HD7900).

The rebrands will not see a 10-15% performance improvement over their older counterparts. The 390X is supposed to be clocked at 1070MHz and if true is not that fast since most R9 290X after market cooled GPUs can hit 1100 easily. The memory clock boost might help but again the manufactures have already done this with their R9 290s.

Fury is the only thing worth considering if you have a R9 290 right now. That or a GTX 980.
 


That hasn't actually turned out to bee a good thing. Cash is meaningless when cash in = cash out. Many, myself included, would argue that **this** is the cause of AMD's current situation. In order to **win** the console market, chip vendors have to slash margins to miniscule amounts. And while on paper, it might *look* profitable, it must be realized that not all actual costs are always reflected in the cost to make a product. Too much of AMDs management and "brain" assets are tied up in a venture that had a high dollar but little profit value. AMD's R & D investment in the past year was the lowest in the company's history .... whereas nVidia spent over 30 cents per dollar of income. That bodes ill for future product releases.

As for HBM, 1) it's not AMDs technology and 2) it as not been shown to be a notable performance advantage at this time tho it will undoubtedly be in the future. A reasonable corollary is that AMDs wide bandwidth gives it an advantage at 4k, but with 4k's market saturation on the gaming market at just 0.6%, (6 in 1000) [Edit 0.06% - 6 in 10,000], it hasn't been able to bring home a financial shot in the arm. Seeing the leaked press releases for the new Fury, it clearly is aiming at the 4k market .....

As far as AMDs financial condition, that is hardly arguable .... the company is in dire financial straits. Their position\in the discrete card market is most accurately reflected in one simple statistic.

There are 2 GTX 970s (3.05% of market) in use for every single R7 (0.52%) or R9 (0.99%) series (all 16 cards) cards in use combined .... and despite the fact that those R7/R9 series has been on the market twice as long. The 960 has outsold (0.78%) all R7 cards combined 1.5 to 1.0 ... so has the 730 (0.58%)

Cost here is going to be quite a quandary ....I expect we will see "Deja Vu all over again:

-AMD will clock the cards in the box so that it will edge the 980 Ti
-The 980 Ti will edge the Fury when both cards are overclocked.
-AMD will no doubt suffer from the misplaced "4 GB is not enough" argument

So with the increased fab costs .... and increased cooler costs (assuming all have CLC), how does AMD price the card next to the 980 Ti ? If they come in at $699, I don't see it selling well enough to change things. At $639 I see it doing well.
 
I think that's a dumb argument for simple logic jump. 1. Only 0.6% of the market is at 4k. 2. The Fury will cost too much. This is the problem with your logic. If I am spending $700 on a GPU, its going to be for a 4k monitor. Now what GPU performs best on 4k? Oh the Fury...
The high end market is the high end market.

I do agree with you on the consoles. The main reason for AMDs current GPU and CPU delays is complying with the consoles. This puts their current GPU and CPU on delay for a year or so. So we are seeing that delay now. The problem is that R&D does not happen sequentially, it happens in tandem to other R&D. So where these current generations are delayed, their next offerings are still on track.
The Excavator Cores from what we have seen thus far are pretty impressive out of the AMD camp and are the first CPUs to really be competitive in IPC for consumer CPUs. If I was a business, I would be wasting money on Intel at this point for the majority of the computers in the office when AMD right now is offering a $200 solution that competes with Intel's $400 solution. The excavator cores should also prove to be the best in gaming once more powerful versions are offered since the games coming off the XBone side will be optimized for them.
 



The high end market provides bragging rights. Not cash income. Who has the higher end automobile Porsche or GM . Porsche 2015 income target is $2.97 billion ... drop in the bucket compared to GMs $155 billion.

Your financial prediction is that AMD will recovery from it's dire financial straits because that 0.06% of the gaming population are going to all dump their old cards and buy new Fury's ?.... Let's say they all do that. How will this minute influx of income affect AMDs fortunes ?

Let's do the math ....

-Assume no one buys any Nvdia cards for 4k and AMD owns that 0.06% of the market
-Meanwhile for the other 99.94% of the market is still buying two 970s for every other card AMD makes put together.

Simply put, the math doesn't work

Consider also that from a % of sales standpoint, benchmarks standpoint and forum posts on the subject, the major sentiment among the enthusiast market feels that 4k is not quite "ready for prime time"

1. Not a single card made today can handle 4k at 60 fps across most major titles, even SLI / CF is a challenge

2. The fact that 4k remains at 0.06% and remains flat .... that is not growing.... no month to month market share increases. Interestingly enough the largest growth has been in the 1600 x 900 market (0.40%), presumably for small laptops.

3. Not a single 4k monitor made today can support 120/144 Hz, and I expect won't until Display Port 1.3 compliant products arrives.

4. Many folks feel that investing $1200 in a 4k monitor and $1,500 on video cards is unwise considering that once 120/144 Hz does arrive at 4k, they will at once making all existing 4k monitors irrelevant.

5. An investment in 4k at 120/144 Hz would warrant cards powerful enough to break 60 fps in the majority of AAA games and that ain't happening until 12-18 months from now.

4k will arrive.... but I don't see it being a significant factor on anyone's income statement for at least 12-18 months.





Well many have been flushed.... and many have left, If they can take a dedicated core and get them focused on higher margin activities, they can turn this around. Right now the lack of cash is hurting R & D .... that's why I think if they cut the price a little, they can make inroads into a wider market segment beyond 4k .... might not make as much per card but they'd sell a lot more cards.



According to Steam, majority of gamers on 1080p with a 1GB card.
Click to expand...

1920 x 1080 = 34.54%
1366 x 768 = 26.33%
1600 x 900 = 7.82%
1440 x 900 = 5.08%
1680 x 1050 = 4.95%
1360 x 768 = 3.02%
1024 x 768 = 2.30%
2560 x 1440 = 1.11%
3840 x 2160 = 0.06%
3440 x 1440 = 0.04%

 

very good points except the market share thing. when dp 1.3 arrives 4K development and adoption will increase (unless the panel makers start shenanigans again). the current options are for early adopters who are willing to pay for the hardware. in reality, current crop of cards aren't useful over single, high refresh rate 1600p displays. radeon fury should be fine for displays <4K.
 


What do you mean "except" ? .... that was exactly my point. It isn't here now and Fury will be "last gen" by the time it does.

So the postualtion being put forth is that increasing the market share from 0.06% to 0.12 % will change the math ? That means that if AMD holds that 0.12%..... that changes their prospects by 1 extra sale in 800. Meanwhile 75.2% of sales are to nVidia ... which gives nVidia 601 sales to every 199 for AMD

2880 x 1800 = 0.01%
3200 x 1800 = 0.01%
3440 x 1440 = 0.04%
3840 x 2160 = 0.06%
5760 x 1080 = 0.06%

Total = 0.12%

960 = 0.78%
970 = 3.05%
980 = 0.84 %

Total = 4.67%

All R9s = 0.99%
All R7s = 0.55%

Total = 1.54%

In short...the Fury isn't going to change AMDs fortunes if it merely sits atop the performance charts at 4k....and everything else above 1600p .... not with one in 800 gamer machines fitting in that category. If they can price it under the 980 Ti and price / maintain the rebrands under the corresponding nVidia cards .... and make money doing so.... then I can see them turning things around.
 

i don't quite understand... i was under the impression that the .06% figure was for 4K monitors (says so in your post), not amd's marketshare. and saying "amd holds that 0.12%" would mean that amd holds all of that 4k marketshare...? i am confused here. that's why i completely sidestepped that part in your post.
even if your analysis made sense (to me), i would not relate marketshare to the viability of the tech amd and nvidia are bringing. marketshare depends on a lot of other things like pricing, availability, seasonality, health of the industry, p.r., customer support and so many other things as well as the tech. i am sorry i missed your point.
 
The question at hand is "Can Fury change AMDs financial position ?" I suggested that for the card to be successful, it should be priced at $639 (cheaper than 980n Ti) and not $699 because capturing the 4k market is not enough to change AMDs market share. based upon leaked performance data, 4k is the segment that Fury will have a lead over the 980 Ti. If you are suggesting that peeps while my a more expensive card that performs less than it's competitor which requires less power and produces less heat, I wouldn't know how to address such an argument.

We have a condition that has been observed in the market. That is this....

In the past 9 months, nVidia has sold twice as many 970s all all AMD R7/R9 cards combined over the past 18 months
In the past 9 months, nvidia has sold 3 cards for every 1 card AMD has sold over the past 18 months.

The position has been put forth that if the "Fury is king of 4k", that will somehow improve AMDs financial position. My answer is that 1 extra sale out of every 800 is not going to change that position.

-The 4k market is small and a review of month to month market penetration day shows it is NOT growing. If every person who owns or buys a 4k monitor in the next year was to buy Fury, that tiny market segment will not imapct AMDs bottom line.

-If the Fury is priced higher than the 980 Ti, what exactly do you think will reverse this 3 to 1 sales trend ?

- The price issue has been addressed; it was the entire basis of my post. If the Fury costs more than the 980 Ti but can not beat it at resolutions below 1600p, it uses more power, it creates more heat, it requires larger PSUs, what of those factors exactly are you suggesting will swap people to pay more for less with the aforementioned disadvantages ?

-As for availability, how exactly are you suggesting availability will change this 3:1 dynamic ? Do you know something we don't ? Is nVidia closing factories ? What exactly under the heading of availability (other than adequate yields on the Fury) can impact this 3:1 dynamic.

-How exactly does seasonality impact this 3:1 dynamic ? Let's look at month to month trends...

970 Market Share Growth in last 3 months: +0.27% / +0.33% / +0.24%
All R9s Market Share Growth in last 3 months: +0.07% / +0.00% / +0.02%

How do you see 'seasonality" affecting those trends ?

- How do you see the "health of the industry" affecting nVidia versus AMD with regard to this 3;1 dynamic ?

-What exactly is it that you know about each player's PR plans in the next 6 months that would change this 3:1 dynamic ?

-What exactly is it that you know about each player's customer support plans in the next 6 months that would change this 3:1 dynamic ?

So again:


We have no reason to suspect than there are any issues of availability on Nvidia's side, they been selling at that 3:1 rate (and growing) for 9 months w/o any shortages..... there's no logic to suggest seasonality and health of the industry affect one side any more than the other or any implication that any changes are forthcoming from either side in PR and customer support.

Simply put:

1. Contrary to what had been suggested, capturing 100% of the 4k (> 1600p) market will have a minuscule impact on anyone's bottom line simply because the market is too small. There are no signs that the 4k market is growing based upon observed month to month market share. It accounts only for 1 in 800 purchases.

2. For the Fury to have an impact below that level, it must:

a) As I suggested, have a lower price point than the comparable competition from nVidia.... I don't see a price point at $699 or above as being viable for 1440p and below, unless

b) The Fury has a substantial performance increase over the the comparably priced competition from nVidia .... leaked test results are not supporting that at this point in time.


Now what has not been discussed as yet is the potential impact of the AMD rebrands.... at $370, the 390 for example seems well positioned to compete with the 970. If AMD can be price, performance and power competitive here, we can see some signs of hope assuming nVidia doesn't want to engage in a price war.
 
What is all this about AMD not putting any money into R&D compared to Nvidia?

We're talking $1.1 Billion (AMD) vs $1.4 Billion (Nvidia), according to each company's latest annual statement (i.s. Form 10-K). That's a gap for sure, but I think people are overstating the size of that gap here. A smaller company putting more than $1B/yr into R&D is far from neglecting research.

AMD 10-K (see table on p. 48) - https://www.sec.gov/Archives/edgar/data/2488/000119312515054362/d871455d10k.htm

Nvidia 10-K (see table on p. 47) - http://files.shareholder.com/downloads/AMDA-1XAJD4/291203888x0xS1045810-15-36/1045810/filing.pdf
 

I think the growing availability of UHD displays below $500 will start changing that next year with some help from cheaper 14/16nm GPUs to drive them.

Of course, by then, AMD's HBM advantage will be gone since both AMD and Nvidia will be using HMB2.
 
Status
Not open for further replies.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom