[Techpowerup] AMD "Zen" CPU Prototypes Tested, "Meet all Expectations"

[Phynaz]

Lets keep it simple, lets keep ti technical,

Does the Intel Sales Team (IST) have any technical arguments to counter the 40% increase of IPC with the available technical details of ZEN Architecture we know today???
Because every time the IST talks in this topic is all about AMDs track record and financial BS.

"We expect across a wide variety of workloads for Barcelona to outperform Clovertown by 40 percent"

"We came to the conclusion that, given the capabilities and performance with the monolithic design, it was clearly the right answer"

-AMD

http://www.cnet.com/news/amd-go-to-barcelona-over-clovertown/

 

[MiddleOfTheRoad]

Believing a hype based on a single unspecified metric from a company who notoriously lied with every new major performance CPU uarch the last 10 years?

You call that open minded?

Everything point to that AMD cant deliver what some people fanatically expect.

As compared to what? A company who notoriously lied with every new major performance CPU by rigging benchmarks which artificially sandbagged competitor's products for the last 15 years?

Although, it is always fun to watch you lecture someone about being open minded.

 

[Abwx]

"We expect across a wide variety of workloads for Barcelona to outperform Clovertown by 40 percent"

"We came to the conclusion that, given the capabilities and performance with the monolithic design, it was clearly the right answer"

-AMD

http://www.cnet.com/news/amd-go-to-barcelona-over-clovertown/

They said clovertown, a server CPU, so they were talking of HPC and professional softs, not of Sysmark...

Intel's own marketing material seems to admit that a Xeon E5472 with 800MHz memory is just as a fast as AMD's quad-core at 2GHz.

http://www.anandtech.com/show/2386/11


Now can you leave this thread alone and spare us your off topic..?.

 

[Yuriman]

I wonder what clockspeeds AMD will even be targeting with Zen? It's undoubtedly going to be a very large chip (in its large core count configurations), and the TDP rumors going around suggest to me that it's going to be sub-3ghz, if the scaling that Intel has with Skylake is any indication.

4c/8t:
Skylake 6700K - 4ghz base, 4.2ghz turbo, 91w TDP
Skylake 6700 - 3.4ghz base, 4.0ghz turbo, 65w TDP
Xeon 1260Lv5 - 2.9ghz base, 3.9ghz turbo, 45w TDP
Xeon 1240Lv5 - 2.1ghz base, 3.2ghz turbo, 25w TDP

^ Basic math suggests to me that you could probably have a 16c/32t Skylake chip with ~2ghz baseclock / ~3ghz+ turbo within the same TDP as a 6700K. Those last mhz really come at a huge power consumption cost. Or, an 8c/16t chip with ~3ghz/4ghz at around the same TDP.

If AMD is going to be releasing 8-16 core chips, what do we really expect the clockspeeds to be? I'm willing to bet it will at the very least be a little lower than Skylake's with similar core counts.

 

[Phynaz]

They said cloverton, a server CPU, so they were talking of HPC and professional softs, not of Sysmark...

HPC isn't a "wide variety of workloads".

"So a former engineer who no longer works on the project says "5% less" and a current engineer who IS working on the project tells me "more IPC."

Why do you choose to believe the one with less access to today's data?"

-AMD
http://forums.anandtech.com/showpost.php?p=30380104&postcount=425

Turns out the former engineer was right.

"I have said it several times. IPC WILL BE HIGHER."

-AMD
http://forums.anandtech.com/showpost.php?p=30346682&postcount=101

AMD's track record speaks for itself.


But yes, let's get back on topic as you say. So let's assume that Zen "meets all expectations" as the thread title says. Show us specific measurable details of what AMD's expectations are of Zen, along with the validation results showing those expectations have been met.

If you can't do that then your posts are all nothing but hot air.

I'll be waiting.

 

[Abwx]

bla bla

Ignore list is the best answer to people who thread crap willfully...

Edit : The thread look much cleaner now...

I wonder what clockspeeds AMD will even be targeting with Zen? It's undoubtedly going to be a very large chip (in its large core count configurations), and the TDP rumors going around suggest to me that it's going to be sub-3ghz, if the scaling that Intel has with Skylake is any indication.


^ Basic math suggests to me that you could probably have a 16c/32t Skylake chip with ~2ghz baseclock / ~3ghz+ turbo within the same TDP as a 6700K. Those last mhz really come at a huge power consumption cost. Or, an 8c/16t chip with ~3ghz/4ghz at around the same TDP.

If AMD is going to be releasing 8-16 core chips, what do we really expect the clockspeeds to be? I'm willing to bet it will at the very least be a little lower than Skylake's with similar core counts.

All these point have already been adressed, just read through the thread, although i must admit that it s not easy given the huge noise..

 

[Phynaz]

Ignore list is the best answer to people who thread crap willfully...

Edit : The thread look much cleaner now...



All these point have already been adressed, just read through the thread, although i must admit that it s not easy given the huge noise..

Ok, so you can't provide anything to back up your posts. Thanks for confirming.

 

[Phynaz]

Lets keep it simple, lets keep ti technical,

Does the Intel Sales Team (IST) have any technical arguments to counter the 40% increase of IPC with the available technical details of ZEN Architecture we know today???
Because every time the IST talks in this topic is all about AMDs track record and financial BS.

The Intel Sales Team is doing a job the ADF envies, considering Intel has a 98% share of the x86 processor market.

You should actually thank them, how much do Intel sales contribute to your personal income? You have said you sell more Intel than AMD systems.

 

[Dresdenboy]

Ok, so you can't provide anything to back up your posts. Thanks for confirming.

The topic's original source couldn't make sure, if those expectations were about a cache or the whole core or even chip.

And absence of evidence is no evidence for absence.

 

[Vesku]

So Intel has 8C/16T + iGPU, 4C/8T + iGPU, 2C/4T + iGPU (?) for 10nm.

What does AMD launch for Zen APU?

4C/8T + "big" iGPU?

2C/4T + "medium" iGPU?

(Or maybe instead of that 2C/4T + medium iGPU they instead launch a 4C/8T with "small iGPU" and bin as 3C/6T for 5W? Or maybe worst case scenario take the 2 best of 4 cores for 5W?)

I can only guess. I'd think they'd try to keep die count down and go with 2C/4T and 4C/8T with reasonably sized iGPUs matched to DDR4 memory speed and to compare favorably to expected Intel iGP performance. Any performance SKUs will most likely come from their server lineup and would mean more cores smaller iGP, maybe 8C/16T or 10C/20T conservatively clocked to work within AM4 (or whatever they will call it) motherboard TDP. Probably try to get Apple to buy a customized APU with HBM.

 

[frozentundra123456]

The topic's original source couldn't make sure, if those expectations were about a cache or the whole core or even chip.

And absence of evidence is no evidence for absence.

Yes, but "the burden of proof is with the positive".

 

[DrMrLordX]

So Intel has 8C/16T + iGPU, 4C/8T + iGPU, 2C/4T + iGPU (?) for 10nm.

What does AMD launch for Zen APU?

4C/8T + "big" iGPU?

2C/4T + "medium" iGPU?

(Or maybe instead of that 2C/4T + medium iGPU they instead launch a 4C/8T with "small iGPU" and bin as 3C/6T for 5W? Or maybe worst case scenario take the 2 best of 4 cores for 5W?)

I don't think anybody really knows. All the APU products are coming under the monickers of Raven Ridge and Basilisk. I expect the low-power solutions for mobile to be Basilisk. Those are probably going to be 2c/4t implementations. There is still that massive 200-300W behemoth APU, for an unknown socket (hopefully AM4 but you never know) with totally unknown specs. But if you look at the power requirements for Fury Nano, for example, you can see a massive GCN 1.2 GPU + HBM fitting inside a sub-200W envelope on a 28nm process. On 14nm LPP, that entire GPU could sit alongside a 95W 8c/16t and fit inside a 300W power envelope. Probably.

What fits in-between is anyone's guess, and the roadmaps are not clear about that.

For the sake of the argument, what is the minimum increase a Zen core must gain in IPC in order to reach the same throughput as XV? (considering SMT brings smaller benefits than CMT)

That's an excellent question, which leads us to question how AMD has chosen to use the acronym "IPC".

The definition favored by those who focus on performance from single-threaded software shows us a very different picture than if we focus on overall performance, that is, throughput.

We know that Zen is being focused on the server market. AMD is likely not all that concerned with how Zen is going to perform when running software that uses n/2 threads, where n happens to be the total thread capacity of any given Zen processor. So, they are more interested in massively-parallel workloads, such as those that can load up 16+ threads, rather than those workloads that struggle to utilize 8 cores or less.

It would genuinely surprise me if their commentary regarding IPC is somehow related to how fast Zen will be when handling a single thread per core vs. XV handling a single thread per module - that is a comparison that has almost no meaning in Zen's target market. Also, from the technical details we have received about Zen, it certainly appears as though Zen has at least as many (if not more) resources available per core versus an XV module. Zen's performance characteristics may be lopsided in favor of the first thread per core thanks to an SMT implementation that may only add 30% total throughput per core, but overall, I would expect a Zen core to outperform an XV module when handling the same two threads.

To reiterate what I've said before, if each Zen core + SMT can not put up 40% better performance than an XV module when handling two threads from some average workload, then AMD has spread misinformation. They should be more clear about exactly what they mean by IPC, especially in a contentious comparison involving a relatively-exotic design like XV.

"We expect across a wide variety of workloads for Barcelona to outperform Clovertown by 40 percent"

"We came to the conclusion that, given the capabilities and performance with the monolithic design, it was clearly the right answer"

-AMD

http://www.cnet.com/news/amd-go-to-barcelona-over-clovertown/

There is nothing technical about that argument. It's a simple matter of trust.

Either you trust AMD to deliver on their promises, or you don't. If you don't, that's fine, but again, that is not based on any technical information we have received to date about Zen or it's 14nm LPP process.

 

[AtenRa]

That's an excellent question, which leads us to question how AMD has chosen to use the acronym "IPC".

The definition favored by those who focus on performance from single-threaded software shows us a very different picture than if we focus on overall performance, that is, throughput.

We know that Zen is being focused on the server market. AMD is likely not all that concerned with how Zen is going to perform when running software that uses n/2 threads, where n happens to be the total thread capacity of any given Zen processor. So, they are more interested in massively-parallel workloads, such as those that can load up 16+ threads, rather than those workloads that struggle to utilize 8 cores or less.

It would genuinely surprise me if their commentary regarding IPC is somehow related to how fast Zen will be when handling a single thread per core vs. XV handling a single thread per module - that is a comparison that has almost no meaning in Zen's target market. Also, from the technical details we have received about Zen, it certainly appears as though Zen has at least as many (if not more) resources available per core versus an XV module. Zen's performance characteristics may be lopsided in favor of the first thread per core thanks to an SMT implementation that may only add 30% total throughput per core, but overall, I would expect a Zen core to outperform an XV module when handling the same two threads.

To reiterate what I've said before, if each Zen core + SMT can not put up 40% better performance than an XV module when handling two threads from some average workload, then AMD has spread misinformation. They should be more clear about exactly what they mean by IPC, especially in a contentious comparison involving a relatively-exotic design like XV.

Well as IPC is actually Throughput, the 40% higher IPC claim could indeed mean 40% higher Throughput of ZEN Core (SMT included) vs Excavator Module (CMT) and has nothing to do with Single Thread Performance.

 

[itsmydamnation]

Yes, but "the burden of proof is with the positive".

Yes, but "the burden of proof is with the negative..............

 

[cbn]

So Intel has 8C/16T + iGPU, 4C/8T + iGPU, 2C/4T + iGPU (?) for 10nm.

What does AMD launch for Zen APU?

4C/8T + "big" iGPU?

2C/4T + "medium" iGPU?

(Or maybe instead of that 2C/4T + medium iGPU they instead launch a 4C/8T with "small iGPU" and bin as 3C/6T for 5W? Or maybe worst case scenario take the 2 best of 4 cores for 5W?)

I can only guess. I'd think they'd try to keep die count down and go with 2C/4T and 4C/8T with reasonably sized iGPUs matched to DDR4 memory speed and to compare favorably to expected Intel iGP performance. Any performance SKUs will most likely come from their server lineup and would mean more cores smaller iGP, maybe 8C/16T or 10C/20T conservatively clocked to work within AM4 (or whatever they will call it) motherboard TDP. Probably try to get Apple to buy a customized APU with HBM.

Going by historical trend, It is somewhat predictable AMD would go 2C/4T Zen at the low end (and then try to compensate with more iGPU than Intel)

But then Intel will have 14nm 2C/4T leftovers of various types plus whatever 10nm they launch on mobile.

 

[bystander36]

Well as IPC is actually Throughput, the 40% higher IPC claim could indeed mean 40% higher Throughput of ZEN Core (SMT included) vs Excavator Module (CMT) and has nothing to do with Single Thread Performance.

Of course all these marketing blurbs are always vague. You never really know what they mean by that until you see the end result.

I assume all news on these topics to be cherry picked, regardless of who says it, or who it is about. Hence the caution. It's hard to take any of this at face value. The only thing I do take away from it all, is that AMD is making progress.

 

[NostaSeronx]

Plan A;
Summit Ridge FX14 -> 8C/16T "Zen"
Raven Rdige FX14 -> 4C/8T "Zen" + 1 Shader Engine "GCN4" [16 CUs]

Plan B;
Viper Gecko 22FDX -> 20C/20T "Tunnelborer"
Basilisk Gecko 22FDX -> 10C/10T "Tunnelborer" + 1 Shader Engine "GCN4" [16 CUs]

 

[Dresdenboy]

Wouldn't disabling SMT add more SKU options?!

Khalid mentions 2-8 core FX... with core count increasing by 2.

 

[Dresdenboy]

Plan A;
Summit Ridge FX14 -> 8C/16T "Zen"
Raven Rdige FX14 -> 4C/8T "Zen" + 1 Shader Engine "GCN4" [16 CUs]

Plan B;
Viper Gecko 22FDX -> 20C/20T "Tunnelborer"
Basilisk Gecko 22FDX -> 10C/10T "Tunnelborer" + 1 Shader Engine "GCN4" [16 CUs]

Did you discover some U-Boat projects at AMD? "Tunnelbohrer" would be German.

 

[Fjodor2001]

AMD Zen FX CPU Product Release Set For Q4 2016 – APUs Coming In 2017 With Comparable Performance To The PS4 And XBOX ONE

http://wccftech.com/amd-zen-launch-q4-2016/

AMD Zen FX CPUs Set To Debut In Q4 2016, APUs In 2017

Today’s update comes after we have finally managed to confirm through sources exclusive to WCCFTech when AMD plans to release its latest inception to the market. Zen is still too far out in time for anyone to accurate pinpoint an exact release date, at this point in time it would be very challenging even for AMD to commit to a specific date internally. Still, we’ve learned that AMD’s planning to debut Zen based desktop FX CPUs by Q4 2016, just in time for the holiday season. While Zen based APUs are expected to debut at a later date in 2017.

AMD Zen FX CPUs Will Come With A Bunch Of New Goodies

Fortunately that’s not all we have for you today, as we’ve managed to learn about a number of additional interesting tidbits as well. There are several SKUs planned based on the Zen CPU die, with 8, 6 and 4 cores with the possibility of even higher core counts later on. The Zen core itself is surprisingly compact and power efficient. The CPU die itself isn’t particularly large and the Zen FX CPUs will target mainstream price points in which they will compete very aggressively with their upcoming Intel counterparts. This is a space that’s been traditionally occupied by Intel’s quadcore I7, i5 and dual core i3 CPUs.

Zen FX CPUs will debut with a brand new chipset on socket AM4 that’s planned to push AMD’s connectivity, storage and entire feature set forward and much closer to parity with Intel. The upcoming FX processors will also include enhanced platform security features built directly into the hardware.

Zen Based APUs To Bring PS4 & XBOX ONE Level Of Performance, Zen Coming To The Enterprise Space In 2017

Moving over to the APU and server side of the Zen equation. APUs and enterprise products based on the new Zen microarchitecture are planned for release sometime in 2017. We’ve also learned of a particularly exciting piece of information about AMD’s next generation APUs and that they will boast performance levels that are comparable with the Playstation 4 and XBOX ONE, even inside frugal notebook chips.

One other exciting tidbit is that AMD is working on a Zen based APU with on-package stacked high bandwidth memory that’s planned for release in 2017. However, whether all Zen based APUs & future iterations will feature HBM is not something that were able to confirm.

 

[looncraz]

-AMD
http://forums.anandtech.com/showpost.php?p=30346682&postcount=101

AMD's track record speaks for itself.

JF-AMD was speaking without permission and soon lost his job (not sure if the two are linked). AMD never officially stated to expect higher IPC from Bulldozer that I can recall.

For your link regarding the original quad core Opteron, AMD specifically stated for virtualization and other specific workloads.

 

[cbn]

AMD Zen FX CPU Product Release Set For Q4 2016 – APUs Coming In 2017 With Comparable Performance To The PS4 And XBOX ONE

http://wccftech.com/amd-zen-launch-q4-2016/

Zen Based APUs To Bring PS4 & XBOX ONE Level Of Performance, Zen Coming To The Enterprise Space In 2017

Moving over to the APU and server side of the Zen equation. APUs and enterprise products based on the new Zen microarchitecture are planned for release sometime in 2017. We’ve also learned of a particularly exciting piece of information about AMD’s next generation APUs and that they will boast performance levels that are comparable with the Playstation 4 and XBOX ONE, even inside frugal notebook chips.

One other exciting tidbit is that AMD is working on a Zen based APU with on-package stacked high bandwidth memory that’s planned for release in 2017. However, whether all Zen based APUs & future iterations will feature HBM is not something that were able to confirm.

GPU-wise the Xbox One has 768 sp and the PS4 has 1152.

 

[frozentundra123456]

What it says is that AMD plans to release it in Q4 2016. But it also says "no firm release date has been set". It does *not* say a release date has been set for 2016.

And lets not forget this is the same publication that said Skylake was going to be a Conroe like jump in performance.

 

[jpiniero]

PS4 and Xbox One level of performance?

I wonder what that means?

GPU-wise the Xbox One has 768 sp and the PS4 has 1152.

But does that mean the Zen APU will have that many stream processors? Or does it mean the CPU will be strong enough to compensate for a weaker iGPU?

Most likely the number of SP. But 768 or even 1024 is not much of a stretch given that they are shipping with 512 right now. You'd obviously need something to increase the bandwidth beyond just dual channel DDR4 2400. HBM would be the obvious answer, but that's going to be way too expensive for OEMs on a chip that's already going to be really expensive to make. That doesn't mean AMD won't try though.

 

[frozentundra123456]

I could see xbone levels, yes. PS4 seems like a lot bigger stretch, at least at a reasonable die size, cost, and power budget. It all depends on availability and cost of HBM like you said. The cpu should at least be a lot stronger than the consoles, so it might be more competitive in cpu demanding titles.

But by the same token, we will have 14/16 nm dgpus by then, so that will raise the bar considerably for APUs compared to discrete.