New Zen microarchitecture details

[mysticjbyrd]

8 cores for mid-range......? So many people wouldn't afford to buy 8 core Zen if what bitsandchips said are true, why call it mid-range product.
But IMO 3.5Ghz for octal-core Zen is pretty likely because ES already hits 3Ghz.

Intel's enthusiast series has no competition, that's why they jack the price sky high. I wouldn't expect intel to drop their prices much, but AMD will likely be far more affordable.

 

[itsmydamnation]

OK, so you think AMD has put together a CPU core with Haswell-class x86 ST perf/clock and will be able to hit much higher clocks in a given power envelope than Intel is seemingly able to with Broadwell?

Is that what you are saying?

sure why not? they have 1/2 the vector width and 1/2 the load store and 1/2 the cache bandwidth requirement. you pay for that in power envelope.

but plz look at the last line you quoted.

 

[frozentundra123456]

How many watts do you suppose those extra two channels gobble up?

Anyway, I think it's probably worth keeping expectations in check around the clock speeds of the 8C/16T 95W Zen chip. It will be a good multithreaded performer most likely but I don't expect particularly high clocks, especially on a mobile-first process. I doubt Samsung had high performance, high frequency CPUs at the top of its mind when it defined and executed 14LPP.

This is a process that was first and foremost aimed at building mobile processors for Apple, Qualcomm, and Samsung System LSI.

Yep, all it takes is an April Fools post to get the hype train going again. 4ghz clocks, Haswell IPC, 8 cores at 95 watts, *and* AMD is going to price it as a bargain. OK....... sounds reasonable to me.

 

[mysticjbyrd]

OK, so you think AMD has put together a CPU core with Haswell-class x86 ST perf/clock and will be able to hit much higher clocks in a given power envelope than Intel is seemingly able to with Broadwell?

Is that what you are saying?

http://ark.intel.com/products/87718/Intel-Core-i7-5775R-Processor-6M-Cache-up-to-3_80-GHz

Intel® Core™ i7-5775R Processor
(6M Cache, up to 3.80 GHz)

TDP = 65 W

 

[Arachnotronic]

Yep, all it takes is an April Fools post to get the hype train going again. 4ghz clocks, Haswell IPC, 8 cores at 95 watts, *and* AMD is going to price it as a bargain. OK....... sounds reasonable to me.

The hype train around Zen is practically unreal, no doubt helped along by Lisa Su and Mark Papermaster. If Zen is going to be so great & be the nail-in-the-coffin for Intel that some of the AMD supporters claim, then why did Rory Read and Jim Keller bail before Zen's launch?

Didn't they want to bask in the glory that this amazing super new CPU core was going to bring them?

 

[mysticjbyrd]

The hype train around Zen is practically unreal, no doubt helped along by Lisa Su and Mark Papermaster. If Zen is going to be so great & be the nail-in-the-coffin for Intel that some of the AMD supporters claim, then why did Rory Read and Jim Keller bail before Zen's launch?

Didn't they want to bask in the glory that this amazing super new CPU core was going to bring them?

No one is claiming that.... That's a ridiculous strawman.

 

[itsmydamnation]

The hype train around Zen is practically unreal, no doubt helped along by Lisa Su and Mark Papermaster. If Zen is going to be so great & be the nail-in-the-coffin for Intel that some of the AMD supporters claim, then why did Rory Read and Jim Keller bail before Zen's launch?

Didn't they want to bask in the glory that this amazing super new CPU core was going to bring them?

You and your ilk have had every opportunity to put forward technical reasons why not. So far all you have got is "coz AMD".....

 

[Arachnotronic]

No one is claiming that.... That's a ridiculous strawman.

Absolutely they are. The bits and chips guy is claiming 3.7-3.8GHz base, 4.1-4.2GHz turbo, 8 cores/16 threads in a 95W power envelope. If this is true, and Zen hits around Haswell/Broadwell perf/clock, this is an Intel HEDT killer plain and simple.

 

[LTC8K6]

As an Intel fan, I hope Zen pushes Intel.

 

[mysticjbyrd]

I agree, that guys speculation seem lofty.

 

[IllogicalGlory]

What incentives would they have to stay at AMD? Maybe they're simply too expensive to keep on any longer than is needed to do their work. Do they get paid extra if Zen really is magical? Jim Keller will be worshiped as a god regardless. If Zen turns out to be, perhaps not magical, but still a great a product, industry insiders will all know who was responsible regardless, and that's what really matters to getting hired.

Still, a 3GHz clock to start with is already a good sign, but yeah I don't expect 3.8GHz base.

 

[Arachnotronic]

You and your ilk have had every opportunity to put forward technical reasons why not. So far all you have got is "coz AMD".....

Yeah, track record matters in the real world. You can go around asking for "technical reasons" all you want, but the reality is that it's not going to be easy for AMD to go from being perpetually behind Intel for about a decade to reaching parity/pulling ahead of them, especially given AMD's relative lack of resources.

AMD has lost a lot of its best engineers over the years and it has dramatically cut R&D spending while not really narrowing its focus. At the same time, it is going from having processes custom-made for them for high performance CPUs to using off-the-shelf foundry processes optimized for the mobile market.

These are all important things to consider, but the "AMD Zen is the second coming" crowd seems to ignore it all and just expects a miracle.

I don't think these expectations are reasonable.

 

[Arachnotronic]

What incentives would they have to stay at AMD? Maybe they're simply too expensive to keep on any longer than is needed to do their work. Do they get paid extra if Zen really is magical? Jim Keller will be worshiped as a god regardless, if Zen is magic, which it won't be.

Let's put it this way: if Zen is the real winner that some people believe, AMD will be in a much, much better financial position over the next couple of years and I am sure for developing such an amazing, company-saving CPU core, Keller would see a big fat raise & a ton of stock options whose value should soar as AMD's fortunes improve.

Instead, he decided to bail and move on to Tesla. It took AMD six months to find somebody willing to take on Keller's role, and it ultimately ended up being some guy nobody has ever heard of who previously led Broadcom's Vulcan CPU efforts.

People can demand "technical explanations" all they want, but following the money and the people is generally the best indicator, IMO.

 

[itsmydamnation]

Yeah, track record matters in the real world. You can go around asking for "technical reasons" all you want, but the reality is that it's not going to be easy for AMD to go from being perpetually behind Intel for about a decade to reaching parity/pulling ahead of them, especially given AMD's relative lack of resources.

AMD has lost a lot of its best engineers over the years and it has dramatically cut R&D spending while not really narrowing its focus.

First of all what are my expectations? My questions have been around with what we know why not?

Second, how much does it cost to develop a high performance CPU, The numbers i have heard is 400 million USD over 4 years, The big thing is time. You cant make these things go faster. Bulldozer was the wrong direction, but it takes 4 years to turn the ship around.

But here is the thing i have said so many times, AMD have a base to build from, how about you start by saying what is wrong with the logic here:

http://forums.anandtech.com/showpost.php?p=37929130&postcount=248

At the same time, it is going from having processes custom-made for them for high performance CPUs to using off-the-shelf foundry processes optimized for the mobile market.

lets look at the clock speeds of the products that bring Intel all the GP and revenue......
Core M, 1.4 to 2.9.
ULV ultrabook chips 2.3 to 2.8
>10 core server parts 2.0 to 3.0
phi ~1.1

funny how they sit largely in the same clock envelop as these "mobile targeted processes", i guess all GPU's are going to suck on this mobile optimized process as well..........

 

[dark zero]

Absolutely they are. The bits and chips guy is claiming 3.7-3.8GHz base, 4.1-4.2GHz turbo, 8 cores/16 threads in a 95W power envelope. If this is true, and Zen hits around Haswell/Broadwell perf/clock, this is an Intel HEDT killer plain and simple.

Intel was killing themselves their HEDT a long time before... the prices are ridiculous to start with. Also few applications can use all the threads (which affects AMD too).

Finally... you forgot optimization. Newer software are made only by Intel and the rest of the X86 companies will need more power to get similar results.

Is similar on the phone area... Intel with higher clocks, struggles a LOT against their ARM counterparts in Android applications (some games lags like crazy and even worse on some games like Hungry Shark Evolution and The King of Figthers XIII for Android) and the reason is optimization. Android was never meant to work flawlessly on Intel and Intel needs more power to compete against them.

 

[monstercameron]

Intel was killing themselves their HEDT a long time before... the prices are ridiculous to start with. Also few applications can use all the threads (which affects AMD too).

Finally... you forgot optimization. Newer software are made only by Intel and the rest of the X86 companies will need more power to get similar results.

Is similar on the phone area... Intel with higher clocks, struggles a LOT against their ARM counterparts in Android applications (some games lags like crazy and even worse on some games like Hungry Shark Evolution and The King of Figthers XIII for Android) and the reason is optimization. Android was never meant to work flawlessly on Intel and Intel needs more power to compete against them.

Wrong, google built android on java. Java is multiplstform.

 

[Abwx]

Not on Haswell-E, not quite, but remember that's 22nm. The 14LPP process that AMD is using is between Intel's 22nm and 14nm processes in terms of feature size, and likely energy efficiency as well.

.

In its current form GF s 14nm LPP LVT is either 15% more efficient or 20% faster than Intel s 14nm, given that the 14nm LPP sLVT will trade a part of the low static power for higher frequency ceiling those 4GHz+ figures are not even surprising...

If you have some number that contradict my estimation please post them, FTR the 14nm LPP LVT is 2.3x more efficient than GF s own 28nm HPP, and the sLVT will extend this number, i guess that people who are relying on deference will have a nice ride for their money...

 

[cbn]

Let's put it this way: if Zen is the real winner that some people believe, AMD will be in a much, much better financial position over the next couple of years and I am sure for developing such an amazing, company-saving CPU core, Keller would see a big fat raise & a ton of stock options whose value should soar as AMD's fortunes improve.

Instead, he decided to bail and move on to Tesla. It took AMD six months to find somebody willing to take on Keller's role, and it ultimately ended up being some guy nobody has ever heard of who previously led Broadcom's Vulcan CPU efforts.

People can demand "technical explanations" all they want, but following the money and the people is generally the best indicator, IMO.

Remember Keller also developed K12, the ARM core.

And I have been wondering if AMD's lack of Linux development was a strong reason for his departure.

No Linux development = no strong path for Keller's baby: K12. So why stay?

P.S. Regarding Linux development, I am primarily thinking of the Linux graphics driver (for a K12 APU) rather than Linux development for a Server K12 chip.

 

[DrMrLordX]

Absolutely they are. The bits and chips guy is claiming 3.7-3.8GHz base, 4.1-4.2GHz turbo, 8 cores/16 threads in a 95W power envelope. If this is true, and Zen hits around Haswell/Broadwell perf/clock, this is an Intel HEDT killer plain and simple.

No, it wouldn't be an "Intel HEDT killer", and no, nobody is making the claim you've inferred. Intel is a multi-billion dollar corporation that has survived fierce competition from AMD in the past. What makes you think Intel's HEDT platform, much less the entire corporation, would vanish into a cloud of dust should Zen actually perform as well as that bitsandchips guy seems to think? They'll just go back to the ol' playbooks and fight AMD the way they always have. They'd emerge with a better product lineup and with much of their market share intact.

Personally I think he estimations are far too optimistic, but we'll see soon enough.

 

[Exophase]

Remember Keller also developed K12, the ARM core.

And I have been wondering if AMD's lack of Linux development was a strong reason for his departure.

No Linux development = no strong path for Keller's baby: K12. So why stay?

P.S. Regarding Linux development, I am primarily thinking of the Linux graphics driver (for a K12 APU) rather than Linux development for a Server K12 chip.

Could be but I have my doubts that Keller really cares about AMD's GPU development. And I doubt there was ever a lot of market potential for an ARM-based SoC running "normal" Linux outside of servers.

He's got a long history of bouncing from company to company. There's been some criticism of AMD's internal structure and management these last several years. Maybe things were a lot worse than they were when he left the first time and he was disillusioned with the company.

 

[Dresdenboy]

Yep, all it takes is an April Fools post to get the hype train going again. 4ghz clocks, Haswell IPC, 8 cores at 95 watts, *and* AMD is going to price it as a bargain. OK....... sounds reasonable to me.

A bargain? Come on, now it's getting unrealistic again. My october guesstimations were a tad lower, but not by much.

 

[Dresdenboy]

Could be but I have my doubts that Keller really cares about AMD's GPU development. And I doubt there was ever a lot of market potential for an ARM-based SoC running "normal" Linux outside of servers.

He's got a long history of bouncing from company to company. There's been some criticism of AMD's internal structure and management these last several years. Maybe things were a lot worse than they were when he left the first time and he was disillusioned with the company.

Maybe he found it problematic, that the Zen leader and top engineer are a couple?

 

[NTMBK]

How many watts do you suppose those extra two channels gobble up?

An awful lot! Compare the Ivy Bridge i7 3770k with i7 4820k- a quad core with marginally higher base clock and same turbo, no integrated GPU, but quad channel memory and lots of PCIe lanes. TDP jumps from 77W to 130W.

I still think BnC are being awfully optimistic, mind you.

 

[AtenRa]

If ZEN will not be able to get at least Haswell IPC, at least the same clocks and better perf/watt with a node advantage (22nm vs 14nm) than Haswell-E (If bellow ZEN mArch is correct), then they better start looking for a new job.

Edit. Seems to me that ZEN is an EVOLUTION of Steamroller/Excavator CMT in to SMT.

Bulldozer and Steamroller

ZEN (by Dresdenboy)
http://dresdenboy.blogspot.gr/

Haswell mArch (exec Engine)

 

[cbn]

Let's put it this way: if Zen is the real winner that some people believe, AMD will be in a much, much better financial position over the next couple of years and I am sure for developing such an amazing, company-saving CPU core, Keller would see a big fat raise & a ton of stock options whose value should soar as AMD's fortunes improve.

Instead, he decided to bail and move on to Tesla. It took AMD six months to find somebody willing to take on Keller's role, and it ultimately ended up being some guy nobody has ever heard of who previously led Broadcom's Vulcan CPU efforts.

People can demand "technical explanations" all they want, but following the money and the people is generally the best indicator, IMO.

Remember Keller also developed K12, the ARM core.

And I have been wondering if AMD's lack of Linux development was a strong reason for his departure.

No Linux development = no strong path for Keller's baby: K12. So why stay?

P.S. Regarding Linux development, I am primarily thinking of the Linux graphics driver (for a K12 APU) rather than Linux development for a Server K12 chip.

Could be but I have my doubts that Keller really cares about AMD's GPU development. And I doubt there was ever a lot of market potential for an ARM-based SoC running "normal" Linux outside of servers.

A K12 APU would not work well without good Linux drivers even if had the best CPU cores. So that is the importance of the GPU as relates to K12.

As far a Linux, I think we are going to see it grow in importance. Just look at all the game developers porting to Linux now. How much of this is being done because they want to break free of x86 Windows vs. How much of this because they want to run ARM in the future?

Anyway, I bring this up because I don't think Zen or Zen's performance is the problem.