AMD Ryzen (Summit Ridge) Benchmarks Thread (use new thread)

[cytg111]

Arh .. those expenses dont evaporate into thin air just cause they dont own the fabs...

 

[unseenmorbidity]

Arh .. those expenses dont evaporate into thin air just cause they dont own the fabs...

No, but the cost is shared. I am not sure if that would be a net positive or not though.

 

[AtenRa]

Arh .. those expenses dont evaporate into thin air just cause they dont own the fabs...

No but i was talking about margins here not prices.

 

[StrangerGuy]

That is a horrendously bad idea... First off, that $1100 CPU is overpriced by over $500, because intel has a monopoly in that market! Instead of dropping their previous flagship core count down a tier in pricing, as they had done in the past, intel just made a new pricing tier, and keep the prices the same.

Secondly, AMD needs a CPU to compete with the mainstream i7, i5, and i3s. The HEDT market is extremely niche, and intel has near complete control of it. The kind of people that are dropping a grand on a CPU aren't going to buy second best! If Intel prices their CPUs to "compete" with the HEDT market, then they won't even take 5% of it away from Intel. Actually, at $900 they wouldn't even get 5%. They would be lucky if they sold maybe a dozen to some raving lunatic AMD fanboys.

They have to take marketshare away from them in the mainstream market by offering consumers more cores. It's a fair trade really, and the only way AMD can recapture marketshare from intel. More Cores vs Higher IPC Higher Frequency and IGPU.

It's painfully clear you have no idea what you are talking about. AMD always had a far worse margin problem than a marketshare problem. The latter without the former is worthless. It is critical for AMD that they have to raise ASPs as high as they possible could with Zen. The notion that AMD has to sell great products but at bargain basement prices is disguised entitlement syndrome.

 

[blublub]

Someone tried to measure IPC compared to Ivybridge:


https://www.youtube.com/watch?v=IN0BjzaP7lA

 

[zinfamous]

Selling ZEN CPUs at $200-600 range will increase AMDs margins by 2-3x or more because today they DONT sell anything above $120-150 (not what the price is but actual products sold)

I don't think Zen cores cost the same to make as their FX chips, no? I guess you get more efficiency per wafer, but it is still a more expensive process, isn't it? The way I see it, this may allow them to get a higher rate of better-binned chips per wafer, so maybe potential greater profits on selling more premium chips at premium prices.

But the sell price of their current chips isn't an indicator of what their margins would be on future chips, with a new process with a different expense.

I definitely think there will be $200-350 chips in the stable, possibly as low as $150, because that is in AMD's wheelhouse and as you said, targets the bulk of that i5 gaming marking. Something that AMD has long claimed as their goal. If they don't have i5 performance/price Zen chips, then they've been lying to us all along.

 

[blublub]

We will see but anything AMD sells with Ryzen will improve their balance sheet because market share is close to 0% !!! - And AMD will make sure they sell Ryzen CPUs below 200$ for the more budget market (maybe not from day 1) to gain MS in any area.

The base 8c 3.4Ghz will either be the leaked 399 or 499$ and anything which might be "+" as mentioned on the Horizon demo might be a bonus of very good chips for some kind of BE - may it either be 3.6 or 3.8 or even 4.0Ghz base clock.
I think those higher chips also won't have a TDP of 95w- If AMD can bring a Ryzen chip in at 3.8Ghz with a TDP of 140w it will be the fastest available 8c for the time being - hence no reason sticking to a slower 8c only for the sake of 95w.
If AMD can bring the fastest 8c on the market key will do it. Hell a 3.8Ghz or even 4.0Ghz chip could take it on with Skylake-X, so no way the won't go for it if they could do it....

 

[CatMerc]

I don't think Zen cores cost the same to make as their FX chips, no? I guess you get more efficiency per wafer, but it is still a more expensive process, isn't it? The way I see it, this may allow them to get a higher rate of better-binned chips per wafer, so maybe potential greater profits on selling more premium chips at premium prices.

But the sell price of their current chips isn't an indicator of what their margins would be on future chips, with a new process with a different expense.

I definitely think there will be $200-350 chips in the stable, possibly as low as $150, because that is in AMD's wheelhouse and as you said, targets the bulk of that i5 gaming marking. Something that AMD has long claimed as their goal. If they don't have i5 performance/price Zen chips, then they've been lying to us all along.

The 32nm process AMD uses for Vishera is incredibly cheap by this point, probably 3000$ a wafer. 14nm wafers probably cost them 6000$-7000$, and that's on the cheap side since it's GloFo.
Vishera die size is 315mm^2, while Zen's die size is supposed to be in the area of 220mm^2. Assuming a similar defect density of 0.01 per sq cm, we're looking at about:
18 dollars per Vishera chip
24 - 28 dollars per Summit Ridge chip

Margins should be quite healthy regardless.

And besides, even if they would sell at FX chip margins, the issue is the very very low volume. If they had Intel volume at FX margins, they would still be making very good money. Zen should improve AMD's situation on BOTH accounts.

 

[.vodka]

Nice find, thanks for posting. That Broadwell-E monster is usually running at around 3Ghz, so Ryzen at ~3.8Ghz should do the job I guess .
BTW January 3rd is in just over 2 weeks from now, not that far away.

Well, at least the CES part of the rumor is confirmed. There they are!

Now we wait, two long weeks ahead...

 

[zinfamous]

The 32nm process AMD uses for Vishera is incredibly cheap by this point, probably 3000$ a wafer. 14nm wafers probably cost them 6000$-7000$, and that's on the cheap side since it's GloFo.
Vishera die size is 315mm^2, while Zen's die size is supposed to be in the area of 220mm^2. Assuming a similar defect density of 0.01 per sq cm, we're looking at about:
18 dollars per Vishera chip
24 - 28 dollars per Summit Ridge chip

Margins should be quite healthy regardless.

And besides, even if they would sell at FX chip margins, the issue is the very very low volume. If they had Intel volume at FX margins, they would still be making very good money. Zen should improve AMD's situation on BOTH accounts.

ah, thanks. The other key point is to look back at their wafer arrangement revision from July~August or whenever that was. IIRC, AMD now has a far more favorable deal (still costs money, but better contract that before) to purchase wafers outside of GoFlo. Clearly they knew what they had back then and it looks like they have been planning all along to maximize foundry space wherever they could to get Zen out in volume.

by the way--that is just cost per wafer paid to GoFlo, right? I don't imagine that $18/chip or certainly not $24-28/chip reflects the costs of R&D, right? If itemized out, I wonder how much AMD has spent up until this point on each Zen chip--I assume that the costs of FX are (at least slightly!) beyond profit by now, right?

 

[blublub]

Yep that WSA makes much 2nd sourcing for 14nm LPP at Samsung more reasonable to expand total wafer availability

 

[Atari2600]

I don't think AMD offering twice the performance/price as Intel is good for them or good for us.

If AMD cannot make their money on Zen, how are they going to fund Zen+?

How are they going to fund whatever comes after the entire Zen line?


If, say, Zen offers 90% of the performance of Intel's best, then I wouldn't want to see it sell for less than 80% of the price of that. Its not healthy.

 

[realibrad]

I don't think AMD offering twice the performance/price as Intel is good for them or good for us.

If AMD cannot make their money on Zen, how are they going to fund Zen+?

How are they going to fund whatever comes after the entire Zen line?


If, say, Zen offers 90% of the performance of Intel's best, then I wouldn't want to see it sell for less than 80% of the price of that. Its not healthy.

AMD stock price will allow them to get more in loans. AMD has cut huge amounts of their debt, and are prime for investment. AMD could very easily offset a reduction in margins and use the investments to supplement things like R&D. AMD could likely sell Zen for $600-$700 at the top end and still make very good margins. If AMD can gain enough market share, then they can leverage that for further investment to supplement cash.

You are thinking in small business terms, and not in corporate terms. If AMD goes after the cash pop, then that means higher prices. Higher prices directly reduce consumption at the higher price point vs a lower price point. That limits market capture which allows Intel to come back if they so choose.

What is important for AMD is to gain market share and lock people into their system. If AMD can get a win in the server market and the next gen is equal or slightly behind Intel, that is a win. That is because the cost of moving platforms is expensive and is only worth while if the cost of moving is worth the benefit. If Intel comes out with its next gen and its only a few % faster than AMDs next gen, then many still upgrade to AMD because that means less hardware to buy. Big companies may have the capital to move, but many will not.

 

[KTE]

I was hoping for the 8c 4 base and 4.5 turbo, but realistically I think a bit more than the 8370E (95W) that clocks 3.3/4.3, so 3.5/4.5 for the first batches and 4/4.8 only if they deliver a 125/140W rig (that i hope will be produced at 95W with these clocks as soon as process matures), because the 8370 (125W) is 4/4.3 and i think that a bit more this is possible, considering that is a 32nm CPU and excavator is sayd more efficient than that (see the famous AMD slide with the +40% IPC and same energy/cycle than XV)...

Hello Planet Earth.

I've stopped replying to you, because:

1. You are on here forever and seem to have nothing else to do. I have a friggin long hours fulltime job, family, research, GTR etc. Sooo little time. I could never compete.

2. Entertaining you is too far on the nonsense scale. Serious rigmarole. Waste of too much precious time.. Should be getting AWS certified instead.

3. When I have spent genuine time replying and pointing to professional literature debunking your ideals, you've not understood even the simplest concept in it, dodged it, failed to reply to it and not even read 99% of it. You've stil carried on parroting your beliefs. That's not an honest intellectual, a scientist or an enthusiast. That's something else we all know.

5. An opinion is an opinion, is an opinion. A belief is just that, a belief. A fact though, is from a different planet and unchangeable by your prejudices or dislikes of it. You don't seem to get this.

6. From the more indepth talk, it was clear your hardware/electrical knowledge is lacking. We all have weaknesses but why make it up stuff you've no idea about? You seem a coder dabbling in other fields.

7. Your reasoning is Abwx based. It's pretty much the same rhetoric, and both jump in to support one another all throughout the threads without questioning each others silly assertions. What does that tell you?

Just look at one of your theories above:

Intel Broadwell 14nm 6950X 10C 140W = 3.0/3.5GHz
Intel Broadwell 14nm 6900K 8C 140W = 3.2/3.7GHz

Let's change the brand to AMD just to help your processing. AMD Broadwell.

2C down and they only gain 200MHz on base and Turbo??

Only at max 3.2base??

3.2GHz base is lower than their P4 130nm 2003?!?

Let's look further...
Intel Broadwell 14nm 6850K 6C 140W = 3.6/3.8GHz

2C less but STILL only 3.8GHz Turbo and 3.6GHz base??


And that is with a MUCH better process than AMDs!
No you didn't claim, you defiantly pushed everything else as absurd except your own belief of 4-4.5GHz base mininum at launch. This is what I clarified from you right at the beginning of your assertions.

Then you'll bring in how the 14nm process is 5000x lower leakage, 5000x faster, 5000x lower capacitance than Intel 130nm at the same FO4, meaning:

5000x better characteristics -> 82W -> 0.0164W per core
95W... 95/0.0164= 5792 Cores

That we will definitely, easily have 5792 Broadwell cores at 16000GHz and 95W AND 500x IPC of Northwood.


So here you arrive from Mars, pontifying BETTER process performance, better core performance, better SMT performance, better power efficiency, better power draw and 1GHz better clocks than Intels latest and greatest...



This is completely detached from reality.

Reminds me of the old days of THG/AMDZone fanboys.

Official: 19 stages integer pipeline.
I was saying this from weeks, but here all are mocking me or denying the obvious.
19 stages are so many that zen can be called speed daemon design...

Official where?

And FP pipe?

Why do you ignore every factor except just the one that suits your belief?

Speed demon means piss poor IPC. Don't forget that nugget.

I am done with this. GLHF

========================

BTW, onliners are doubting Zen IPC, which is 1T performance. Chief of them is Intels mouthpiece:

Altho Trump like sensationlized, what he says is possible. MT perf might be far better in scaling if ST is not as good. Specifically for branchy and cache sensitive code.

But I still think his 'bad ST' refers to 'not the same as SkyLake', which we already know. Non-competitive, sure, but anything over 15% IPC better than XV cannot be bad, in any respect.

Sent from HTC 10
(Opinions are own)

 

[KTE]

I was only trying to illustrate that AMD doesnt need to sell at the same price/margins as Intel and have a huge profit.

Agreed mate. They just need better prices/margins than in the last 11 years to generate more revenue and help start making a solid profit.

Sent from HTC 10
(Opinions are own)

 

[zinfamous]

Hello Planet Earth.

[holy crap stuff] deleted

For someone who has "no time to deal with this," not only is this post tiresome, it provides zero value to the thread while taking up an inordinate amount of space in doing so.

please stop.

 

[Fanatical Meat]

Agreed mate. They just need better prices/margins than in the last 11 years to generate more revenue and help start making a solid profit.

Sent from HTC 10
(Opinions are own)

They'll most likely under price it just to attract people like me who have felt burned by AMD's chips lately.
I'm still concerned about their partners motherboard quality, hopefully Ryzen has some serious quality behind it.

 

[BeepBeep2]

For someone who has "no time to deal with this," not only is this post tiresome, it provides zero value to the thread while taking up an inordinate amount of space in doing so.

please stop.

All he did was make counterpoints to the argument others were making. I agree that he wrote a book but some of his points were legitimate.

Personally, I think Zen will be a great CPU for AMD. I just don't think it's going to launch with mid 4 GHz turbo like some are suggesting. Look at the 65nm Phenom launch vs K8...now, I don't think it won't be even 5% as bad, I think Zen will be great and put up a fight with Bdw-E. But I think some people have their expectations unrealistically high.

I love AMD, and I hold a huge amount of stock in AMD, but I think those claiming Zen will have an outrageously high turbo speed are claiming that without much logical basis.

As far as Francois, he was under the impression that Bdw-E might be in single channel during the demos, so yeah I would suppose at least he is slightly surprised by AMD's Ryzen performance in multi-thread too.

The Stilt compiled a new version of Blender (newer than 2.78a binaries) too that takes full advantage of AVX256 and shows 30+% performance gain on intel depending on the architecture so it will be interesting to see what Ryzen would do on that build.

 

[Atari2600]

AMD stock price will allow them to get more in loans.

Crazy.

They are just about to try to make their returns on a massive R&D investment and you are saying they should be looking to increase their gearing.

You are thinking in small business terms, and not in corporate terms. If AMD goes after the cash pop, then that means higher prices. Higher prices directly reduce consumption at the higher price point vs a lower price point. That limits market capture which allows Intel to come back if they so choose.

Do you really think I (and others) haven't considered the happy medium?
Why do you think I (and others) advocate AMD selling for less than Intel rather than going for the same price/perf?


Anyone buying an ultra HEDT will know fine well which is best, they're going to be techno-aware.

Business sales are a bit different, with small/med business likely to be more ignorant to AMD, and large business having specialists who will be well aware of Zen.

The average desktop user doesn't have a clue... and will buy whatever they're told or whatever they recognise anyway.


Dressing up a daft decision as "corporate" business rather than good business reflects the idiocy that often prevails in large multinationals.... who usually seem to endure despite the f**kwits at the top, not because of them.

What is important for AMD is to gain market share and lock people into their system.

Useful. But not the be-all and end-all.

If AMD can get a win in the server market and the next gen is equal or slightly behind Intel, that is a win.

Completely different market from consumer, and yes, agreed. If AM4 can last several processor generations AMD would be well served by adjusting their prices for Zen1 down a bit in order to build share and gain that back on Zen2/Zen3/etc.

 

[Atari2600]

Can anyone who is benching blender please do a number of runs seeing the scaling on any one machine for a set number of threads please?
Particularly with and without SMT.

It'll give us some very quick and extremely dirty idea of where Zen single thread may be.

 

[qookap]

some guys live in the dream?
I wont buy their product with high price. no way.
a company with fail product? A company is gone but the earth is still turning.

this is the reason:
1. for now all of benchmark, I still thinking Intel better(if drop the pirce should be nice)
2.BMW vs Ford, anyone agree if Ford have high price than BMW? Brand value different.
if any company just push garbage few years and asking customer pay for it. did you accept?

you guys really crazy fans, I was an AMD fans with lot of AMD CPU.
now...go hell...the damn price with mouthful of lies.
I just bought RX 480 with fully regret, bye bye Vega no more again.
Terrible Watt kill the polar bear and pls follow PCI-SIG.
you cant design something without criterion.

the only good is the driver, no more crap CCC no more blue screen, nothing more.
pascal driver really redactedthis year.








Threadcrapping, inflammatory posting and profanity are not allowed.


esquared
Anandtech Forum Director

 

[sirmo]

Can anyone who is benching blender please do a number of runs seeing the scaling on any one machine for a set number of threads please?
Particularly with and without SMT.

It'll give us some very quick and extremely dirty idea of where Zen single thread may be.

Blender benefits from SMT. Someone did a test on a Skylake and it was about 40%.

 

[USER8000]

All he did was make counterpoints to the argument others were making. I agree that he wrote a book but some of his points were legitimate.

About half his post,is the Intel principle engineer badmouthing a competing product. He seems spooked for some reason,to go on such a massive rant on Twitter. If you look at the scores for the FX8350 in Blender,Intel CPUs absolutely destroy it.

If the AMD chip is as bad as he thinks it is,he would have no need to actually rant like that,since the benchmarks at launch would tell the full story. You never saw a similar rant like that before Bulldozer was launched,right??

 

[bjt2]

I'll respond only on technical questions, avoiding to respond to personal attacks, that qualify yourself.

I'll only say that is you that have ignored fact that i posted (like FO4, BD and XV speeds and TDP, and finally denying progress bewteen 28nm and 14nm ff).

Intel Broadwell 14nm 6950X 10C 140W = 3.0/3.5GHz
Intel Broadwell 14nm 6900K 8C 140W = 3.2/3.7GHz

Let's change the brand to AMD just to help your processing. AMD Broadwell.

2C down and they only gain 200MHz on base and Turbo??

Only at max 3.2base??

3.2GHz base is lower than their P4 130nm 2003?!?

8C is only 20% less than 10C, and there is the NB and memory controller that is the same, so no surprises here.
The P4 comparison is absurd as it is a completely differend architecture with more stages and way less FO4.
I suspect that you don't even know what FO4 is and i will not explain you again, because already did that (maybe not to you, but maybe you read my posts only if i reply to you)
I can say then that INTEL sucks, because IBM on 65nm produced the POWER6+ that went at 6GHz...

Let's look further...
Intel Broadwell 14nm 6850K 6C 140W = 3.6/3.8GHz

2C less but STILL only 3.8GHz Turbo and 3.6GHz base??


And that is with a MUCH better process than AMDs!

Again here seems you don't know what FO4 is.

No you didn't claim, you defiantly pushed everything else as absurd except your own belief of 4-4.5GHz base mininum at launch. This is what I clarified from you right at the beginning of your assertions.

4-4.5 means 4 base and 4.5 turbo. I don't believe that Zen 8c can arrive at 4.5 base. Don't distort my claims to prove your point, please

Maybe the 4C, since the 7700K has 4.2 base and this last ES has base clock higher than the 6900k, so i can hope that at same core number this advantage can be retained.

Then you'll bring in how the 14nm process is 5000x lower leakage, 5000x faster, 5000x lower capacitance than Intel 130nm at the same FO4, meaning:

5000x better characteristics -> 82W -> 0.0164W per core
95W... 95/0.0164= 5792 Cores

That we will definitely, easily have 5792 Broadwell cores at 16000GHz and 95W AND 500x IPC of Northwood.

Where I said that? Again distorting my claims to prove your point. Readers can judge how your rethoric is flawed...

So here you arrive from Mars, pontifying BETTER process performance, better core performance, better SMT performance, better power efficiency, better power draw and 1GHz better clocks than Intels latest and greatest...



This is completely detached from reality.

No this is based on available data, that i put in this thread and that you skipped. I don't repeat myself anymore. WHo has read and understood my arguments, can judge also these your sentences...

Official where?

http://www.linleygroup.com/newsletters/newsletter_detail.php?num=5577
Google, first result, but you don't even bother to read, let alone verify my posts. Here also says that 14nm FF is 30% more efficient than 28nm.

And FP pipe?

Here we don't have data, except than BD/XV FPU had FMACs, and Zen not, so the FO4 could be even lower than BD...

Why do you ignore every factor except just the one that suits your belief?

Same i can say of you, apparently.

Speed demon means piss poor IPC. Don't forget that nugget.

This is a prejudice, and I already explained that, but maybe you missed the post.
Anyway we know for sure that the INTEL process is better, still a preproduction Zen still manages to clock higher than the 6900k. THis means that Zen is a speed daemon design, because can clock higher than INTEL on a worse process. But Zen has also the same performance in blender and better in Handbrake! So it is not a low IPC design, since Broadwel isn't.

CUT nasty images

Yes, ST IPC can be lower, since all points to an higher gain from SMT on Zen, so what's the problem? It can be compensated with slighly higher clocks as i don't think the difference is abysmal.

 

[realibrad]

Crazy.

They are just about to try to make their returns on a massive R&D investment and you are saying they should be looking to increase their gearing.

That is 100% what I am suggesting. AMD woke the sleeping dragon last time, and Intel threw massive amounts of money at R&D. Intel was able to do this because of the size of their company. That gap is even greater now. AMD needs to leverage this win for cash to do more R&D, not just for desktop, but other markets as well. If AMD tries to soak up cash now at the expense of not gaining market share, it only makes it easier for Intel to step back in. Intel at any time can flip the switch and likely retake the market. Intel is massive and can easily outspend AMD in R&D. So, even if Zen is a win, it wont last if AMD cannot continue to improve. What you are suggesting is that AMD take the cash over the life of Zen, when they could take debt and invest in the future. Selling chips its going to be much slower. Cash now allows for R&D to grow for the future. Debt is for the future.

Do you really think I (and others) haven't considered the happy medium?
Why do you think I (and others) advocate AMD selling for less than Intel rather than going for the same price/perf?

Not sure what your point is here.

Anyone buying an ultra HEDT will know fine well which is best, they're going to be techno-aware.

And you would be wrong.

Business sales are a bit different, with small/med business likely to be more ignorant to AMD, and large business having specialists who will be well aware of Zen.

And they will look at the price and go AMD in that case.

The average desktop user doesn't have a clue... and will buy whatever they're told or whatever they recognise anyway.

Yep.

Dressing up a daft decision as "corporate" business rather than good business reflects the idiocy that often prevails in large multinationals.... who usually seem to endure despite the f**kwits at the top, not because of them.

You are free to think that. This is basic finance here. Its why companies are willing to sell long term debt for cash now. It allows you to take the cash now, and do even more. That results in a net gain over the long run. Not sure what else to say.

Useful. But not the be-all and end-all.

Never said it was. In my opinion, market share is very important. AMD should be able to make large profits and still undercut Intel to gain market share.

Completely different market from consumer, and yes, agreed. If AM4 can last several processor generations AMD would be well served by adjusting their prices for Zen1 down a bit in order to build share and gain that back on Zen2/Zen3/etc.

Okay.