AMD Further Unveils Zen Processor Details

H

HardOCP News

[H] News
Joined
Dec 31, 1969
Messages
0
The crew at HotHardware attended an AMD press event that showed, among other things, a Summit Ridge processor pit against an Intel Extreme Edition Broadwell-E CPU (both clocked at 3GHz) running Blender and playing Deus Ex: Mankind Divided. Hit the link for the complete rundown. You can see the demonstrations in the video below.
 
I posted about this earlier today.

Toms stated as far as they could tell, that unless they were to use a stopwatch, that there was no discernable differences.
I wonder if the guy from Toms had a doobie beforehand, or if the guy at HH is The Flash?
 
Be interesting to see how it shakes out. I haven't had an AMD processor since the XP 1900.
 
Its also open source. So who knows if they recompiled and optimized, or lack of same. And using a custom image. Not to mention all the other information they didn't give out.

Seems like a quick put together response to the AOTS benchmark leak.
 
Clock to clock speed doesn't matter when the standard clock speed of a 6900K is 3.2 (turboed to 3.7 under load) compared to a zen at 3.0.

Clock speed is fairly irrelevant when a zen will have trouble going higher than 3.0 ghz.
 
  • Like
Reactions: raz-0
like this
It seems Blender is very cache speed sensitive rather than actual performance. FX for example performs really badly in it for the same reason.

If we imagine a Zen 8C/16T is to match a 6900K/5960X in average. Then we can also guess the price AMD will take for it. If so and at 95W. It will be a 800-1000$ CPU.
 
Hypergreatthing said:
Clock to clock speed doesn't matter when the standard clock speed of a 6900K is 3.2 (turboed to 3.7 under load) compared to a zen at 3.0.

Clock speed is fairly irrelevant when a zen will have trouble going higher than 3.0 ghz.
Click to expand...

How so? So, you are saying a clock for clock comparison is irrelevant? How so? Also, do you have proof that the Zen 8c 16t processor will not go higher than 3.0 GHz? Interesting but, I guess if Zen had not edged out the 6900k in this example, you would not be so quick to judge then, eh? It absolutely astounds me how the justification occurs when Intel is possibly threatened. Were you also one of those folks who were saying you must down clock the FX processors to get a more realistic comparison?

Personally, I find this information encouraging since Zen will be my next upgrade one way or another. (Not till next year though.)

Shintai said:
It seems Blender is very cache speed sensitive rather than actual performance. FX for example performs really badly in it for the same reason.

If we imagine a Zen 8C/16T is to match a 6900K/5960X in average. Then we can also guess the price AMD will take for it. If so and at 95W. It will be a 800-1000$ CPU.
Click to expand...

No, it will not be 800 to 1000$ in my opinion. However, I could see $300 to $450 as long as the 4c/8t one does ok.
 
We also had benchmarks with Bulldozer beating Intel with 40% and Barcelona beating Intel with 50%. So I wouldn't put too much in it yet. Specially not in a benchmark that can easily be tampered and a custom image run on unknown hardware and settings. All you know is its a Zen CPU, its a 6900K and it runs some sort of Blender.

The AOTS benchmark leak had much more details to compare.
 
Shintai said:
We also had benchmarks with Bulldozer beating Intel with 40% and Barcelona beating Intel with 50%. So I wouldn't put too much in it yet. Specially not in a benchmark that can easily be tampered and a custom image run on unknown hardware and settings.
Click to expand...

Really? Sorry but, I did not see any such thing so, perhaps you just remember incorrectly?
 
ManofGod said:
How so? So, you are saying a clock for clock comparison is irrelevant? How so? Also, do you have proof that the Zen 8c 16t processor will not go higher than 3.0 GHz? Interesting but, I guess if Zen had not edged out the 6900k in this example, you would not be so quick to judge then, eh? It absolutely astounds me how the justification occurs when Intel is possibly threatened. Were you also one of those folks who were saying you must down clock the FX processors to get a more realistic comparison?
Click to expand...
Don't you think if they could get their engineering sample to run at 3.7 ghz they would have compared it directly to intel's offering that's base speed is 3.7?

It's not justification, i'm not an intel fan boy. I really do hope that zen beats intel at a good price and i'll be buying one. However, the leaked benchmarks aren't encouraging when it comes to performance.
 
ManofGod said:
Really? Sorry but, I did not see any such thing so, perhaps you just remember incorrectly?
Click to expand...

Like this?
AMD FX 8150 Looks Core i7-980X and Core i7 2600K in the Eye: AMD Benchmarks

Or this?
AMD posts blatantly deceptive benchmarks on Barcelona | ZDNet

Hypergreatthing said:
Don't you think if they could get their engineering sample to run at 3.7 ghz they would have compared it directly to intel's offering that's base speed is 3.7?

It's not justification, i'm not an intel fan boy. I really do hope that zen beats intel at a good price and i'll be buying one. However, the leaked benchmarks aren't encouraging when it comes to performance.
Click to expand...

True, if they could show any higher they would. It seems both the 4C and 8C models is stuck at the same clocks. 2.8Ghz base, 3.05Ghz all core turbo and 3.2Ghz peak turbo.

At this point in time, there isn't going to be any changes till release.
 
Shintai said:
Like this?
AMD FX 8150 Looks Core i7-980X and Core i7 2600K in the Eye: AMD Benchmarks

Or this?
AMD posts blatantly deceptive benchmarks on Barcelona | ZDNet



True, if they could show any higher they would. It seems both the 4C and 8C models is stuck at the same clocks. 2.8Ghz base, 3.05Ghz all core turbo and 3.2Ghz peak turbo.

At this point in time, there isn't going to be any changes till release.
Click to expand...


I dont think they would show the clock speeds "if they could".
1. They can't give us everything all at once. Bit by bit(lol) to keep the Zen talk alive.
2. Early silicon, early mobo, especially if they are still talking late Q4/16.
3. Why would they show the shipping clock speeds so far away from launch. Do you wanna give your competitor time to blow you away before you even launch? No, you don't. You announce shipping clock speeds close to launch/availability. You want to attempt to catch your competitor with their pants down, even for just a brief moment.

I don't think you understand how prototyping and marketing work.
 
Hypergreatthing said:
Don't you think if they could get their engineering sample to run at 3.7 ghz they would have compared it directly to intel's offering that's base speed is 3.7?

It's not justification, i'm not an intel fan boy. I really do hope that zen beats intel at a good price and i'll be buying one. However, the leaked benchmarks aren't encouraging when it comes to performance.
Click to expand...

I would not think that they would reveal their final clocks so early since we are still 6 months out from actually being able to purchase one, more or less. I would have to say that giving out your final clocks so early would be a bad move on AMD's part because it would give Intel an upper hand on what is coming. One for one is a good comparison in my opinion.
 
alxlwson said:
Why would they show the shipping clock speeds so far away from launch. Do you wanna give your competitor time to blow you away before you even launch? No, you don't. You announce shipping clock speeds close to launch/availability. You want to attempt to catch your competitor with their pants down, even for just a brief moment.

I don't think you understand how prototyping and marketing work.
Click to expand...

When samples are out to partners, mobo makers etc. Then there are no secrets anymore. Specially not when you are fabless. Its not prototypes anymore.
 
ManofGod said:
I would not think that they would reveal their final clocks so early since we are still 6 months out from actually being able to purchase one, more or less.
Click to expand...
Nonsense. If they are going to reveal anything at all, they will put the best possible face on it.

Ask yourself what the purpose of an advance announcement is. It's not to convince OEMs to design in your parts, you do that in NDA-covered private showings. Is it to prop up the stock price so execs can cash out their options? Is it to get consumers to wait for your new product, rather than upgrading to an Intel part now? If it is for either of those reasons, then you want to show the fastest parts you can without outright lying about it. Cherry-picked LN2-cooled samples running hand-tuned benchmarks - don't tell people that, just say "this much performance on this benchmark at this clock."

As we used to say back in my R&D days, paraphrasing Arthur C. Clark, "Any sufficiently advanced technology is indistinguishable from a rigged demo."

ManofGod said:
I would have to say that giving out your final clocks so early would be a bad move on AMD's part because it would give Intel an upper hand on what is coming.
Click to expand...
Don't be naive. Intel knows better than we do what's coming from AMD. Engineers talk, and they move from company to company. Plus Intel has competitive analysis teams whose entire job is predicting what the competition will roll out, and when.
 
ManofGod said:
I would not think that they would reveal their final clocks so early since we are still 6 months out from actually being able to purchase one, more or less. I would have to say that giving out your final clocks so early would be a bad move on AMD's part because it would give Intel an upper hand on what is coming. One for one is a good comparison in my opinion.
Click to expand...

Agreed, run 'em the same for now. You can't tell, it might overclock more than 50% and really compete with Intel when it's all said and done. I haven't seen full specs, but I think I heard dual channel DDR4 vs quad channel for the Intel, so the RAM might overclock nicely, too.
 
Shintai said:
When samples are out to partners, mobo makers etc. Then there are no secrets anymore. Specially not when you are fabless. Its not prototypes anymore.
Click to expand...

You don't think AMD has in-house fabs for when they're protyping new silicon?
 
ManofGod said:
I would not think that they would reveal their final clocks so early since we are still 6 months out from actually being able to purchase one, more or less. I would have to say that giving out your final clocks so early would be a bad move on AMD's part because it would give Intel an upper hand on what is coming. One for one is a good comparison in my opinion.
Click to expand...

That's where things get tricky. I hope that AMD learned their lesson with the hyperbole associated with the FX lineup prior to launch and that they are teasing information like you suggest, but there is always that sliver of doubt in the back of many people's minds because they have been burned before.
 
Steve said:
I think everyone is really hoping for the Zen to be stellar so that AMD once again gives Intel a run for their money and, in turn, spurs healthy competition and innovation.
Click to expand...

I for sure hope so. Nothing groundbreaking has happened from either company since AMD dropped the Operton bombshell on everyone at MillionManLAN 2 back in 2003. Sure, dual core and stuff, that wasn't nearly as exciting as the performance gains we got with the on-die memory controller.
 
  • Like
Reactions: N4CR
like this
Shintai said:
Like this?
AMD FX 8150 Looks Core i7-980X and Core i7 2600K in the Eye: AMD Benchmarks

Or this?
AMD posts blatantly deceptive benchmarks on Barcelona | ZDNet

True, if they could show any higher they would. It seems both the 4C and 8C models is stuck at the same clocks. 2.8Ghz base, 3.05Ghz all core turbo and 3.2Ghz peak turbo.

At this point in time, there isn't going to be any changes till release.
Click to expand...

There's a different between a marketing slide vs showing actual performance. Of course things should be taken with a grain of salt but it seems you are extrapolating more than what has been shown. I wasn't aware that they showed a 4C model? Which ES sample was that?
 
alxlwson said:
You don't think AMD has in-house fabs for when they're protyping new silicon?
Click to expand...

At this point in time, the chips must be validated and ready. And partners and mobo makers must have them. Hence we see things like the AOTS leak. If not so, then its going to be a late 2017 release at best.
 
"We don’t have actual clocks and power consumption details to share just yet" says HotHardware. Well, that's not encouraging. If Zen can't make it in the transactions-per-joule sensitive rack-mount server market, it's in trouble.
 
Tying a Broadwell-E chip at 3Ghz is great. It shows that they might be over-delivering on their IPC goals.

Of course the test may be highly selectively picked, including down to the point of selecting a version that gives them as much advantage as possible, as intel recently did vs Nvidia, but it still looks promising.

What is alarming to me is that it is a 3Ghz part. The rumor out there is that GloFo's 14mm FinFet process can't produce parts that go much beyond 3Ghz.

it's great to go head to head with Intel at 3Ghz, but we are going to need parts that can hit 4Ghz and beyond as well.
 
This will not be like the bulldozer debacle,,, i hope
Looking forward to ZEN/Vega ...need to build a new rig
 
kac77 said:
There's a different between a marketing slide vs showing actual performance. Of course things should be taken with a grain of salt but it seems you are extrapolating more than what has been shown. I wasn't aware that they showed a 4C model? Which ES sample was that?
Click to expand...

Did they show actual performance vs a slide? You can only tell 3 things. There is a 8/16T Zen, a 6900K and some version of Blender with a custom image. Did they recompile Blender, did they optimize it? Did they remove optimizations? What version was used? What hardware was used? And so on.

20160811_144627_980.jpg

AMD Zen处理器规格曝光：已知共有四种 - AMD,Zen处理器,CPU - IT之家
 
Last edited:
alxlwson said:
You don't think AMD has in-house fabs for when they're protyping new silicon?
Click to expand...
I don't think so. First, even a small fab line is damned expensive. Second, you prototype on the process you intend to go into production on. You're just wasting time otherwise.

Now, do I believe that AMD might hot-lot prototypes through GloFo with instructions to skew the processing towards faster parts at the expense of yield? They might. It depends on how desperate they are. After the overhyping of RX-480, and given Intels complete dominance of the high-margin CPU markets, they may be very desperate.
 
Shintai said:
Did they show actual performance vs a slide? You can only tell 3 things. There is a 8/16T Zen, a 6900K and some version of Blender with a custom image. Did they recompile Blender, did they optimize it? Did they remove optimizations? What version was used? What hardware was used? And so on.

20160811_144627_980.jpg
Click to expand...

Those questions go both ways. You don't know it is not. Like I said before you are extrapolating more than whats' shown. Did they say one way or another? Nope then you really can't prove the negative if you can't prove the positive.

So you see a listing and you assume that's what they are launching with?
 
alxlwson said:
I dont think they would show the clock speeds "if they could".
1. They can't give us everything all at once. Bit by bit(lol) to keep the Zen talk alive.
2. Early silicon, early mobo, especially if they are still talking late Q4/16.
3. Why would they show the shipping clock speeds so far away from launch. Do you wanna give your competitor time to blow you away before you even launch? No, you don't. You announce shipping clock speeds close to launch/availability. You want to attempt to catch your competitor with their pants down, even for just a brief moment.

I don't think you understand how prototyping and marketing work.
Click to expand...

Look man, I'm running some AMD harware at home now, I own and have owned both intel and amd. I'm a fan of competition as well because I'm a consumer and that competition always benefits me in both performance and price. But The intel cpu at 3.2/3.7Ghz isn't some future part to compete with in the future, it's a current part, and those numbers aren't likely to stay put as time moves forward. If they can't beat a box stock part NOW, they probably won't beat what is there THEN either.

AMD chose unknown and unknowable benchmarks, hobbled the competition, and THEN came out basically even with a part that is supposed to be competitive upon release in six months. If it makes the new deadline and all the other usual launch caveats.

This isn't showing off, this is damage control to tide them over until they announce pricing, because that's really what they compete on these days. Not being too unusably slow while being cheap.
 
Shintai said:
I know where he works. But is he wrong?
Click to expand...
Yes. Because it's subjective to a point, especially if you want to go down the road of compiling and compilation optimizations. For example you can compile handbrake for a 2600k with every optimzation and do the same for a an 8150 (this creates a custom code path BTW). In every bench I've run when I do this a 8150 will edge out a 2600K. Yet, does a 8150 have an overall IPC advantage? Nope.

This is why it's best to take these new benches with a grain of salt and wait. Not immediately say one way or the other if the product is great or bad. Personally the results tell me that they have working silicon and nearing release and that's it. I'm not going to say that performance will be this or that until everyone can test it.
 
What I would say the test shows is that AMD fixed their weak caches. Rest we have to wait for. :)
 
Hypergreatthing said:
Don't you think if they could get their engineering sample to run at 3.7 ghz they would have compared it directly to intel's offering that's base speed is 3.7?

It's not justification, i'm not an intel fan boy. I really do hope that zen beats intel at a good price and i'll be buying one. However, the leaked benchmarks aren't encouraging when it comes to performance.
Click to expand...
You do have a point, as AMD would have clocked it higher if they could, and they didn't. We know from Polaris that AMD's 14nm process doesn't promote clock speeds very much. The BroadWell-E CPUs don't clock well either, and turbo as high as 3.5Ghz. Otherwise 3.0Ghz is the rated base speed for a 5960X.

The good thing about this Zen unveil is that AMD is on the right track and clock for clock the Zen CPU is on par. Getting higher clock speeds is just a matter of time with tweaks and better manufacturing. More importantly is the price as a 5960X is $1000. If AMD were to sell a 3.0Ghz Zen for $300, I don't think anyone would complain about the clock speed.
 
You must log in or register to reply here.
Back
Top