[Deustche Bank Conference] AMD's New x86 Core is Zen, WIll Launch WIth K12

[MiddleOfTheRoad]

By the way, to answer your point of "why would they do that" -- it's easy. Intel doesn't sell game console chips, so comparing total X86 share is a bit of a misleading metric. It's like saying that Intel has lower share of the ARM server market than AMD does (which is a true statement, but a meaningless one).

Again, not true. Intel did sell game console chips, the original Xbox was powered by an Intel 733 Mhz Processor (based on a Celeron).

Intel marketshare had those Xbox sales in their x86 numbers as well -- you can't have it both ways, dude.

AMD has stated that the goal is to have over 50% of revenue from outside PC sales -- which is a smart strategy. They are not there yet, but it is a very wise move to not heavily depend on sales in a declining industry (which was their old business model of the 1990's - 2000's). I think the PC Market will continue to shrink as better smart TV's, game consoles, tablets and smartphones replace a portion of traditional desktop usage (web surfing/casual games). Those game consoles wins keeps AMD in the mix for these replacement devices.

 

[DrMrLordX]

FWIW AMD has been stretching itself even thinner in the last three years. They have introduced a new proprietary gaming API, they are trying to improve their professional drivers, they are introducing a new architecture based on a different instruction set that they will have to support with specific tools... All those products should demand more resources, not less, and less is the prevalent trend regarding AMD engineering team. Since Rory Read already stated that further "optimizations" should be coming in the next 3-4 quarters, I'd be really skeptical of AMD's ability to deliver all the software promises they are making.

While true, I think the greater concern is that AMD has a killer hardware product selling on stores right now that languishes in near-perpetual underutilization. That product is Kaveri.

Take a look at your average benchmark suite used to test CPUs in 2014 (example: any of the recent Haswell-E or Devil's Canyon reviews, or even the 8370E reviews) and ask yourself: how many of these benchmarks could benefit from HSA? I would think that any benchmark that is fp-heavy and scales well regardless of how many cores you throw at it would be a prime candidate for HSA.

So uh, where's the beef?

 

[Phynaz]

People said the exact same thing during the launch of the Athlon 64 and Athlon X2... That it's over for Intel. it wasn't true then and it's not true now.

AMD still beats the heck out of Intel if the user is running the integrated graphics -- Intel's best stuff i5 / i7, still can't compete with AMD's $150 APU's in graphics performance.

AMD's marketshare has always been around 10 - 20% of the market and it will probably always been that way. Which makes a lot of sense, since Intel is a company that is 5 times larger.

Intel has a vested interested in helping AMD stay in the market -- otherwise, Anti-Monopoly laws would force a breakup of Intel. All you're saying is just empty banter.

You completely missed the point.

And Intel helping AMD? Lol, no. Intel has has about million concerns more important than AMD.

Anti-monopoly laws...You should go read them.

 

[frozentundra123456]

While true, I think the greater concern is that AMD has a killer hardware product selling on stores right now that languishes in near-perpetual underutilization. That product is Kaveri.

Take a look at your average benchmark suite used to test CPUs in 2014 (example: any of the recent Haswell-E or Devil's Canyon reviews, or even the 8370E reviews) and ask yourself: how many of these benchmarks could benefit from HSA? I would think that any benchmark that is fp-heavy and scales well regardless of how many cores you throw at it would be a prime candidate for HSA.

So uh, where's the beef?

No, they do not have a killer hardware product on the shelves. They *might* have if everything used HSA. But the point is, very few apps use HSA. In the current software market they have an overpriced, mediocre performing CPU with more gpu than the average person needs, but not enough to beat even a low/midrange discrete card if you really need graphics performance. You can always say "what if" but unfortunately reality is what really counts.

 

[krumme]

This is definitely a concern, and more so for the shareholder than the consumer/customer.

But this is precisely what "fading to irrelevance" looks like. We've already witnessed it so many times in x86, most recently with Via (which is irrelevant in all but name only these days).

Sure they are holding on, but the product pipeline is resource starved for lack of corporate revenues which is a vicious self-perpetuating cycle. Once you fall off that wagon, rarely do you succeed in getting back on.

What do we actually know about ressource starving? Nada. The devil is in deep complexity. And i dont think amd with cowboy RR is trying to reinvent the bd wheel. Its pros give them the slack the financial results deserve.

At least the fading to irrelevance is more nonsense than ever for amd. Except a few years by a pure chance of outside factors like buying half baked k7, p4 and soi 90/60nm they have lost money all the way. They are more important in the market than ever as the small company they always were, but they are a long way from the Intel copies of the 90ies. Remember history. K7-K8 was a short period.

 

[mrmt]

Look at AMD's financials -- they still very much rely on the PC market, even as it shrinks as a % of revenue because the business continues to contract.

It's funny to see a CEO stating a 50% revenue share goal when that very goal depends on shrinking revenues from the rest of the company.

 

[mrmt]

What do we actually know about ressource starving? Nada.

You just can't cut your R&D budget by 25% and expect to develop projects of the same scope as before, and that's exactly what AMD did in the past three years.

 

[Arachnotronic]

It's funny to see a CEO stating a 50% revenue share goal when that very goal depends on shrinking revenues from the rest of the company.

Well, I think the intention is to grow everything else to get there, but what's happening is that PC share is shrinking, other stuff is growing. This allows AMD to get closer to its 50/50 ahead of schedule

 

[pw257008]

Anti-monopoly laws...You should go read them.

Or just understand that anti-monopoly cases are often political in nature and dependent on the political leanings of the judges involved (or if you prefer, on the way these judges interpret the laws, which is by its nature often related to said judges' political leanings). In other words, Intel could be vulnerable to anti-monopoly suits in the right circumstances.

 

[mrmt]

Well, I think the intention is to grow everything else to get there, but what's happening is that PC share is shrinking, other stuff is growing. This allows AMD to get closer to its 50/50 ahead of schedule

I don't think they were naive enough for this. They axed their server/FX line up, they castrated their APU line up with DDR3 only solutions, they didn't try to break into the mobile market... I don't think they had any illusions about sustaining both their market and revenue share in the x86 market. They had to know well beforehand that their core PC business would erode in the process of shifting resources to the new frontiers. It has to be a conscious decision.

And while the shrink surely adds to the speed of which the 50/50 revenue goal will be reached, it surely messes with the other financial parameters they need to sustain, no wonder Rory Read is talking about further OPEX cuts for the next quarters.

 

[mrmt]

While true, I think the greater concern is that AMD has a killer hardware product selling on stores right now that languishes in near-perpetual underutilization. That product is Kaveri.

Take a look at your average benchmark suite used to test CPUs in 2014 (example: any of the recent Haswell-E or Devil's Canyon reviews, or even the 8370E reviews) and ask yourself: how many of these benchmarks could benefit from HSA? I would think that any benchmark that is fp-heavy and scales well regardless of how many cores you throw at it would be a prime candidate for HSA.

Killer hardware without killer hardware support is meaningless, and this is exactly HSA problem here. None of the big software companies are backing AMD. It's the complete opposite situation of the x64 situation, when Microsoft backed AMD solution.

 

[TurtleCrusher]

The anti-AMD circle jerk is real; usually with the same participants every time. When you guys stop drinking the brand loyalty Kool-Aid you'll see that AMD isn't screwed and their products keep relatively stable prices throughout the market for us consumers.

AMD's 8 core FX line hits the $100 price point fairly often. Can you say the same about Intel's i5 lineup? When you're able to buy an 8 core FX CPU, ATX motherboard and 8GB of DDR3 for $225-250, something is seriously wrong with you if you can't see the competitiveness that AMD brings to the market. Every time someone even suggests an AMD product, it's the same few people coming into these threads constantly move goalposts back and forth to suit their agenda.

The funny thing is that unless you're a shill or a stockholder, you have zero financial interest in either companies, yet you all bicker and argue regurgitated talking points you learned from the next fanboy. I'm a union electrician, and Intel is the largest employer of union electricians in Oregon. My livelihood depends on their success and you won't see me shilling or pulling one way or the other. I'll still buy the best value, regardless of brand. Many of you are looking like basement dwelling children with your "always negative" stance towards anything that can change the status quo in the CPU market.

 

[jpiniero]

Because most of the people here only really use their computer's processing power for games. Otherwise they would just be fine with a tablet.... and Bulldozer does terribly in games, especially ones that aren't great threaded.

 

[mrmt]

Because most of the people here only really use their computer's processing power for games. Otherwise they would just be fine with a tablet.... and Bulldozer does terribly in games, especially ones that aren't great threaded.

Don't forget that some people here need performance/watt, and AMD offers are simply unacceptable on this department.

 

[mrmt]

AMD's 8 core FX line hits the $100 price point fairly often. Can you say the same about Intel's i5 lineup? When you're able to buy an 8 core FX CPU, ATX motherboard and 8GB of DDR3 for $225-250, something is seriously wrong with you if you can't see the competitiveness that AMD brings to the market. Every time someone even suggests an AMD product, it's the same few people coming into these threads constantly move goalposts back and forth to suit their agenda.

And what does this has to do with AMD *capabilities* to stay on the market, which are the crux of the discussion on this topic?

 

[piesquared]

The anti-AMD circle jerk is real; usually with the same participants every time. When you guys stop drinking the brand loyalty Kool-Aid you'll see that AMD isn't screwed and their products keep relatively stable prices throughout the market for us consumers.

AMD's 8 core FX line hits the $100 price point fairly often. Can you say the same about Intel's i5 lineup? When you're able to buy an 8 core FX CPU, ATX motherboard and 8GB of DDR3 for $225-250, something is seriously wrong with you if you can't see the competitiveness that AMD brings to the market. Every time someone even suggests an AMD product, it's the same few people coming into these threads constantly move goalposts back and forth to suit their agenda.

The funny thing is that unless you're a shill or a stockholder, you have zero financial interest in either companies, yet you all bicker and argue regurgitated talking points you learned from the next fanboy. I'm a union electrician, and Intel is the largest employer of union electricians in Oregon. My livelihood depends on their success and you won't see me shilling or pulling one way or the other. I'll still buy the best value, regardless of brand. Many of you are looking like basement dwelling children with your "always negative" stance towards anything that can change the status quo in the CPU market.

The bolded part is the key, I've been saying it for years. Enthusiasts/ hobbyists have been displaced with more... financially motivated posters. They don't want a successful AMD, they want to profit from their failure. Short interest in AMD is enormous, and those 'people' want AMD dead. There of course are others, but those people are the most vocal in voicing negative sentiment in any way possible. There's no doubt.

 

[TurtleCrusher]

Because most of the people here only really use their computer's processing power for games. Otherwise they would just be fine with a tablet.... and Bulldozer does terribly in games, especially ones that aren't great threaded.

Terribly in games? PS4/XB1 games are finally trickling in to the PC side and it's shaking things up. Just download the demo to Fifa 15 and tell me the FX series can't play next generation games. 60fps solid the entire time, save for cut-scenes that are automatically locked to 30FPS. Unless you're running a top-end crossfire or SLI PC, the FX line is not going to hold you back one bit.

Starcraft 2, a game that notoriously run one 1 and a half threads at best, performance is much higher than it used to be with patches. I wouldn't play SC2 with my Q8200 @2.8Ghz because it would bog down in heavy multiplayer games. The 8350? Hardly breaks a sweat. Battlefield 3/4 multiplayer? Nearly unplayable in 64 man matches. FX-8350? Feels like single player. NBA 2K14 would drop down from 80fps to 50fps during intense parts of the game on the Q8200, yet the FX stays at 120 and every once in a while drops down to 110.

The point is, not a single game I've played with it has poor performance, save for GTA4 and their useless porting ability. Future games make the FX line competitive even on the e-peen level. And the price? $100-125 vs $200? An actual enthusiast and not a fanboy shill is going to put that extra towards an even better video card or an SSD.

 

[positivedoppler]

The bolded part is the key, I've been saying it for years. Enthusiasts/ hobbyists have been displaced with more... financially motivated posters. They don't want a successful AMD, they want to profit from their failure. Short interest in AMD is enormous, and those 'people' want AMD dead. There of course are others, but those people are the most vocal in voicing negative sentiment in any way possible. There's no doubt.

It benefits consumers to have AMD and everyone else do well and Intel take a back seat. Worst case nightmare scenario for all would be Intel gains an unsurmountable lead in it's fab and then all chips from cell phones, to computers, and microwaves are made by Intel, free to charge whatever it wants.

Look what happens to an ecosystem free of a Intel Monopoly. Look at tablets. Qualcommn, Apple, Mediatek, and countless others making cheap arm tablets as low as $69. This would never happen if Intel took out everyone and is the only guy in town.

 

[TurtleCrusher]

It benefits consumers to have AMD and everyone else do well and Intel take a back seat. Worst case nightmare scenario for all would be Intel gains an unsurmountable lead in it's fab and then all chips from cell phones, to computers, and microwaves are made by Intel, free to charge whatever it wants.

Look what happens to an ecosystem free of a Intel Monopoly. Look at tablets. Qualcommn, Apple, Mediatek, and countless others making cheap arm tablets as low as $69. This would never happen if Intel took out everyone and is the only guy in town.

This is absolutely spot on. I have a vested interest for the success of Intel due to my profession and location, but even I know that if they were the only dominant player (and they're not) that prices for middle class consumers would be almost completely out of reach.

It'd be almost kind of funny though to see all these people that cheered for every win they made by moving goalposts for Intel, to suddenly be shut out with the ridiculously high prices that would result.

 

[Enigmoid]

The anti-AMD circle jerk is real; usually with the same participants every time. When you guys stop drinking the brand loyalty Kool-Aid you'll see that AMD isn't screwed and their products keep relatively stable prices throughout the market for us consumers.

AMD's 8 core FX line hits the $100 price point fairly often. Can you say the same about Intel's i5 lineup? When you're able to buy an 8 core FX CPU, ATX motherboard and 8GB of DDR3 for $225-250, something is seriously wrong with you if you can't see the competitiveness that AMD brings to the market. Every time someone even suggests an AMD product, it's the same few people coming into these threads constantly move goalposts back and forth to suit their agenda.

The funny thing is that unless you're a shill or a stockholder, you have zero financial interest in either companies, yet you all bicker and argue regurgitated talking points you learned from the next fanboy. I'm a union electrician, and Intel is the largest employer of union electricians in Oregon. My livelihood depends on their success and you won't see me shilling or pulling one way or the other. I'll still buy the best value, regardless of brand. Many of you are looking like basement dwelling children with your "always negative" stance towards anything that can change the status quo in the CPU market.

I think you have realism confused with anti-AMD sentiment.

We have been hearing for years (since BD) how BD was going to blow everyone away, how HSA was going to take off, how the APU would reign supreme, how Piledriver and Steamroller will fix all of Bulldozer's flaws, how kaveri would bring 20% IPC increase and demolish everything based on LibreOffice HSA tests, how mantle would blow DX away, how the cat cores were going into tablets, how hUMA and HSA would usher in a new era of computing, how freesync would be awesome, and how OpenCL and compute was/is going to displace CUDA and/or become important to the average consumer.

Most of that didn't pan out (BD, PD, SR, HSA/hUMA, Kaveri 20% IPC, jaguar tablets, freesync - still MIA and unknown variable refersh, Lano-> Trinity-> Richland -> Kaveri the cost went up for top model, APUs still don't get more perf/$ than a pentium + 260X combo, etc). Some did (mantle seems to have merit despite caveats and OpenCL is gaining dominance) but much didn't. Some is still to be determined.

Much was also said about how intel's graphics were going to surpass AMD's, how Haswell was going to be a major jump, how intel would stay on their tick-tock schedule, how intel would be dominating in tablets and phones by now, the pumped up projections of Sofia, Mooresfield, Merrifield, how CT would massively increase graphics over BT due to having 4x EUs, etc.

Some of that didn't occur. Sofia, Moorsefield, Merrifield are massively behind schedule if they are going to be released. 14nm was delayed and tick-tock looks to be permanently off schedule. CT looks only to have 2x GPU gain over BT (measured using Egypt HD). Though Iris Pro is good, the regular GT2 graphics are still lagging behind Kaveri.

However, intel has also delivered a lot as well. Massive efficiency gains in servers. Massive increase in performance and/or battery life for mobile devices. Greatly improved graphics drivers and hardware (SB couldn't even do texture filtering properly, Haswell can't do AA efficiently but is much improved). At least positive progression in performance (PII x4 980 does better in applications vs the 7850k).

http://www.hardware.fr/articles/913-7/cpu-performances-applicatives.html

Now broadwell is being showcased and being hyped (IMO quite a bit).

My point is that AMD hasn't really delivered much on the CPU front in the last few years. The GPU division is solid but the company is floundering and barely making a profit.

Personally I have never seen the 8xxx series CPUs near $100. Your way of thinking is admirable but a company selling a product at such low margins is a company that is not doing well. You could simply say that the product is so bad they have to sell it so cheap because if they sold it at the price they wanted (8150 for $245) nobody would buy it.

I like AMD and I wish them well but nothing about that CPU division screams that they are going to rise like a phoenix out of the ashes (certainly possible for them to pull a conroe but unlikely given the small R&D budget). I certainly have interest in the financial health of AMD, intel, Nvidia, Qualcomm, Samsung, etc. If they go broke, they don't produce any more products for me to buy.

Its very fine to always go for value but don't be penny wise pound foolish. Its fine to believe in value but value is meaningless without some base of quality, for the same reason why I never buy the really cheap toilet paper.

 

[mrmt]

It benefits consumers to have AMD and everyone else do well and Intel take a back seat. Worst case nightmare scenario for all would be Intel gains an unsurmountable lead in it's fab and then all chips from cell phones, to computers, and microwaves are made by Intel, free to charge whatever it wants.

The funny thing is that it is that despite Intel bleeding AMD dry in a price war, you and others think AMD is responsible for the low CPU prices. If CPU prices are that low today, that's because of Intel and Intel alone, not AMD. AMD would rise its prices if it could, but since the line up is obsolete/weak, they simply cannot charge too much money.

 

[positivedoppler]

The funny thing is that it is that despite Intel bleeding AMD dry in a price war, you and others think AMD is responsible for the low CPU prices. If CPU prices are that low today, that's because of Intel and Intel alone, not AMD. AMD would rise its prices if it could, but since the line up is obsolete/weak, they simply cannot charge too much money.

To easily counter your point. AMD is not competitive, and that is the whole problem right now and why the desktop and laptop cpu market is very unhealthy. Tablet ecosystem has been so healthy without Intel, Intel needed to practically give away BT to buy in.

How happy are people with Devil's Canyon? How long is Intel going to wait to bring 14nm to desktop? The answer is they can wait and continue milking Haswell at a high margin because AMD has no answer. This clearly was not the case in the days of the original Athlon forced the P4 into a corner.

You somehow seem to be under the impression that the cpu world is a better place with only Intel. I don't know what path of reasoning lead you down that train of thought but it defies everyone's understanding that competition is good.

 

[mrmt]

You somehow seem to be under the impression that the cpu world is a better place with only Intel. I don't know what path of reasoning lead you down that train of thought but it defies everyone's understanding that competition is good.

I'm not under the impression that the CPU world is better with Intel competing alone, that ship has sailed already and is a fact we must get used to it. Intel *is* already competing alone in a lot of market brackets, and is well on its way to get all the x86 market for itself. Intel is able to dictate the prices on most segments of the x86 market, and they are able to get record margins/profits while doing so. AMD, on the other hand, must answer every new launch from Intel with cuts in their ASPs, and are being pushed lower and lower in the value chain, accruing losses on almost each quarter. In reality, it is AMD that desperately needs higher ASPs on the market, while Intel is pushing down the price of their CPUs in preparation to compete against the ARM ecosystem.

AMD is no competition for Intel, whatever competition we are going to see will come from the ARM chip makers like Apple, Qualcomm, Samsung and others.

 

[Enigmoid]

It benefits consumers to have AMD and everyone else do well and Intel take a back seat. Worst case nightmare scenario for all would be Intel gains an unsurmountable lead in it's fab and then all chips from cell phones, to computers, and microwaves are made by Intel, free to charge whatever it wants.

Look what happens to an ecosystem free of a Intel Monopoly. Look at tablets. Qualcommn, Apple, Mediatek, and countless others making cheap arm tablets as low as $69. This would never happen if Intel took out everyone and is the only guy in town.

I know what you are saying but this argument doesn't make sense. If a microwave chip costs $.50 what kind of magic sauce is going to make it worth $15 to put it in a microwave. Consumers aren't idiots. If the $15 chip doesn't add enough value nobody will buy it regardless of marketing and appliance makers won't use it. You could have a quantum CPU that has more power than all the computers on earth but if your microwave cost more and didn't do more than I felt it was worth, I wouldn't buy the microwave.

Just playing devil's advocate here but bear with me on this thought experiment. Take two systems A and B. A is a monopoly and B is not. Companies B1, B2, B3, and B4 sell similar products at $2.50, $3.25, $2.25, and $2.70 respectively. Company A (monopoly) sells a similar product to the B companies for $1.50. Irrespective of the fact that system A is a monopoly A has a better price. Monopolies are not necessarily bad, its the fact that companies, when guaranteed of profits reduce expenditures to maximize those profits.

It benefits nobody for intel or any company to take a backseat.


I can't take you seriously when you say Apple is making cheap arm tablets let alone when $69 is in the same sentence. Those tablets are crap that don't work well as soon as you open the box; race to the bottom. In a way apple illustrates the opposite of what you are saying. Only so many people are willing to pay so much for a phone/tablet. Apple's A9 could be as powerful as the 5960X in a 1.5W envelope but if they priced it at $49,999 few would buy it.

@Scholzpdx : At this point in time, intel is competing primarily with itself. Its not going to raise prices because people will stick to legacy. In the history of computing devices, CPUs have never been cheaper.

 

[gdansk]

I honestly can't imagine what they'd do to design a good chip. It can't be as efficient as Intel due to the node difference. It probably won't be wider since that needs a lot of transistors and AMD is constantly cash-starved. What on earth do they plan to do? The only interesting thing would be a core with 6-wide ALU and SMT4. Might be able to fit 8 on a 100W chip with 20nm. 32 x64 threads in a single chip would at least be interesting. Still doubt Zen will be competitive in performance, it'd just cost too much to develop. Competitive power consumption is probably impossible.

I'm more excited for their ARM core, which has some possibility of being faster than some current ARM chips.