AMD Zen (Roadmaps, Quad Core Unit, Block Diagram, 32 Core Opteron)

[patrickjp93]

Looking at that, Intel don't use the extra space from removing the iGPU on Xeons to increase the size of the cores, do they?

They do, but it primarily goes into cache and the quad-channel memory controller as well as ECC and some security features stuff useful for businesses.

[Opcode]

Updated roadmaps added. Also a single Zen core will be below 10 mm².

 

 

[patrickjp93]

Updated roadmaps added. Also a single Zen core will be below 10 mm².

 

 

-double snip-

That makes me heavily suspect they're going to keep HDL going in addition to the process shrinks, which actually makes a ton of sense when it comes to trying to put a dent in Intel's stranglehold on the HPC market. I suspect HDL is going to become a commonly used technique since Dannard (sp.?) Scaling is gone, especially if AMD starts looking remotely performance-competitive with Intel in desktop and server markets again.

[Humbug]

Wow, so the new roadmap confirms that the high end desktop replacement for FX piledriver will be a pure CPU.

The new FX will be an 8-core Zen CPU, replacing fx-8350.

I thought it would be an APU.

[Opcode]

That makes me heavily suspect they're going to keep HDL going in addition to the process shrinks, which actually makes a ton of sense when it comes to trying to put a dent in Intel's stranglehold on the HPC market. I suspect HDL is going to become a commonly used technique since Dannard (sp.?) Scaling is gone, especially if AMD starts looking remotely performance-competitive with Intel in desktop and server markets again.

I imagine so as Excavator measures 14.48 mm² per core pair (Steamroller 18.61 mm²). Although the core should be bigger than most expect (14nm + HDL) as K10 was just under 10 mm² on 32nm. Zen will probably offer double the performance over Piledriver at lower clocks (I would expect 3.5 GHz range for reference clocks). I wouldn't say performance scaling is gone as Zen has the opportunity to surpass Intel in IPC only validating that if you actually do something more than just rely on die shrinks for performance you can actually make larger advances with your product (*cough* Intel *cough*). It's not like Intel has hit a performance wall or silicon limitation as that is not even a remote possibility. Being stagnant like they are for so long could be a problem for the company. Even their mobile market that they lose tons of money in has been stagnant for so long. Intel didn't push x86 hard enough by offering a truly solid product for manufactures. With AMD getting into the low powered mobile SoC ecosystem with their ARM based products if Intel doesn't do something quick they could be pushed right out of the ecosystem not only by AMD but also by the latest ARM efforts of all companies (Nvidia, Apple, ARM, etc). I made the prediction a while back that ARM wasn't going anywhere and that prediction is living up to its premise.

 

Wow, so the new roadmap confirms that the high end desktop replacement for FX piledriver will be a pure CPU.

The new FX will be an 8-core Zen CPU, replacing fx-8350.

I thought it would be an APU.

I think AMD is going after Intel's enthusiast platform (LGA 2011-v3) with Summit Ridge.

[patrickjp93]

I imagine so as Excavator measures 14.48 mm² per core pair (Steamroller 18.61 mm²). Although the core should be bigger than most expect (14nm + HDL) as K10 was just under 10 mm² on 32nm. Zen will probably offer double the performance over Piledriver at lower clocks (I would expect 3.5 GHz range for reference clocks). I wouldn't say performance scaling is gone as Zen has the opportunity to surpass Intel in IPC only validating that if you actually do something more than just rely on die shrinks for performance you can actually make larger advances with your product (*cough* Intel *cough*). It's not like Intel has hit a performance wall or silicon limitation as that is not even a remote possibility. Being stagnant like they are for so long could be a problem for the company. Even their mobile market that they lose tons of money in has been stagnant for so long. Intel didn't push x86 hard enough by offering a truly solid product for manufactures. With AMD getting into the low powered mobile SoC ecosystem with their ARM based products if Intel doesn't do something quick they could be pushed right out of the ecosystem not only by AMD but also by the latest ARM efforts of all companies (Nvidia, Apple, ARM, etc).

 

 

I think AMD is going after Intel's enthusiast platform (LGA 2011-v3) with Summit Ridge.

Intel's under no threat. Think 6 moves ahead instead of 5 and the answer becomes obvious: let AMD catch up, get itself on good financial footing by 2019 when the debt comes due, and open the floodgates again to deliver a new Sandy Bridge. Jim Keller is far from a God above every other chip designer the world has to offer. He has his equals, and most of them are employed by Intel, and IBM has all but 1 of the others. Nvidia may have the last, but it's not yet totally obvious.

 

If Intel can keep AMD in a state of good competition and pressure Nvidia by proxy on two fronts, it's victory for team blue even if AMD does better than in previous years. Intel just has to get AMD to the point it has enough money to live but not enough to speed up its rather slow development cycles, and then Intel can just jump ahead and stay ahead with no risk of gaining IBM or Samsung as an x86 competitor.

[Delphi1662]

Wow, so the new roadmap confirms that the high end desktop replacement for FX piledriver will be a pure CPU.

The new FX will be an 8-core Zen CPU, replacing fx-8350.

I thought it would be an APU.

I dont think the FX is a Zen CPU it says Piledriver their not sure why they are doing a refresh of this platform so close to the launch of Zen 

[Eroda]

Intel's under no threat. Think 6 moves ahead instead of 5 and the answer becomes obvious: let AMD catch up, get itself on good financial footing by 2019 when the debt comes due, and open the floodgates again to deliver a new Sandy Bridge. Jim Keller is far from a God above every other chip designer the world has to offer. He has his equals, and most of them are employed by Intel, and IBM has all but 1 of the others. Nvidia may have the last, but it's not yet totally obvious.

 

If Intel can keep AMD in a state of good competition and pressure Nvidia by proxy on two fronts, it's victory for team blue even if AMD does better than in previous years. Intel just has to get AMD to the point it has enough money to live but not enough to speed up its rather slow development cycles, and then Intel can just jump ahead and stay ahead with no risk of gaining IBM or Samsung as an x86 competitor.

 

yeah i suppose the last thing intel wants is someone with a wallet as big as them to buy up AMD and start pouring in the cash to make them have a really tough rival.. but it could be that intel has been a little bit lax and trying to develop for other markets leaving the desktop market a little bit vulnerable. but that as well could be part of a plan

[Opcode]

Intel's under no threat. Think 6 moves ahead instead of 5 and the answer becomes obvious: let AMD catch up, get itself on good financial footing by 2019 when the debt comes due, and open the floodgates again to deliver a new Sandy Bridge. Jim Keller is far from a God above every other chip designer the world has to offer. He has his equals, and most of them are employed by Intel, and IBM has all but 1 of the others. Nvidia may have the last, but it's not yet totally obvious.

 

If Intel can keep AMD in a state of good competition and pressure Nvidia by proxy on two fronts, it's victory for team blue even if AMD does better than in previous years. Intel just has to get AMD to the point it has enough money to live but not enough to speed up its rather slow development cycles, and then Intel can just jump ahead and stay ahead with no risk of gaining IBM or Samsung as an x86 competitor.

If you think Intel is under no threat than you may be mistaken. AMD is not coming back swinging at just the x86 ecosystem but also in markets that Intel confides in. What do you think is going to happen if AMD prices K12 extremely competitively. Intel has yet to shed a profit in the mobile market space and their massive net loss will only continue to grow as other companies suck up market share. Effectively reducing Intel's budget to invest (or really lose) into that segment. ARM has always been a better mobile architecture anyways as Intel cannot match its performance per watt (A8X and Denver are proof of that). If AMD can push Amur hard this year to where it makes it into a few well known cell phones and tablets then they will have the relations ready for next years K12. The same for Summit Ridge that is targeting the 2011-v3 platform just at a much cheaper price point. Even if Zen is ~10% slower than Skylake people will still favor it due to it's massive price difference. Then you got Bristol Ridge with full HSA 1.0 compliance and we've yet to know what the iGPU has to offer but DDR4 support will be a massive help at relieving memory bandwidth bottlenecks. A high performance product that sells well may finally be what it takes to influence the x86 software ecosystem towards HSA. Then you also got the 32 core Opteron and 16 core HPC APU that will be hitting the server ecosystem along with K12. Keep in mind Intel's success is really only a success based on competitive downfalls. Without competitive products they can monopolize the market which is why their architecture is only taking baby steps to milk their consumers. Every company is threatened by every company that's how the industry works. Otherwise there would be no competitive factor and a true monopoly would be established.

 

I dont think the FX is a Zen CPU it says Piledriver their not sure why they are doing a refresh of this platform so close to the launch of Zen 

I don't think they are planning on refreshing the FX lineup. It's on the chart more than likely as a placeholder to show what the company will be offering throughout 2015.

[Dabombinable]

I don't think they are planning on refreshing the FX lineup. It's on the chart more than likely as a placeholder to show what the company will be offering throughout 2015.

Which is pretty much the same CPU's they've been peddling for the last 3-4 years.

[Opcode]

Which is pretty much the same CPU's they've been peddling for the last 3-4 years.

Yep, they still got to broadcast it somehow as it's still a product they need to sell regardless to do away with inventory.

[Humbug]

I dont think the FX is a Zen CPU it says Piledriver their not sure why they are doing a refresh of this platform so close to the launch of Zen

The desktop roadmap has two columns. On the left is 2015; it shows 8 core piledriver fx as the top performance part. On the right shows 2016; piledriver is finally gone and replaced with an 8-core Zen performance CPU. That will be the new top performance desktop part. Obviously they have not named this product yet; but we can speculate that it will be the new FX since it directly replaces the old FX. Either way doesn't matter what they call it the important thing is that we are finally getting a real 100% new replacement for piledriver FX, rather that tweaks and iterations.

[cesrai]

" The Opteron needs all the silicon space for the L2, L3 cache as well as its Zen x86 cores."

The Desktop CPU's need the exact same thing,yet we get apu fucking morons.

You know AMD offers 860k, 760K, 750K ???

[Dabombinable]

Yep, they still got to broadcast it somehow as it's still a product they need to sell regardless to do away with inventory.

Which sucks for us-the consumer, considering their original FX line wasn't that impressive in the first place-and the fact that it barely improved over time.

[Opcode]

Which sucks for us-the consumer, considering their original FX line wasn't that impressive in the first place-and the fact that it barely improved over time.

I think Zen will launch sooner than expected so conserving money and staying away from a Bdver3 or Bdver4 FX was a good idea.

[Dabombinable]

I think Zen will launch sooner than expected so conserving money and staying away from a Bdver3 or Bdver4 FX was a good idea.

Yep. TBH as long as they make something that can beat my X5450 and i5 4440 in every single way, I'll be considering a Zen CPU, considering the flexibility that's planned for the FM2/FM2+ sockets.

[7850OC]

nice 

[SeriousDad69]

I think Zen will launch sooner than expected so conserving money and staying away from a Bdver3 or Bdver4 FX was a good idea.

When do you think Zen is coming out? I'm hoping for Q1 2016. I'm already setting some money aside for an upgrade and Intel and Nvidia can suck me.

[That Norwegian Guy]

You know AMD offers 860k, 760K, 750K ???

 

Which are slower than a 5 year old Intel quad core (i5-2400)

[cesrai]

Which are slower than a 5 year old Intel quad core (i5-2400)

Wow such point much response.

[Guest Generallee]

 

Yes, I'm aware of hypocrisy very much (looks at Intel's 16-core)

 

My fucking sides

[Sharkyx1]

You know AMD offers 860k, 760K, 750K ???

Which also use die space for gpus but they are deactivated, it's not any better

[Opcode]

Quick update, the slides posted for both the desktop and mobile roadmaps are supposedly fake says AMD.

[TechFan@ic]

Quick update, the slides posted for both the desktop and mobile roadmaps are supposedly fake says AMD.

They asked for these "fake" slides to be taken down. Even though they traditionally never address rumors.

This only makes them appear more valid in my estimation.

[Eroda]

Quick update, the slides posted for both the desktop and mobile roadmaps are supposedly fake says AMD.

 

 

fake  maybe the timelines are fake but amd moving to a single unified socket has been talked about for a long time now. i cant see AMD waiting till 2017 for Zen hopefully zen drops Q1 2016 that would make me happy... also  the affect of DDR4 on APU's is meh, i want to see HBM Memory on an APU 2Gb of HBM imagine the possibilities