AMD CPU speculation... and expert conjecture

[noob2222]

Spoiler

I don't believe someone on a forum postulating that ARM will rule the world when that's exactly what everyone else is saying will not happen...(behind closed doors or not...).

EDIT: From a diplomatic standpoint, with many ARM partners on the HSA group as founders with AMD, they would need to show a sort of support to those partners who have helped them to write the HSA standards. Now, this is exactly what this looks like..."See, ARM can work in servers!! *mumbling under breath* how long do we have to do this Rory?? I need to get back to work on steamroller..."

LOL. First you claimed people was misinterpreting roadmaps, then changed to my friend said me Seattle is only an "experiment" for testing market. Now it has changed to kind of "diplomatic" release for satisfying HSA ARM members.

I'll backup what 8350 said. I was talking to a couple AMD engineers last month and when the topic of the new ARM chips came up they rolled their eyes. A lot of that initiative is just to satisfy investors so they can add ARM to their checklist of available options. It's a sizable company with many groups competing for their piece of the pie and wanting to keep their jobs.

People give Rory Reed a lot of credit for "turning things around" but what's keeping AMD alive now is a deal made well before he came on board, and that's the x86 APUs.

Funny, now we have a fourth explanation: they are released for satisfying investors.

I don't know what engineers you asked (x86?, GPU?). But I know that AMD head for servers, Feldman, says a completely different history:

http://www.eweek.com/servers/amds-feldman-arm-will-quickly-gain-market-share-in-servers/

I think we all need to set down with you and explain to you what marketing is i know it will be disappointing but it needs to be done, so far we had a couple people including myself explain why(based on employers from Amd that are not on a article or news site) Arm is more of a another product vs the main product i have no information whatsoever on steamroller FX CPU's but i do know Amd cares MUCH MUCH more about their x86 design vs arm which MANY MANY others make.

I think all of you need to set down first to decide which of the four explanations you are giving is the correct: First the roadmap was misinterpreted, then the explanation changed to the product is an "experiment", then changed again to the product is a "diplomatic" release for satisfying HSA members, and finally changed to released for satisfying investors.

.

ya, because the word introduce actually means replace. If AMD migrates away from the only product line that is distinct, they will lose business. period.

ARM will gain market share, there is no way it can't since there aren't any ARM servers yet. from 0% to 1% is a gain in market share. Between the big (for the sake of arguing just say 5) ARM manufacturers, they will each get 0.2% of that 1% market share if they all split sales evenly. Now say samsung spends a small fraction of their company to advertise. now they double their sales while the others get a smaller chunk.

AMD can't out-advertise multi-billion dollar companies. Thats why they are currently in the position they are in with Intel. Instead of competing with 1 other company, it will be 5 or more. (5 others already hold arm a-57 lisences)

Because they count it as APU. CPU alone the improvement in efficiency is of only 10-15%, which is not enough to compete with Intel.

so now SR cores are only 10-15%? I thought a 4-core Kaveri would compete with the I5 on the cpu level and trash it on the APU side ... but a 16 core SR opteron can't compete with a xeon?

Logic fails this one again.

While (Juanrga = wrong) {printf("JUANRGA IS NEVER WRONG")};

I know the four explanations are plain wrong. Up to now my sources have shown that are very reliable and others (including friends working at) are not so.

this source?

 

[anxiousinfusion]

looks like i am the only one without a friend working for amd.

TT_TT

This is starting to sound a lot like grade school "My dad works for X company."

 

[The Q6660 Inside]

Holy world records! http://www.overclock.net/t/1436561/various-amd-radeon-r9-290x-reviews-thread/370#post_21046990 and http://www.overclock.net/t/1436561/various-amd-radeon-r9-290x-reviews-thread/380#post_21047021
AMD and Intel working together in the same build, seems legit 😛...
DAT CROSSFIRE BRIDGE!

 

[blackkstar]

Spoiler

Regarding AMD APUs keeping them afloat. I think the past decisions of AMD are double sided. Yes APUs are pulling AMD rather well right now, but the same people who made that great move also made the move to make Bulldozer a low IPC, high frequency part that didn't clock well and wasn't much better than Phenom 2.

Rory gets the credit for turning the APU business into something where AMD creates semi-custom solutions for companies. PS4 and XBone are huge profit makers for AMD right now and it's because AMD was able to provide a custom solution for Sony and MS that they are making that money.

I highly doubt AMD would have been as lucrative to both had AMD only been able to offer stock APUs. MS and Sony wanted their own custom parts tacked onto the parts and they got them.

Which is why I think solid x86 core lineup (low power chips and high performance chips) is integral to AMD's future. They are building a company where Cray can come up and say they want a custom APU with Jaguar + massive GCN cores and it can happen. They can order massive SR CPU with dGPU HSA.

There are a lot of options for AMD on the table with the new strategy and it seems like it would be in AMD's best interests to keep as many of those going as possible. Specially given AMD's history of relying only on CPU, releasing a dud, and then having profits spiral into a death swirl as a company with a tenth of the competitions R&D budget tries to out-smart the competition.

With the new strategy, if AMD comes up with a new product and it sucks, they can just focus on something else. Low and behold, there's rumors of two versions of SR floating around and AMD is only focusing on APUs. I'm suggesting that SR wasn't competitive enough with Intel so it got canned, and instead of this turning into another generation of AMD only offering budget CPUs and earning profits around that, they can go mess around with APUs and low power and make money while the big CPU division gets their stuff together.

It is a really good strategy and it makes sense, it reminds me of that old saying, "don't put your eggs in one basket." From the sound of it, AMD had a basket that didn't work (SR FX chips). Normally, AMD would be doomed, but now they can go to their APU basket and be just fine. Doesn't mean the CPU basket is dead, dying, never coming back, etc. The other baskets can be used to levitate the basket that is not healthy.

I agree with most of this.

Of course SR is not competitive enough with Intel. That is why there is no SR Warsaw to compete against Xeons. If AMD planned SR Warsaw, and then derived SR FX CPU from them (as some here believed), then AMD would be back in the red numbers and probably out of business in one or two years.

SR is only competitive when HSA is enabled and used in APUs. That is why AMD is replacing 8-core Opterons by 4-core Berlin APUs and then deriving Kaveri APU from Berlin APU. Why do you believe that AMD quotes Berlin 8x efficiency over previous Opterons? Because they count it as APU. CPU alone the improvement in efficiency is of only 10-15%, which is not enough to compete with Intel.

In fact I agree with other analysts on that AMD comeback to HPC would be with APUs. Berlin or some product derived from it could be a good choice for Cray and other HPC:

http://techreport.com/news/24973/berlin-warsaw-are-the-future-of-amd-x86-server-lineup

http://www.theinquirer.net/inquirer/news/2275443/amd-reveals-details-of-berlin-and-seattle-chips

The problem here is that AMD is again too late to the party (thanks to the myopia of the former head). A Berlin APU could be fantastic years ago. But in late 2014 it will be competing against Intel new products such as the improved Xeon Phi in socket format (this eliminates PCIe bottleneck) and against Nvidia and its new ARM-CUDA products.

juanrga not to let you down but Arm is not something Amd wants to do all in its ONE of their future products and this was confirmed from my friend at Amd in fact he warns me Amd is not the only person making Arm processors and they don't perceive a big profit margin in doing it, it would be like competing in the cell phone market when so many already do.

Of course AMD will be competing against other players. Their only possibility is the combination of their expertise in servers/HPC with a custom core + GCN. Seattle will be released with standard cores, because AMD had no enough time to develop custom cores. They are doing it.

ARM is their only choice. There is no possibility that AMD gain significant market share using x86. In the first place AMD has no competitive x86 CPU product. In the second place, the market is migrating away from x86 monopoly. Feldman is right in that ARM is the future.

Ah, but you forget, AMD is also competitive in dCPU gaming PCs when it comes to modern games. You also forget AMD has basically gotten every game developer who is making a multi-platform to optimize their games for an AMD architecture.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/7

Look at a semi-modern engine like Frostbite 2. AMD FX 8350 offering best frametime experience (smoothest game experience) and some of the best frame rates.

This is why I am putting forth the idea that AMD is done with the server -> HEDT transition and instead might want to just flat out go straight dCPU for high end gaming platform with a high end Radeon card. We might from now on see AMD go APU/dCPU -> server.

Oh look, that's already happened with AMD turning their desktop APUs into server APUs.

AMD is working backwards now from how they have been. People are still anticipating SR Opteron to come first and it's not going to happen. It makes far more sense for AMD to go SR or EX gaming CPU first and then trickle over to server.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/8

Now look at this. AMD APU in different performance class than AMD CPU. FX 8350 fitting in just fine with 2600k and IB CPUs.

Top end APU fitting in below Core i3.

Do you see the point I am trying to make?

Gaming is important to AMD. APU puts AMD's CPU offerings below Intel's Core i3 in most situations. FX 8350 is in Intel $300+ territory most of the time.

AMD going APU only is basically throwing that all away. They are not going to do that.

 

[The Q6660 Inside]

Spoiler

Regarding AMD APUs keeping them afloat. I think the past decisions of AMD are double sided. Yes APUs are pulling AMD rather well right now, but the same people who made that great move also made the move to make Bulldozer a low IPC, high frequency part that didn't clock well and wasn't much better than Phenom 2.

Rory gets the credit for turning the APU business into something where AMD creates semi-custom solutions for companies. PS4 and XBone are huge profit makers for AMD right now and it's because AMD was able to provide a custom solution for Sony and MS that they are making that money.

I highly doubt AMD would have been as lucrative to both had AMD only been able to offer stock APUs. MS and Sony wanted their own custom parts tacked onto the parts and they got them.

Which is why I think solid x86 core lineup (low power chips and high performance chips) is integral to AMD's future. They are building a company where Cray can come up and say they want a custom APU with Jaguar + massive GCN cores and it can happen. They can order massive SR CPU with dGPU HSA.

There are a lot of options for AMD on the table with the new strategy and it seems like it would be in AMD's best interests to keep as many of those going as possible. Specially given AMD's history of relying only on CPU, releasing a dud, and then having profits spiral into a death swirl as a company with a tenth of the competitions R&D budget tries to out-smart the competition.

With the new strategy, if AMD comes up with a new product and it sucks, they can just focus on something else. Low and behold, there's rumors of two versions of SR floating around and AMD is only focusing on APUs. I'm suggesting that SR wasn't competitive enough with Intel so it got canned, and instead of this turning into another generation of AMD only offering budget CPUs and earning profits around that, they can go mess around with APUs and low power and make money while the big CPU division gets their stuff together.

It is a really good strategy and it makes sense, it reminds me of that old saying, "don't put your eggs in one basket." From the sound of it, AMD had a basket that didn't work (SR FX chips). Normally, AMD would be doomed, but now they can go to their APU basket and be just fine. Doesn't mean the CPU basket is dead, dying, never coming back, etc. The other baskets can be used to levitate the basket that is not healthy.

I agree with most of this.

Of course SR is not competitive enough with Intel. That is why there is no SR Warsaw to compete against Xeons. If AMD planned SR Warsaw, and then derived SR FX CPU from them (as some here believed), then AMD would be back in the red numbers and probably out of business in one or two years.

SR is only competitive when HSA is enabled and used in APUs. That is why AMD is replacing 8-core Opterons by 4-core Berlin APUs and then deriving Kaveri APU from Berlin APU. Why do you believe that AMD quotes Berlin 8x efficiency over previous Opterons? Because they count it as APU. CPU alone the improvement in efficiency is of only 10-15%, which is not enough to compete with Intel.

In fact I agree with other analysts on that AMD comeback to HPC would be with APUs. Berlin or some product derived from it could be a good choice for Cray and other HPC:

http://techreport.com/news/24973/berlin-warsaw-are-the-future-of-amd-x86-server-lineup

http://www.theinquirer.net/inquirer/news/2275443/amd-reveals-details-of-berlin-and-seattle-chips

The problem here is that AMD is again too late to the party (thanks to the myopia of the former head). A Berlin APU could be fantastic years ago. But in late 2014 it will be competing against Intel new products such as the improved Xeon Phi in socket format (this eliminates PCIe bottleneck) and against Nvidia and its new ARM-CUDA products.

juanrga not to let you down but Arm is not something Amd wants to do all in its ONE of their future products and this was confirmed from my friend at Amd in fact he warns me Amd is not the only person making Arm processors and they don't perceive a big profit margin in doing it, it would be like competing in the cell phone market when so many already do.

Of course AMD will be competing against other players. Their only possibility is the combination of their expertise in servers/HPC with a custom core + GCN. Seattle will be released with standard cores, because AMD had no enough time to develop custom cores. They are doing it.

ARM is their only choice. There is no possibility that AMD gain significant market share using x86. In the first place AMD has no competitive x86 CPU product. In the second place, the market is migrating away from x86 monopoly. Feldman is right in that ARM is the future.

Ah, but you forget, AMD is also competitive in dCPU gaming PCs when it comes to modern games. You also forget AMD has basically gotten every game developer who is making a multi-platform to optimize their games for an AMD architecture.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/7

Look at a semi-modern engine like Frostbite 2. AMD FX 8350 offering best frametime experience (smoothest game experience) and some of the best frame rates.

This is why I am putting forth the idea that AMD is done with the server -> HEDT transition and instead might want to just flat out go straight dCPU for high end gaming platform with a high end Radeon card. We might from now on see AMD go APU/dCPU -> server.

Oh look, that's already happened with AMD turning their desktop APUs into server APUs.

AMD is working backwards now from how they have been. People are still anticipating SR Opteron to come first and it's not going to happen. It makes far more sense for AMD to go SR or EX gaming CPU first and then trickle over to server.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/8

Now look at this. AMD APU in different performance class than AMD CPU. FX 8350 fitting in just fine with 2600k and IB CPUs.

Top end APU fitting in below Core i3.

Do you see the point I am trying to make?

Gaming is important to AMD. APU puts AMD's CPU offerings below Intel's Core i3 in most situations. FX 8350 is in Intel $300+ territory most of the time.

AMD going APU only is basically throwing that all away. They are not going to do that.

Agreed, perhaps they are just finding a mature process for SR FX\whatever it will be called (inb4AMDSpectre 😛), maybe they do not need Opteron to make FX profitable and are just waiting for Piledriver to give way, the 8320 is a superb value after all. I am going to assume it will launch in a Phenom II-style on AM3+ 990FX, followed by AM4 1090FX whenever it launches.

 

[-Fran-]

Problem is... Although there is a market for high performing desktops thanks to games and some open/free applications, the real money is made in the server space. Not just because of the CPUs you sell, but the whole computing platforms around them. It all adds. We could use some real numbers in this, but I suspect the brute numbers might be closer than many suspect. For each server (1U, for instance) there's like 2 or 3 high performing desktops out there.

That's why we actually see more performing CPUs year over year. Not because of the consumer space, but because of server space. Where mobile is the new land where the gold rush is in full force; maybe winding down now, but still big enough to sway Intel into it.

Cheers!

 

[tracker45]

Is steamroller ever coming out ???

And who wants an APU ? they are just low performing CPUs combined with a low performing GPU.

I would rather just get a high performing CPU.

 

[The Q6660 Inside]

Is steamroller ever coming out ???

And who wants an APU ? they are just low performing CPUs combined with a low performing GPU.

I would rather just get a high performing CPU.

OEMs, their customers and of course, budget builders... The iGPU in Kaveri is quite good for the price, 7750 levels and pretty good per-core performance...

 

[griptwister]

Is steamroller ever coming out ???

And who wants an APU ? they are just low performing CPUs combined with a low performing GPU.

I would rather just get a high performing CPU.

OEMs, their customers and of course, budget builders... The iGPU in Kaveri is quite good for the price, 7750 levels and pretty good per-core performance...

And MANTLE could mean HD 7870 Performance 😀

 

[noob2222]

^^ ya. but on mantle, that 7870 would be like the 290X

 

[jdwii]

looks like i am the only one without a friend working for amd.

TT_TT

This is starting to sound a lot like grade school "My dad works for X company."

Is steamroller ever coming out ???

And who wants an APU ? they are just low performing CPUs combined with a low performing GPU.

I would rather just get a high performing CPU.

Unfortunately i can't even get my friend to say much about it and he works at Amd so i have to say i'm quite puzzled if it will come out

 

[jdwii]

looks like i am the only one without a friend working for amd.

TT_TT

This is starting to sound a lot like grade school "My dad works for X company."

Well i think its a well known fact that my friend at Amd can beat up Noobs friend anyday and it probably happens on the job all the time hell i even think they have his friend working on the arm project just to get rid of him

 

[Cazalan]

It is a really good strategy and it makes sense, it reminds me of that old saying, "don't put your eggs in one basket." From the sound of it, AMD had a basket that didn't work (SR FX chips). Normally, AMD would be doomed, but now they can go to their APU basket and be just fine. Doesn't mean the CPU basket is dead, dying, never coming back, etc. The other baskets can be used to levitate the basket that is not healthy.

The only problem with this is what does AMD expect people to buy for R290x CF or TF? They still should have a HEDT class CPU.

 

[Cazalan]

Funny, now we have a fourth explanation: they are released for satisfying investors.

I don't know what engineers you asked (x86?, GPU?). But I know that AMD head for servers, Feldman, says a completely different history:

http://www.eweek.com/servers/amds-feldman-arm-will-quickly-gain-market-share-in-servers/

The people I were talking to were long time x86 designers. When some executive comes in and says "hey, stop working on those American designed x86 cores, and go slap a few of these British ARM cores together". They're not exactly happy about it. Executives/management comes and goes but some of the old x86 designers are still around from the glory days.

What's the easiest way of satisfying investors who have been reading ARM hype for years? Offer ARM servers.

 

[con635]

Do amd really need to upgrade fx right now? I remember years ago how they moaned software didnt use the full power of their processors vs intel and I think they've played a blinder this year getting 8 core cpus into consoles. There was 75million (google) ps3s sold for example so now lets say a year from now how many 8 core cpus will there be in homes over the world? I think the fx 8***s will shine and maybe alot of fx 9***s will be used because of price drops and shine also in the 'hardware review world' in upcoming games. This should drive sales a bit and take the bad light off them allowing amd to be in a better position to call a new chip an 'fx'.

I have a few questions, does there even exist a game coded in a way that is uses 100% or near of every core on an 8 core chip? If there was such a game would that mean it would perform much better than an intel 4c/4t chip?

 

[juanrga]

Of course AMD will be competing against other players. Their only possibility is the combination of their expertise in servers/HPC with a custom core + GCN. Seattle will be released with standard cores, because AMD had no enough time to develop custom cores. They are doing it.

ARM is their only choice. There is no possibility that AMD gain significant market share using x86. In the first place AMD has no competitive x86 CPU product. In the second place, the market is migrating away from x86 monopoly. Feldman is right in that ARM is the future.

Ask yourself are they migrating away or is their another server laying around? X86 is here to stay Arm will not be controlling the world of every server and the people making the Arm processors won't make a living off of that alone. Most people at Amd knows this and again that is why this is one of their products not their only product. Its just another thing their getting into Arm servers might replace cheaper servers but not mid-range servers or high-end servers not a network administrator alive would want to use Arm on a high-end scale unless they like servers crashing a lot. I even have my doubts about the low-end range since you would need Arm compatible software and a lot of software was made by a small software company that only support x86 on windows or unix and this would now include software cost and hardware cost for changing for almost no apparent reason.

Yes, they are migrating because customers will be migrating (see quote below). That is the reason why Warsaw uses a PD refresh: there is no enough demand/market for AMD releasing a 8/12/16 core SR CPU for servers.

Warsaw provides an easy/cheap upgrade to customers of previous Opteron products. Warsaw is not aimed to take market share from Intel, because it is not competitive enough.

The head of the server division of AMD has already said that ARM will win. I already reproduced this slide before

Also when asked explicitly about Warsaw Feldman says:

He also told us that “Warsaw”, AMD’s follow up in the 2P/4P server market, will cater for a declining but still large market with institutional customers that will take longer to migrate to an ARM ecosystem.

Yes he says "migrate". Note he also says "declining". Of course this massive migration from x86 to ARM will not happen tomorrow neither next month. It will take years, but it will happen. It is not only AMD who is starting the migration to x86, HP, Dell and lots of other server makers are already preparing ARM servers.

Now i don't know what your saying Juanrga when you state (Amd has no competitive x86 products) This is simply untrue X86 is in their APU's which are extremely competitive with Intel if not better for the price of a celeron you can get a 3.8Ghz unlocked dual core processor with 160 radeon cores(i think that's the amount) which is perfect for a 1080P HTPC. For 130-140$(A10) you can get a processor with more compute and gaming power then any comparing product. A 6 core FX Piledriver is better then a I3 in overall gaming and normal application performance is a good deal.

So basically Amd has competitive APU's and CPU's for 200$ and under.

Please, read the whole paragraph; it is all about servers. No competitive x86 CPU for servers. That is why are replacing Opteron CPUs by Berlin APUs in servers. That is why AMD quotes the performance and efficiency of the whole APU = CPU+GPU in their presentation of server strategy/products. The efficiency and performance of the CPU alone are not enough for competing with Intel in servers.

 

[jdwii]

if i remember correctly piledriver was still on top of the list for performance line server CPU's which means Arm is not replacing the top of the line performers

 

[juanrga]

ARM will gain market share, there is no way it can't since there aren't any ARM servers yet. from 0% to 1% is a gain in market share. Between the big (for the sake of arguing just say 5) ARM manufacturers, they will each get 0.2% of that 1% market share if they all split sales evenly. Now say samsung spends a small fraction of their company to advertise. now they double their sales while the others get a smaller chunk.

AMD can't out-advertise multi-billion dollar companies. Thats why they are currently in the position they are in with Intel. Instead of competing with 1 other company, it will be 5 or more. (5 others already hold arm a-57 lisences)

Thank by the laugh. First, introducing a product in the market doesn't guarantee that market share increases. If the product is bad then the share goes from 0% to... 0%, because nobody buys it.

Second, that predicted 20% market share gain means that Intel will lose market share by a similar amount. There is no possibility that AMD can use x86 products to take server market share from Intel. That is why AMD is migrating to ARM to recover market share in servers:

http://www.slashgear.com/amd-seattle-chips-ditch-x86-for-arm-to-undermine-intels-server-market-18286798/

http://www.zdnet.com/amd-bets-arm-server-chip-experience-means-market-gains-7000016857/

http://www.pcworld.idg.com.au/article/465171/amd_reboots_server_strategy_first_arm_chips/

some relevant quotes:

The company said its first ARM server processors - which will be released in the second half of next year - will be faster and more powerful than its existing low-power x86 server processors.

aka why jaguar-based Opteron-X servers are being replaced by ARM-based Seattle servers. ARM is a superior product.

AMD will sell its first ARM processors alongside x86 server processors, which were also updated with new high-end Opteron chips due next year. But AMD expects ARM processors to outrun x86 chips in the long term.

aka Warsaw is only for legacy customers.

Observers have said the ARM-based chips could give AMD an edge over Intel, and the move could be as significant as its introduction of 64-bit server chips in 2003, and dual-core chips in 2004. Both moves gave AMD a competitive advantage over Intel, but server chip delays and the failure of chips based on the Bulldozer core ultimately cost AMD market share.

aka migrating to ARM is AMD only possibility to come back to servers. This explain why there is no 8/12/16-core SR Opterons.

By 2016 or 2017, ARM CPUs will have 20 percent of the server market, Feldman said. Right now the server market is dominated by Intel's x86 chips, but Feldman said large data centers will start porting software from x86 to ARM in 2015.

Aka even large customers will migrate to ARM fast.

ARM CPUs, which are cheaper and ship in higher volumes compared to the more expensive x86 chips, will prevail in the end, Feldman said.

I know that ARM will win in servers. I also know that ARM will win in supercomputers and that one day all desktop computers will be ARM-based.

Because they count it as APU. CPU alone the improvement in efficiency is of only 10-15%, which is not enough to compete with Intel.

so now SR cores are only 10-15%? I thought a 4-core Kaveri would compete with the I5 on the cpu level and trash it on the APU side ... but a 16 core SR opteron can't compete with a xeon?

Logic fails this one again.

While (Juanrga = wrong) {printf("JUANRGA IS NEVER WRONG")};

I know the four explanations are plain wrong. Up to now my sources have shown that are very reliable and others (including friends working at) are not so.

http://missryansgcseenglish.files.wordpress.com/2012/05/imaginary_friend.jpg

this source?

Thanks by another laugh.

Efficiency != performance

20--30% performance and 10-15% efficiency, but please continue misunderstanding everything as you always do.

 

[juanrga]

Ah, but you forget, AMD is also competitive in dCPU gaming PCs when it comes to modern games. You also forget AMD has basically gotten every game developer who is making a multi-platform to optimize their games for an AMD architecture.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/7

Look at a semi-modern engine like Frostbite 2. AMD FX 8350 offering best frametime experience (smoothest game experience) and some of the best frame rates.

This is why I am putting forth the idea that AMD is done with the server -> HEDT transition and instead might want to just flat out go straight dCPU for high end gaming platform with a high end Radeon card. We might from now on see AMD go APU/dCPU -> server.

Oh look, that's already happened with AMD turning their desktop APUs into server APUs.

AMD is working backwards now from how they have been. People are still anticipating SR Opteron to come first and it's not going to happen. It makes far more sense for AMD to go SR or EX gaming CPU first and then trickle over to server.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/8

Now look at this. AMD APU in different performance class than AMD CPU. FX 8350 fitting in just fine with 2600k and IB CPUs.

Top end APU fitting in below Core i3.

Do you see the point I am trying to make?

Gaming is important to AMD. APU puts AMD's CPU offerings below Intel's Core i3 in most situations. FX 8350 is in Intel $300+ territory most of the time.

AMD going APU only is basically throwing that all away. They are not going to do that.

I am not forgetting that. If you read my previous posts I already shared my expectation to see AMD releasing a high-end 8-core ~200W APU in future. I also said that 8-core APU could be complemented with a dGPU.

That 8-core high-performance APU could be Berlin replacement in servers and would compete against future Intel Socketed Xeon Phi and Nvidia ARM-CUDA SoCs in the HPC market.

What can be doing AMD today? I see two strategies:

1) Drops FX-4000 series and continue selling FX-6000/8000/9000 series for gaming PCs. Power consumption is not a concern on HEDT, and those chips get a performance upgrade with new software: consoles ports + MANTLE.

2) Drops FX-4000 series and upgrade FX-6000/8000/9000 series for gaming PCs. Those upgrades would be based in PD refresh. The new Warsaw comes only as 12/16 core (there is no 8-core Warsaw), which means that uses only 3M/4M PD. AMD could use those new dies for refreshing the six-core FX and octo-core FX series. They can even change the name. The same that "Opteron" has changed to "Warsaw". FX can be changed to a new brand.

But both strategies are for the short run. I expect AMD to migrate to a full APU strategy: APUs for laptops, desktops, servers, supercomputers...

This is my opinion.

 

[The Q6660 Inside]

Ah, but you forget, AMD is also competitive in dCPU gaming PCs when it comes to modern games. You also forget AMD has basically gotten every game developer who is making a multi-platform to optimize their games for an AMD architecture.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/7

Look at a semi-modern engine like Frostbite 2. AMD FX 8350 offering best frametime experience (smoothest game experience) and some of the best frame rates.

This is why I am putting forth the idea that AMD is done with the server -> HEDT transition and instead might want to just flat out go straight dCPU for high end gaming platform with a high end Radeon card. We might from now on see AMD go APU/dCPU -> server.

Oh look, that's already happened with AMD turning their desktop APUs into server APUs.

AMD is working backwards now from how they have been. People are still anticipating SR Opteron to come first and it's not going to happen. It makes far more sense for AMD to go SR or EX gaming CPU first and then trickle over to server.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/8

Now look at this. AMD APU in different performance class than AMD CPU. FX 8350 fitting in just fine with 2600k and IB CPUs.

Top end APU fitting in below Core i3.

Do you see the point I am trying to make?

Gaming is important to AMD. APU puts AMD's CPU offerings below Intel's Core i3 in most situations. FX 8350 is in Intel $300+ territory most of the time.

AMD going APU only is basically throwing that all away. They are not going to do that.

I am not forgetting that. If you read my previous posts I already shared my expectation to see AMD releasing a high-end 8-core ~200W APU in future. I also said that 8-core APU could be complemented with a dGPU.

That 8-core high-performance APU could be Berlin replacement in servers and would compete against future Intel Socketed Xeon Phi and Nvidia ARM-CUDA SoCs in the HPC market.

What can be doing AMD today? I see two strategies:

1) Drops FX-4000 series and continue selling FX-6000/8000/9000 series for gaming PCs. Power consumption is not a concern on HEDT, and those chips get a performance upgrade with new software: consoles ports + MANTLE.

2) Drops FX-4000 series and upgrade FX-6000/8000/9000 series for gaming PCs. Those upgrades would be based in PD refresh. The new Warsaw comes only as 12/16 core (there is no 8-core Warsaw), which means that uses only 3M/4M PD. AMD could use those new dies for refreshing the six-core FX and octo-core FX series. They can even change the name. The same that "Opteron" has changed to "Warsaw". FX can be changed to a new brand.

But both strategies are for the short run. I expect AMD to migrate to a full APU strategy: APUs for laptops, desktops, servers, supercomputers...

This is my opinion.

The 4000 series is just failed Vishera(s), you trash em or you sell them..

 

[8350rocks]

Spoiler

Exactly, everyone is misinterpreting the roadmaps and slides, including AMD representatives and head. The next quote from the above link is also wrong

A few weeks ago, AMD introduced X-Series Opteron CPUs, based on recently launched "Jaguar" microarchitecture. The roadmap states that these processors will be on the market until the second half of 2014, at which time AMD is going to replace them with "Seattle" systems-on-a-chip, built on ARM Cortex-57 IP core.

Your own words "This is not meant to even replace Opteron-X offerings, but more as an option for some who want ARM" are the TRUTH.

That quote is representative of the goal for Seattle; however, AMD is not so sure it's going to take off like a rocket. So much so, that they're not willing to bet the farm on it, and will keep x86 as an open option after their initial offering to the market. In fact, the ARM server SoC is an experiment, that's why no custom cores, that's why very little effort into adding just the barebones necessary for a server platform. That way, if (or more like when) it bombs, they're not out retarded figures worth of research dollars, and Opteron-X will still be there sitting on the back burner. If it takes off like a rocket, great, they make money. If it doesn't (like every other ARM SoC for enterprises out there)...then they're not out massive dollars for the research and development of a custom processor.

It's called "hedging bets"...if you've ever heard the term before.

EDIT: Do you think they would openly produce something and come out and say...(A)"We doubt this is going to work, but here you go...try it out...what do you think?", or, (B) is it far more likely that they would promote the hell out of it (see: PR slides you constantly quote), and try to get something, that they don't think is going to be any more than a niche, to take off so they can capitalize on untapped markets?

(A) or (B) you tell me...

I am going to say (B)

First it was an misinterpretation of the slide/roadmap. Now you have changed to the slide/roadmap only reflects the goal. Wow!

Several analysts, including AnandTech, have said that Seattle is a fantastic product. And everyone is predicting that ARM will take on Intel servers. The only problem for AMD comes from competence, who will be releasing custom cores.

AMD has already projects about custom ARM cores. The reason why are releasing Seattle with standard cores is because custom cores take time and money and AMD expect to have their server product first, before competitors release their custom chips. Sure you hear the "who strikes first strikes twice".

If AMD was to release both Opteron-X and Seattle, then they had not made that official roadmap where one replaces other. They had presented a server roadmap as the embedded roadmap where both products, Eagle (x86) and Hierofalcon (ARM), are offered in parallel and customers choose one.

In fact, if they present now a server roadmap where a product is replaced by other and latter recover the replaced product because the new didn't work then I can say you that it is better that AMD abandon servers, because would lose credibility.

As usual I will take your "my AMD friend said so" with a 99% of skepticism in base to history of your failed predictions.

What have I ever predicted and failed? I have made no predictions...

 

[8350rocks]

Ah, but you forget, AMD is also competitive in dCPU gaming PCs when it comes to modern games. You also forget AMD has basically gotten every game developer who is making a multi-platform to optimize their games for an AMD architecture.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/7

Look at a semi-modern engine like Frostbite 2. AMD FX 8350 offering best frametime experience (smoothest game experience) and some of the best frame rates.

This is why I am putting forth the idea that AMD is done with the server -> HEDT transition and instead might want to just flat out go straight dCPU for high end gaming platform with a high end Radeon card. We might from now on see AMD go APU/dCPU -> server.

Oh look, that's already happened with AMD turning their desktop APUs into server APUs.

AMD is working backwards now from how they have been. People are still anticipating SR Opteron to come first and it's not going to happen. It makes far more sense for AMD to go SR or EX gaming CPU first and then trickle over to server.

http://techreport.com/review/23750/amd-fx-8350-processor-reviewed/8

Now look at this. AMD APU in different performance class than AMD CPU. FX 8350 fitting in just fine with 2600k and IB CPUs.

Top end APU fitting in below Core i3.

Do you see the point I am trying to make?

Gaming is important to AMD. APU puts AMD's CPU offerings below Intel's Core i3 in most situations. FX 8350 is in Intel $300+ territory most of the time.

AMD going APU only is basically throwing that all away. They are not going to do that.

I am not forgetting that. If you read my previous posts I already shared my expectation to see AMD releasing a high-end 8-core ~200W APU in future. I also said that 8-core APU could be complemented with a dGPU.

That 8-core high-performance APU could be Berlin replacement in servers and would compete against future Intel Socketed Xeon Phi and Nvidia ARM-CUDA SoCs in the HPC market.

What can be doing AMD today? I see two strategies:

1) Drops FX-4000 series and continue selling FX-6000/8000/9000 series for gaming PCs. Power consumption is not a concern on HEDT, and those chips get a performance upgrade with new software: consoles ports + MANTLE.

2) Drops FX-4000 series and upgrade FX-6000/8000/9000 series for gaming PCs. Those upgrades would be based in PD refresh. The new Warsaw comes only as 12/16 core (there is no 8-core Warsaw), which means that uses only 3M/4M PD. AMD could use those new dies for refreshing the six-core FX and octo-core FX series. They can even change the name. The same that "Opteron" has changed to "Warsaw". FX can be changed to a new brand.

But both strategies are for the short run. I expect AMD to migrate to a full APU strategy: APUs for laptops, desktops, servers, supercomputers...

This is my opinion.

Let me clue you in here, AMD's market share will fall faster than a lead brick dropped out of an airplane if they abandon the Gaming PC market. In terms of desktops...that's their largest % market share. They're waiting for steamroller FX or successor to PD FX.

They are not abandoning HEDT. It is not happening, and there will be no HEDT ARM or HE Server ARM.

Opteron-X is not a high end part, if they equate ARM to that product, we're discussing micro servers running on ULP low end laptop grade parts (i.e. kabini grade solutions). That doesn't run a massive bank of servers at a financial institution or a super computer.

Your delusions are ever so grandiose on this one...

 

[jdwii]

"ARM will win in supercomputers" "all desktop computers will be ARM-based."

- "Thanks by another laugh."

 

[The Q6660 Inside]

"ARM will win in supercomputers" "all desktop computers will be ARM-based."

- "Thanks by another laugh."

inb4PowerPCbetakinovertheworld

 

[8350rocks]

"ARM will win in supercomputers" "all desktop computers will be ARM-based."

- "Thanks by another laugh."

inb4PowerPCbetakinovertheworld

LOL!!!

But you find PPC in servers and cars all over the world!!! Surely it will overtake x86 sometime soon!! /sarcasm.