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

[AtenRa]

Zen+ was already in development in parallel and is likely to soon tapeout for a 2018 launch.

I have a feeling ZEN+ will come later with 7nm.

 

[Ajay]

I have a feeling ZEN+ will come later with 7nm.

That said, I think the absolute best case scenario is Zen coming in as slightly better than Sky/Kabylake (which would really mean moderately better if AMD had access to Intel's process tech). AMD had a truly massive technological advantage during the K8 era which might not ever be the case again. The situations aren't exactly comparable.

When is 7nm @ GFL? I've heard they already have EUV units arriving or were about to, my source wasn't clear. He was stoked about Polaris and Zen production (@ Malta, IIRC).
I would expect AMD to go with a 12-15 month 'tick' if Zen sells well* (gotta keep momentum running). Minor uArch improvements + in situ process improvements.


* I really, really want Ryzen to do well - hasn't been much excitement in the x86 CPU world like this for a long time - it feels so good. Even if CPU improvements eventually grind back to 5-10% per gen, it'll be more fun with AMD and Intel trading places every six months.

 

[AtenRa]

This, agree with it totally but just to add to that the wow effect will have to come from performance & not the cheap(er) price tag. If they can't sell enough SR7's initially, even with performance IMO between 5960x & 6900K, for 700~800$ then they'll most likely not be able to tide the Intel crowd over to their side with a 600$ price tag either.

$600 is not low price, only 1% of the TAM will ever consider to spend more than $300-400 for the CPU alone. 99% of Intels non Server CPUs sold the last 2-3 years have iGPUs.

The big bucks today are in the $200 to $350-400 range with Core i5 and mainstream Core i7 (Socket 1151) for the desktop segment.

We also have to take in to account the platform features, AM4 will not have the same features as 2011-v3. AM4 is targeting up to high-end mainstream not HEDT which have quad channel memory (and im sure not that much PCI-e lanes).

And 8C 16T zen die will be significantly smaller than 10C 20T Broadwell-E , making the $600 price having ridiculously high margins even for Intel standards (imagine Intel selling 4C8T + iGPU 120-140mm2 at $500).

AMD will launch the top 8C 16T SKU first in January, HEDT people will want one but it will be too expensive for the rest of the TAM. The rest of the people will opt for the 6C12T and 4C8T SKUs at way lower prices that will come later in the end of Q1 early Q2. So in the entire Q1 the demand will not be high enough with only $600 (if that is the price) CPU available but it will generate anticipation for the cheaper models that will follow the coming months.

Selling a 4C 8T ZEN with Core i7 4770K performance at $200-250 and 95W TDP, will certainly generate a Wowww effect in Q1 2017 when Intel will sell a unlocked dual core Core i3 7350K at $170.

 

[qookap]

damn, you killed a polar bear and blame AMD?

I mean, wat?

I bought many AMD things, CPU&GPU.
It's really hot and increase Celsius temperature above 3 degree in the summer.
I can't open this one with 39 degree here...

 

[AtenRa]

When is 7nm @ GFL? I've heard they already have EUV units arriving or were about to, my source wasn't clear. He was stoked about Polaris and Zen production (@ Malta, IIRC).
I would expect AMD to go with a 12-15 month 'tick' if Zen sells well* (gotta keep momentum running). Minor uArch improvements + in situ process improvements.


* I really, really want Ryzen to do well - hasn't been much excitement in the x86 CPU world like this for a long time - it feels so good. Even if CPU improvements eventually grind back to 5-10% per gen, it'll be more fun with AMD and Intel trading places every six months.

I will put it around 2019.

 

[DrMrLordX]

Ehh, I know Francois, his social media accounts are not to be taken so seriously. He always errs on the side of AMD sux, but he has buddies working across the line. He posted later that he wishes nothing but the best for those working hard at AMD and knows competition keeps things spicy.

Heh! Fair enough. Some of it was heading into low IQ territory, but as long as everyone playing along knows the drill . . .

 

[R0H1T]

$600 is not low price, only 1% of the TAM will ever consider to spend more than $300-400 for the CPU alone. 99% of Intels non Server CPUs sold the last 2-3 years have iGPUs.

The big bucks today are in the $200 to $350-400 range with Core i5 and mainstream Core i7 (Socket 1151) for the desktop segment.

We also have to take in to account the platform features, AM4 will not have the same features as 2011-v3. AM4 is targeting up to high-end mainstream not HEDT which have quad channel memory (and im sure not that much PCI-e lanes).

And 8C 16T zen die will be significantly smaller than 10C 20T Broadwell-E , making the $600 price having ridiculously high margins even for Intel standards (imagine Intel selling 4C8T + iGPU 120-140mm2 at $500).

AMD will launch the top 8C 16T SKU first in January, HEDT people will want one but it will be too expensive for the rest of the TAM. The rest of the people will opt for the 6C12T and 4C8T SKUs at way lower prices that will come later in the end of Q1 early Q2. So in the entire Q1 the demand will not be high enough with only $600 (if that is the price) CPU available but it will generate anticipation for the cheaper models that will follow the coming months.

Selling a 4C 8T ZEN with Core i7 4770K performance at $200-250 and 95W TDP, will certainly generate a Wowww effect in Q1 2017 when Intel will sell a unlocked dual core Core i3 7350K at $170.

It's not but the point is if people aren't willing to spend 700~800$ (let's say 700$ just for the sake of argument) for something that's fairly close to 6900K then what makes you think that 600$ will sway them the other way? Also I do believe that octa core is still pro/gamer territory, although hexa core is mainstream IMO just that Intel doesn't want to sell it as such & would rather waste the die space of an i7 x700K on IGP, so the people looking for high end should be willing to spend that little bit extra considering the cash grab that 6950x or even 6900K is atm.

Indeed a quad core without IGP for ~250$ is perfect at that price, a vast majority of gamers out there have an i7 x700K & it's really a waste of a die with IGP that's barely used, if at all.

 

[unseenmorbidity]

I have a feeling ZEN+ will come later with 7nm.

I seriously doubt it. They will almost certainly do at least another run at 14nm.

7nm exists, but from what I heard they have just begin testing it. It's likely still years away.

How cool would it be if AMD beat intel to 7nm though!?

 

[AtenRa]

It's not but the point is if people aren't willing to spend 700~800$ (let's say 700$ just for the sake of argument) for something that's fairly close to 6900K then what makes you think that 600$ will sway them the other way? Also I do believe that octa core is still pro/gamer territory, although hexa core is mainstream IMO just that Intel doesn't want to sell it as such & would rather waste the die space of an i7 x700K on IGP, so the people looking for high end should be willing to spend that little bit extra considering the cash grab that 6950x or even 6900K is atm.

Indeed a quad core without IGP for ~250$ is perfect at that price, a vast majority of gamers out there have an i7 x700K & it's really a waste of a die with IGP that's barely used, if at all.

Well half the price for almost the same performance I believe its a good reason to sway someone to buy the AMD CPU. But I will say again that AMD doesnt aim to sway those users (1% of the TAM) away from Intel, 8C 16T top SKU at $600 will only serve as a lure for the rest of the $200-400 segment.

 

[AtenRa]

I seriously doubt it. They will almost certainly do at least another run at 14nm.

7nm exists, but they have just begin testing it. It's likely still years away.

They will do the same they did with Piledriver or what Intel does with Kabylake. They will bring a new SKU with 5% higher IPC and perhaps 100-200MHz higher clocks in 2018 and then in 2019 ZEN+ will arrive at 7nm.

 

[bjt2]

I remember buying an athlon 64 x2 for a fairly decent price. The gouging was on the gold sample high bin parts, but if you go back in the history you'll find a64 x2 were still priced better than Intel at the time.

I think we might see a similar thing this time. A black edition Ryzen top bin part 3.6Ghz+ base will probably cost a pretty penny. But the regular 3.4Ghz base will probably be pretty decently priced. Also I do hope all Ryzen chips are unlocked like the FX chips were.

According to a leaked slide they are full unlocked, except, maybe the TDP. There is an overclock software that let you change the target TDP of the auto OC feature and that's it for the overclock. I imagine that non top models have TDP locked somehow (e.g. top bin unlimited TDP and lower bin with TDP locked to e.g. 125/140W).Maybe some other knob like temperature, voltage, current and clock limits, also limited in low bin parts.

 

[unseenmorbidity]

It's not but the point is if people aren't willing to spend 700~800$ (let's say 700$ just for the sake of argument) for something that's fairly close to 6900K then what makes you think that 600$ will sway them the other way? Also I do believe that octa core is still pro/gamer territory, although hexa core is mainstream IMO just that Intel doesn't want to sell it as such & would rather waste the die space of an i7 x700K on IGP, so the people looking for high end should be willing to spend that little bit extra considering the cash grab that 6950x or even 6900K is atm.

Indeed a quad core without IGP for ~250$ is perfect at that price, a vast majority of gamers out there have an i7 x700K & it's really a waste of a die with IGP that's barely used, if at all.

Because it's a $100 cheaper!?

But seriously, I think you are misunderstanding the point of expensive items.

"Introduce a new item that’s priced even higher. Providing a higher price point for comparison, sometimes called “anchoring,” makes the original item’s price seem lower. Anchoring works because the new, higher-priced item changes customers’ expectations and perception of value where the original product is concerned. A classic example is Williams-Sonoma’s difficulty selling a $275 bread maker, which it solved by offering a similar product priced at $429. (For the greatest success, do not support the highest-priced item with a lot of inventory or marketing effort, because it may languish on store shelves.)"

If you happen to sell a few of the top tier items that have asinine markups on them, that's great, but it's not where the real money is. The real money is in that middle tier. In this case, the $200-$400 range.

 

[AtenRa]

How cool would it be if AMD beat intel to 7nm though!?

GloFos 7nm will be in the same category as Intels 10nm.
Intel will launch 10nm SKUs (mobile) in H2 2017. I believe both will use 7nm (AMD) and 10nm(Intel) for Desktop SKUs in 2019 at some point.

 

[R0H1T]

Because it's a $100 cheaper!?

But seriously, I think you are misunderstanding the point of expensive items.

"Introduce a new item that’s priced even higher. Providing a higher price point for comparison, sometimes called “anchoring,” makes the original item’s price seem lower. Anchoring works because the new, higher-priced item changes customers’ expectations and perception of value where the original product is concerned. A classic example is Williams-Sonoma’s difficulty selling a $275 bread maker, which it solved by offering a similar product priced at $429. (For the greatest success, do not support the highest-priced item with a lot of inventory or marketing effort, because it may languish on store shelves.)"

If you happen to sell a few of the top tier items that have asinine markups on them, that's great, but it's not where the real money is. The real money is in that middle tier. In this case, the $200-$400 range.

I'd like to see how that works out for AMD, I foresee review sites & users panning them for such a move, if they were to introduce a more expensive SR7 after they've launched a 600$ octa core.

I get what both of you are saying but IMO the 700~800$ launch price just seems good enough, they can always bring it down to more reasonable levels if the sales aren't that great &/or Intel retaliates with their own round of price slashing. This also takes into account the assumption that SR7 will be within 5~15% of 6900K, after including the OC headroom.

 

[bjt2]

Is that possible? To get it to 3.9 or 4.0 with only 3 months before they hit their self imposed deadline? We know they reached 3.4 base. For those of you who understand the manufacturing part, how many months and respins does it normally take to improve performance that much? Admittedly we don't know what the boost clocks are today, we only know minimum base. Is it even reasonable to expect a CPU that can only reach 3.4 reliably on all cores, but be able to boost up to 3.9?

I posted a graph, many weeks ago, of power consumption vs frequency of a neon test chip. With LVT transistors, the power started to explode past 4.1-4.3 GHz. Until then the graph was almost quadratic, as one would expect when under the limits of the process. And that was a graph of two years ago. If the power density is not too high, turbo core max on one core allows to put about 5-6x the power than at base clock on a single core. I don't think that the clock increment is only a few hudred MHz...

 

[unseenmorbidity]

GloFos 7nm will be in the same category as Intels 10nm.
Intel will launch 10nm SKUs (mobile) in H2 2017. I believe both will use 7nm (AMD) and 10nm(Intel) for Desktop SKUs in 2019 at some point.

Yah, I forgot about that... It's so annoying that the same terms have different meanings.

 

[AtenRa]

I'd like to see how that works out for AMD, I foresee review sites & users panning them for such a move, if they were to introduce a more expensive SR7 after they've launched a 600$ octa core.

I get what both of you are saying but IMO the 700~800$ launch price just seems good enough, they can always bring it down to more reasonable levels if the sales aren't that great &/or Intel retaliates with their own round of price slashing. This also takes into account the assumption that SR7 will be within 5~15% of 6900K, after including the OC headroom.

People should understand that AMD with ZEN doesnt aim the HEDT segment but up to high-end mainstream ($300-400 max), the 8C 16T SKU at $600 will be the Intel 6950X or NVIDIAs TITAN XP equivalent. They will not introduce a more expensive SKU later because 8C 16T ZEN die/performance doesnt allow them. Even at $600 its very expensive for 99% of the TAM, and with less than 10% of the Desktop market share , they better aim to sell millions of $200-300 CPUs than a few thousand $600 SKUs.

IF Intel lower prices then they can still sell a top bin 8C 16T SKU in 2018 at $300-400 when Intel will launch a mainstream 6-Core SKU.

 

[Nothingness]

I've never used the program before, so I mighta gotten lost somewhere. Step by step instructions would help; make sure I provided the right info.

I'm afraid I'm not able to help, I'm not familiar with that tool

 

[R0H1T]

People should understand that AMD with ZEN doesnt aim the HEDT segment but up to high-end mainstream ($300-400 max), the 8C 16T SKU at $600 will be the Intel 6950X or NVIDIAs TITAN XP equivalent. They will not introduce a more expensive SKU later because 8C 16T ZEN die/performance doesnt allow them. Even at $600 its very expensive for 99% of the TAM, and with less than 10% of the Desktop market share , they better aim to sell millions of $200-300 CPUs than a few thousand $600 SKUs.

This is where our opinions diverge, I do believe that with an octa core ZEN AMD is targeting the HEDT line, for me personally the high end mainstream stops at six core chips. Should AMD introduce a ten core successor to the SR7, it'll just as well be aimed at the enthusiast segment & priced accordingly.

Yes but as I've said around 600$ is where IMO the enthusiast segment starts & I see them happily paying 1000$ for Titan XYZ all the time, should they find the SR7 to their liking a 100$ more (even if it's just the launch price) ought not to be a deal breaker for them.

That they will & so the (IGP less) quad core ZEN should be priced competitively, the APU however could command a premium since AMD is clearly superior in the graphics department.

 

[krumme]

If it cost more than $600 I'd just wait for Skylake-E as I originally planned. I know Skylake would be at least a little faster and if I'm spending the big money on a CPU then I'm getting the absolute best and I'll wait for it.

Not only will skle probably be a bit faster. It will also be a bit more futureproof due to quad channel mem and 256avx code. I prefer not to buy a gpu that can not do asynch properly and is not optimal for dx12 vulcan. I only do that if i get a discount for the perf/ $ i get. The same principle applies to cpu.
Right now i can do without 256 fpu and quad mem because it comes at a gigantic cost for a total solution. But if amd tries to sell this at 600 usd its 100% a no go for my part.

 

[krumme]

I can not understand FO4 and can not find places that is sufficient simple to understand. Please help and gladly in relation to zen uarch and gf 14lpp.?

 

[bjt2]

I can not understand FO4 and can not find places that is sufficient simple to understand. Please help and gladly in relation to zen uarch and gf 14lpp.?

FO4 metric is a measure of how slow (or fast) a circuit is.

Absolute FO4 metric is the delay of the reference test cirucit on a given process.

All is normalized to the reference circuit. So changing the process, you may only measure the delay of the test circuit and know moreless the delay of all other circuits.

The test circuit is an inverter that feed other 4 identical inverter. Then the delay of the propagation of a switch is measured.

The FO4 numbers that we see around are RELATIVE FO4. If I say FO4 17, i am saying that that circuit has a delay 17 times the reference circuit.

Now let's talk of FO4 applied to pipelined processor.

Each stage implements different logic and of each stage a total FO4 (input of the stage/output of the stage) can be calculated.

A good design requires that all stages have the same or similar FO4, because the max clock is limited by the stage with the highest FO4, because when it ceases to function, all the CPU does not work.

The relation is roughly that the max clock (excluding TDP limits) achievable is inversely proportional to FO4.

How we can estimate FO4? We can only do this relatively. But this is enough.

If i have an x86 design that have X stages, another x86 design with Y stages has roughly X/Y the FO4 of the first, because both are x86 CPUs and do the same work. This is especially true with similar architectures.

INTEL has an advantage on the process, but since Zen managed to equal or slightly beat the clock, we can conclude that has lower FO4.

More stages means more penality for branch mispredition. But as can be seen by a table posted some page ago, on blender the branches are 7% of the instructions and of these only 5% are mispredicted.

So even if we double the stages (and so we can increase the clock), the loss of IPC is minimal, because we double the time only on 5% of 7% of the instructions...

The only problem is that this increase the power consumption, so we must do a balance.

And anyway we can't go too low because there is an overhead of 2-3FO4 to break the stages, so under 10FO4 is not advisable to go...

 

[Nothingness]

I've never used the program before, so I mighta gotten lost somewhere. Step by step instructions would help; make sure I provided the right info.

On Realworldtech:
http://www.realworldtech.com/forum/?threadid=163466&curpostid=163575

Blender branch prediction MPKI (misprediction / kilo instr) is not that high. So this benchmark likely does not tell us a lot about branch prediction performance of Ryzen.

 

[BeepBeep2]

You know what I just remembered? Bdw-E has a coldbug, overclocks at best mid 4 GHz on air / water, and 5-5.2 GHz on LN2. Haswell-E does well over 6 GHz on LN2.
As an extreme overclocker, it will be really interesting to see where Zen falls there for me.

Also:
https://edgeup.asus.com/2016/06/17/broadwell-e-overclocking-guide/6/

^^^This is exactly what I was worried about when I was explaining my theory on Zen's turbo clocks that got me huge backlash a few pages back, and this is also the reason you don't see Bdw-E with high turbo clocks.

If Zen gets 4 GHz+ Turbo and performs nearly as well across the board as AMD has shown so far, it'll be a smash hit. If not, and it gets between 3.7-4.0 GHz turbo like I had guessed, it will still be impressive imho.

I know a lot of people here are more concerned about Skylake / Kaby Lake and single thread performance, but even if the Ryzen CPUs can not compete there, this will be a very good processor for AMD in the HEDT space, which they are targeting. (I have no idea why people are claiming that they aren't targeting the HEDT space, it has been stated multiple times that they are.)

/drunkpost

 

[Insert_Nickname]

Folks that have been sitting out of the market for awhile are wildcards. You would think that anyone looking to pay $800+ for a CPU would have picked up Haswell-E or Broadwell-E by now. Remaining buyers are probably on the fence between a 5820k and 6700k and don't necessarily 100% like all of what they see, but don't want to step up and pay the big bucks for the stronger LGA-2011 v3 chips.

They might pay a little more for Summit Ridge than they would a 6700k, but not THAT much more.

I obviously can't speak for anybody else, but my own beef with moving up to the HEDT isn't so much CPU cost, but the cost of 2011v3 mainboards. That and the fact you need 4 DIMMs for optimal performance (can get away with just two). You can get a perfectly serviceable 1151 board for half of what a decent 2011v3 board will set you back.

Combining an 8C/16T CPU with a mainstream platform might work out very well for AMD.

Was planning to go Skylake-E eventually, but from what I've seen so far I am beginning to revise that notion.