AMD "Zen" CPU Prototypes Tested, "Meet all Expectations"

[MxPhenom 216]

So is Zen going to be on a new socket? Also a new chipset to go with it with native USB 3.0 and PCIe 3.0 support, and all the new current stuff that has been going into Intel boards?

 

[john_]

It is 4 years from the introduction of Bulldozer architecture and from that day until today I still can't believe their colossal stupidity.

 

[ThomasDM]

Saw the news a couple of days ago. The source is basically some random guy on the Internet claiming he talked to someone who allegedly worked at AMD and heard from an ex-colleague that Zen is doing great.

Here's the original source:

By: lurker (lurker9000.delete@this.realemail.mail), October 29, 2015 3:12 pm

Regarding Zen performance, a guy who worked for AMD (at least his linkedin profile says that) and who, as he claims, worked on designing L2 cache for Zen and K12 said that their focus was to be competitive against Intel. He no longer works there but apparently his old colleague who still works there said Zen chips have already been tested and so far "it has met all expectation" and they "haven't found any significant bottlenecks". Apparently they haven't finalized the specifications for the clocks and TDP, but their partners in server market are "very excited".
It's not much detail, but I think if there was a problem from having only 2 AGUs, it would count as a significant bottleneck.
Also this is my first post ever, I just usually lurk here and this is the first time I have something useful to add to the discussion. Please no bully.

http://www.realworldtech.com/forum/?threadid=154302&curpostid=154823

 

[Fiery]

So is Zen going to be on a new socket? Also a new chipset to go with it with native USB 3.0 and PCIe 3.0 support, and all the new current stuff that has been going into Intel boards?

Yes, correct. AM4 is set to be rolled out in cca. 6 months from now, with the Bristol Ridge APU that still uses Excavator cores. Bristol Ridge is basically a desktop Carrizo on steroids. How much steroids? Hard to tell at this point, but I wouldn't expect much of a performance gain over the current Carrizo. Then, around the end of next year Summit Ridge CPU, using up to 8 Zen cores should be rolling out as a high-end desktop offering -- competing with the likes of Broadwell-E. Then, in the middle of 2017 we should see the first APUs using Zen cores, most likely 4 Zen cores, as Raven Ridge. All those 3 processors use the same AM4 platform with the Promontory chipset (well, FCH, so a single-chip chipset, just like Intel PCHs).

AMD will also roll out some very interesting many-core (16? 24? 32? we'll see) server CPUs as well, probably only in 2017, along with a Zen-inspired ARM core based APU for Android devices. Exciting times ahead, but it all will ride on how great Zen performs. It will make or break AMD.

Oh, and let's not forget Stoney (or Stoney Ridge), a low-end APU that will also arrive in the new AM4 socket (and also BGA). Imagine a Carrizo cut in half, that will be basically it. The point? Hard to tell. Probably some very low-power parts for ultrabooks and tablets, taking over the role of Jaguar/Puma core based APUs...

 

[RejZoR]

While I did jump to X99 platform early and didn't wait for Zen, I'm happy Zen is turning out the way AMD wanted. Because that's good for us consumers and AMD in general. Looking forward for the benchmarks

 

[AsRock]

I would Sooner Wait for Information from sources not connected to AMD publicity machine its been proven not to be Trusted

Sadly lately only to true.

 

[DeadSkull]

Hopefully Intel won't throw a pansy fit and throw billions at OEM pc makers forcing them to ignore AMD chips.

That's what happened when K7 and K8 lineup was dominating. AMD couldn't get any marketshare because all the OEM PC makers were forced to enter exclusionary contracts with Intel, aka Intel pulled a Mafia move.

 

[RejZoR]

And same happened with K7 and same happened with AXP's and same happened with Athlon64's etc etc. That's why AMD never really gets any decent share which is load of crocked BS. Still, at the end of the day, we face the price/performance ratio dilemma and that's why I have an Intel CPU in my PC even though I sometimes absolutely hate Intel for what they do.

 

[cdawall]

Hopefully Intel won't throw a pansy fit and throw billions at OEM pc makers forcing them to ignore AMD chips.

That's what happened when K6 was dominating. AMD couldn't get any marketshare because all the OEM PC makers were forced to enter exclusionary contracts with Intel, aka Intel pulled a Mafia move.

That issue was a lot more prevalent during K7 and K8 days. K6 1/2/3 were all very close performers to intel and k6-2 honestly sucked.

 

[Dave65]

didn't we hear the same hype from AMD when Bulldozer was on the horizon?
I hope for AMD's sake it is a true winner..

 

[rooivalk]

Hopefully Intel won't throw a pansy fit and throw billions at OEM pc makers forcing them to ignore AMD chips.

That's what happened when K6 was dominating. AMD couldn't get any marketshare because all the OEM PC makers were forced to enter exclusionary contracts with Intel, aka Intel pulled a Mafia move.

I don't remember K6 dominating?

Maybe the price? but Celeron 300A @ 450Mhz seems a better deal.

 

[Dent1]

I don't remember K6 dominating?

Maybe the price? but Celeron 300A @ 450Mhz seems a better deal.

Yeah because 1 old chart with 1 specific game proves your point.

Tom's Hardware 1997:

"The K6 233 will be priced lower than the Pentium MMX 200 and it's faster than this CPU in almost every respect"

"The current choice is only 'K6 or Pentium Pro' . Even when the Pentium II comes out officially (I know you can buy it already, but you can't buy boards, haha) it will be much too expensive for its little performance increase over the Pentium Pro."

"The Pentium Pro is a very good CPU and its internal L2 cache is making it unique. However the K6 is still even cheaper than that, it is just as fast"

http://www.tomshardware.com/reviews/intel,22-10.html

The Celeron 300A came out a full year after the K6. By then AMD was readying the K6-2.

 

[buildzoid]

So is Zen going to be on a new socket? Also a new chipset to go with it with native USB 3.0 and PCIe 3.0 support, and all the new current stuff that has been going into Intel boards?

Kaveri support PCI-e 3.0 natively.

 

[Ferrum Master]

Yeah because 1 old chart with 1 specific game proves your point.

I also don't remember it dominating really. OK Thunderbird was the first 1GHz stone but truly K8 was the first one to mop the floor with P4.

First of all no motherboards could hold a candle to i440BX and later i815, only when nforce2 came AMD got a good performing chipset.

 

[cdawall]

Yeah because 1 old chart with 1 specific game proves your point.

Tom's Hardware 1997:

"The K6 233 will be priced lower than the Pentium MMX 200 and it's faster than this CPU in almost every respect"

"The current choice is only 'K6 or Pentium Pro' . Even when the Pentium II comes out officially (I know you can buy it already, but you can't buy boards, haha) it will be much too expensive for its little performance increase over the Pentium Pro."

"The Pentium Pro is a very good CPU and its internal L2 cache is making it unique. However the K6 is still even cheaper than that, it is just as fast"

http://www.tomshardware.com/reviews/intel,22-10.html

One CPU? K6-2 sucked balls. Intel didn't really limit anything from AMD back then it was k7/k8 when there were issues.

 

[Dent1]

One CPU? K6-2 sucked balls. Intel didn't really limit anything from AMD back then it was k7/k8 when there were issues.

My point is the K6s were pretty dominate. It wasn't the sided Intel affair we see today.

 

[cdawall]

My point is the K6s were pretty dominate. It wasn't the sided Intel affair we see today.

K7 and K8 dominated. K6-2 wasn't really competitive and K6-3 wasn't stable.

 

[DeadSkull]

I don't remember K6 dominating?

Maybe the price? but Celeron 300A @ 450Mhz seems a better deal.

My bad.

K7 and K8 so from 01 to 06 when Core architecture first showed up AMD had excellent competitive performance.

 

[lilhasselhoffer]

This is PR fluff.

Tell me objectively:
1) What these expectations were.
2) What percentage of the initial goal is sufficient to "meet" them.
3) Quantify how production parts will match to the limited sample batches.


The entire article is AMD trying to say that Zen will be amazing. They're promising us this after they said the same about bulldozer, and they think we're stupid enough to believe it without a single question. I'd be insulted, if I had any respect for the AMD PR department left.


I'm hoping that manufacturing and engineering can pull off a genuine win here. Lord knows, we don't need another two generations of Intel offerings that make very little progress. That said, I'm not dumb enough to take this at face value. I'll only accept it once AMD demonstrates some independent verification of their PR.

 

[cdawall]

Considering who designed it I will be astounded if they are a flop for performance.

 

[GhostRyder]

AMD CPU engineer: "It didn't catch fire when we applied power."
AMD CPU engineering manager: "That's a wrap folks, start shipping 'em!"
AMD marketing: "To put a positive spin on the product we're going to focus on its fireproof capabilities in our brochures."
AMD board: "RAMP UP TO FULL PRODUCTION IMMEDIATELY!!!"

Someone's still mad they only have 3.5gb of VRAM.

Considering who designed it I will be astounded if they are a flop for performance.

I agree, mostly that this should be a competitive product especially if the price is right which is what they really need in the future. They need a chip that more OEM's will start putting in machines again that can offer consumers real advantages over other chips more than a better integrated GPU (Which is mostly only better on the mobile market).

 

[Dent1]

I hope AMD doesn't fall into the trap in thinking high performance will solve all their problems. AMD's biggest issue is internal, too much spending, borrowing and repaying loans at high interest. Their second biggest issue is the failure to capture a wide audience marketing wise, and then thirdly its CPU performance. They need to find a way of bringing all these changes together.

 

[HumanSmoke]

This is PR fluff.

Yes, but by "leaking" through anonymous "sources", AMD have plausible deniability. The company have played it this way since the whole Barcelona fake benchmarks and overoptimistic hype debacle. Much easier to feed rumour sites and watch social media spread the gospel. You can pretty much guarantee that official performance expectations will be few and rather vague. The "leaked" performance will however be off the charts. Signal-to-noise ratio should be predictably very low.

A good indicator of the architectures actual performance will be HPC/Server/Data Center contracts being signed (or at least MoU's being announced) well in advance of any launch, just as big iron companies queuing up to jettison AMD ahead of Zambesi/Vishera's entrance was a portent of their relative attributes.

 

[Musaab]

Maybe AMD is rely going back? Maybe we can see K7/K8 era again? Maybe we are looking a K9? Maybe Lisa Su is not BSing like the other AMD CEOs. I am waiting.

 

[alucasa]

Well, it would be nice to build an AMD rig. I think it's been years... Actually, I haven't touched AMD after i7-920. I am stuck with Nvidia for GPU tho due to CUDA rendering.