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

[realibrad]

Not funny, really - not for the long term health of the company. Lolz all around would be getting 4.4 GHz OC from the second tier bin for $600. And I'm being serious.
You would be on the same page if you want a competitive Zen+, Zen++, etc. Now if Intel drops pricing eventually, then everything changes. I don't see Intel dropping prices significantly in the short run - these are high margin parts for Intel (and that's the only reason they exist).

AMD goes aggressive, gains huge market share. Software gets optimized for their hardware. People spread the word how AMD offers stuff just as good if not better for much cheaper. AMD gets cash for R&D. All those things would feed off of each other.

If you think that is crazy, then I would point out the console situation. AMD got some cash, software was optimized for their chips. That secured more investment. They are not making huge profits on consoles, but it's worth it.

If they can gain market share while making a profit, then long run they are better off than a pop in cash. Remember, they have a stock price that can bring in cash. Becoming relevant in the market is better in the long run.

Also, AMD has something Intel does not. AMD with an efficient CPU makes their APUs far more interesting for laptops such as the surface and Mac book. If AMD matches Intel on CPU then that is big.

 

[Thunder 57]

Imagine the chaos if this thing OC'd nicely to 4.6 and came in at under $600? LOLz abound.

Summit Ridge will split the difference as they try for higher margins than in the past while chasing after higher marketshare where it counts. The top-end Summit Ridge will go after Intel HEDT buyers with better-than-6900k pricing with better-than-6900k performance (er, we think). $500 makes sense here. Though we may see prices as high as $800-$900 at launch if we get NewEgg effect . . .

I am predicting some more expensive [golden sample] high bin parts at like 3.6Ghz base clock. But I think the base 3.4Ghz 8c/16t part will be highly competitive. As low as $399.

Bad idea all around. AMD needs the money. If it performs like an $1100 Intel CPU price it at $900. A slower locked 8C version could come in around $5-600.

I don't understand why people think AMD needs to sell these chips at such a bargain. Even if it performs better than higher priced part people may question why it is priced lower, and think it is inferior somehow. Price it appropriately!

 

[Ajay]

People here are thinking as consumers, I guess. Sure, I'd love a top binned Ryzen @ $400, but just covering COGS+Operations+R&D won't get AMD to 25% of the market. Big Ad money will do more than word of mouth. That said - AM4 better be a damn solid performer as well.

 

[CHADBOGA]

AMD goes aggressive, gains huge market share. Software gets optimized for their hardware.

How different is software "optimised for AMD", over software either "optimised" or "optimised for Intel"?

Often AMD gets their highest benchmark score on software that has been compiled with the Intel compiler.

AMD might get closer to Intel if something is recompiled with GCC but that will normally result in the score being lower overall.

 

[sirmo]

Bad idea all around. AMD needs the money. If it performs like an $1100 Intel CPU price it at $900. A slower locked 8C version could come in around $5-600.

I don't understand why people think AMD needs to sell these chips at such a bargain. Even if it performs better than higher priced part people may question why it is priced lower, and think it is inferior somehow. Price it appropriately!

I don't disagree, but I think AMD is going to be disruptive. The margins aren't an issue as Ryzen is roughly around the same size as Polaris 10, for which AMD charges less than $150 (when you delete all the GPU card components), which presumably has a 30% margin. Ryzen at $399 would still have like 600% better margins than the Polaris 10 chips, built on the same 14nm FF process.

Base model 8 core at $399 would shake up the whole market, while still having a hefty margin. It would basically cost as much as a 6800K and offer two extra cores. The 6 core Ryzen if there is a such a chip could be $299, and the quad a $199. Basically in every segment AMD would offer 2 extra cores. Higher binned 8c 16t part will still be expensive due to its exclusivity, but I really think this is how AMD is going to price the whole lineup.

I think people forget how long Intel has been unchecked in this space. The dies have shrunk but the prices stayed the same.

 

[Thunder 57]

I don't disagree, but I think AMD is going to be disruptive. The margins aren't an issue as Ryzen is roughly around the same size as Polaris 10, for which AMD charges less than $150 (when you delete all the GPU card components), which presumably has a 30% margin. Ryzen at $399 would still have like 600% better margins than the Polaris 10 chips, built on the same 14nm FF process.

Base model 8 core at $399 would shake up the whole market, while still having a hefty margin. It would basically cost as much as a 6800K and offer two extra cores. The 6 core Ryzen if there is a such a chip could be $299, and the quad a $199. Basically in every segment AMD would offer 2 extra cores. Higher binned 8c 16t part will still be expensive due to its exclusivity, but I really think this is how AMD is going to price the whole lineup.

I think people forget how long Intel has been unchecked in this space. The dies have shrunk but the prices stayed the same.

Absolutely dies have shrunk under Intel. Let's not forget they ditched including a HSF for the k models as well. They have been making huge profits on these parts. AMD needs to do the same to keep up. An 8 core at $400 is just bad business I think. Remember AMD has had horrible margins for years now on some of their CPUS's. They also have a lot of debt coming due as well (http://seekingalpha.com/article/3221666-amd-bonds-how-much-time-is-really-left-on-the-clock). They need high margins more than Intel.

If AMD had no debt, sure, they could make a lot of money at $399. They need money to cover their bills as well as future R&D. They have been bleeding money for too long. Of course, maybe I'm completely wrong. I think they would be selling themselves too short by having a $399 or even $499 price ceiling.

 

[DrMrLordX]

I thought Intel's problem with the 6900K is that they didn't want to tick off their server customers who are paying well above $1,100 for what exactly, I am unsure.

That may well be the case . . .

What is the difference between a 6900K and the server equivalent?

I was under the impression that it was mostly platform-related.

I don't understand why people think AMD needs to sell these chips at such a bargain. Even if it performs better than higher priced part people may question why it is priced lower, and think it is inferior somehow. Price it appropriately!

Intel has already had Broadwell-E on the market for awhile. Anyone with a 6900k or better (or even a 5960x, honestly) is probably not going to need the top-end Summit Ridge. AMD's main target will be people on the 6700k and 7700k, and their price threshold for a CPU is around $350 - look how many balked when the 6700k was $400 or more due to shortages.

Some (not all, but some) of those people will step up to a $500 CPU if they know they are getting twice the cores and some good-to-great IPC to go with it, especially if the clocks match. The remainder will be looking for the lower-end 8c/16t chip that launches in March or so, and they will look to stay in the $350 or less price range. AMD may also poach a few 5820k/5830k owners in this price range.

Right now I do not think there is a large market for a $900 CPU from AMD, since the people who are willing to buy something that expensive may want more PCIe lanes than can be provided by AM4/Summit Ridge and likely already have something nearly as good anyway.

If AMD tries to go over $500, they will get some pushback.

How different is software "optimised for AMD", over software either "optimised" or "optimised for Intel"?
.

In this case, I do not think that AMD chips will react to software about the same way as Intel chips except where AVX256/512 is involved. Sort of like the K7 days. AMD has ditched FMA4 and XOP, they have ditched CMT in favor of SMT, etc etc. There's a lot they've done in moving to Zen to make themselves more Intel-like.

They need high margins more than Intel.

They need sustainable cash flow. $500 for their flagship will put them way ahead of ~$225 for the 9590. And it will help them build brand loyalty for Zen+ and future iterations of AM4 (or AM5 or whatever).

I expect that if Summit Ridge is a success, we should expect higher prices on Zen+ whenever that comes out.

 

[zinfamous]

Absolutely dies have shrunk under Intel. Let's not forget they ditched including a HSF for the k models as well. They have been making huge profits on these parts. AMD needs to do the same to keep up. An 8 core at $400 is just bad business I think. Remember AMD has had horrible margins for years now on some of their CPUS's. They also have a lot of debt coming due as well (http://seekingalpha.com/article/3221666-amd-bonds-how-much-time-is-really-left-on-the-clock). They need high margins more than Intel.

eh, that is a really old article and in terms of their current debt load, is now irrelevant. AMD shaved half of that debt back in September with $1billion in new stock offerings, much of it back to Mubadala. This allowed them to save a further $85m/qtr in interest going forward, and the remaining 1 billion is now due, something like 2020.

They still aren't in the greatest shape, but considering where they were back in February, they are really in phenomenal shape--again, it's all relative, heh. You're right--I think that 8 core really needs to be performance/quality priced appropriately. Considering that AMD got there with roughly 1/20th??? the R&D budget of Intel--and this includes R&D in GPUs? wtf--they are at a remarkable advantage of winning back significant value in terms of their current near-0 market share.

Intel doesn't really want to get into a price war at this point. They probably have more to lose than AMD has to gain, so it behooves Intel to let AMD regain 5%+ of that market without much response. Intel made this bed with their current pricing, and now AMD has a lot of leverage in pricing their lineup. Kabylake looks to be a rather unexciting update to SL, but I'm not sure if AMD can really capitalize on that--Intel marketing just too strong. The majority of the market just needs to hear "New Intel = more better for reasons!" and they are sold, heh.

 

[zinfamous]

They need sustainable cash flow. $500 for their flagship will put them way ahead of ~$225 for the 9590. And it will help them build brand loyalty for Zen+ and future iterations of AM4 (or AM5 or whatever).

I expect that if Summit Ridge is a success, we should expect higher prices on Zen+ whenever that comes out.

Shit, if they price this top-binned 8c CPU at $500, I'll buy two, bronze the second one and mount in on my wall.

...and I don't think I've ever spent more than $250 on a CPU.

 

[jihe]

I thought Intel's problem with the 6900K is that they didn't want to tick off their server customers who are paying well above $1,100 for what exactly, I am unsure.

What is the difference between a 6900K and the server equivalent?

ECC memory support.

 

[HurleyBird]

I can see AMD having some ~$1K flagship that is very well binned, they've done so before, but I'd put money on the lowest end 8c16t Zen chip being priced at or under $450

 

[jihe]

I hope AMD goes the core unlock route again. Will instantly change all my gear to AMD.

 

[moonbogg]

If they price the chip too high, then this thing is pointless. They're biggest problem is that people have forgotten about them. They need new customers to put an AMD chip in their rig, and people need to be excited about it. If the chip costs $800, then 5 people will buy it and good for them, but that's all that will happen.
AMD used to charge 1k for their flagships, but that was when the AMD name meant high performance. Right now they need to get consumer confidence back. Its a risk buying a Zen CPU because of AMD's embarrassing history. They need to price it low enough so people actually buy the damn thing.

 

[DrMrLordX]

Shit, if they price this top-binned 8c CPU at $500, I'll buy two, bronze the second one and mount in on my wall.

...and I don't think I've ever spent more than $250 on a CPU.

Be sure to test them for max OC before bronzing! You wouldn't want to mount a golden chip on your wall, at least not until after burning it out from use.

If they price the chip too high, then this thing is pointless. They're biggest problem is that people have forgotten about them. They need new customers to put an AMD chip in their rig, and people need to be excited about it. If the chip costs $800, then 5 people will buy it and good for them, but that's all that will happen.
AMD used to charge 1k for their flagships, but that was when the AMD name meant high performance. Right now they need to get consumer confidence back. Its a risk buying a Zen CPU because of AMD's embarrassing history. They need to price it low enough so people actually buy the damn thing.

Your sentiments are basically correct. Even Intel can only sell so many CPUs at $1k or more.

And I must correct myself: AMD's current flagship is now down to $199 on NewEgg:

http://www.newegg.com/Product/Product.aspx?Item=N82E16819113347&nm_mc=AFC-C8Junction&cm_mmc=AFC-C8Junction-PCPartPicker, LLC-_-na-_-na-_-na&cm_sp=&AID=10446076&PID=3938566&SID=

Take a guess as to why? Regardless, AMD's name is mud in the mind of many right now, for many reasons. They need to rebuild their brand.

 

[Thunder 57]

...Intel has already had Broadwell-E on the market for awhile. Anyone with a 6900k or better (or even a 5960x, honestly) is probably not going to need the top-end Summit Ridge. AMD's main target will be people on the 6700k and 7700k, and their price threshold for a CPU is around $350 - look how many balked when the 6700k was $400 or more due to shortages...

I don't think that is necessarily the case. Someone here already mentioned that there are a lot of people still running a 2500k or 3570k (that would be me) that are looking for the next worthwhile upgrade (6 core, perhaps). Those with a 2600k or 3770k would be in the market for an even more expensive CPU, say the 8 core variant.

eh, that is a really old article and in terms of their current debt load, is now irrelevant. AMD shaved half of that debt back in September with $1billion in new stock offerings, much of it back to Mubadala. This allowed them to save a further $85m/qtr in interest going forward, and the remaining 1 billion is now due, something like 2020.

They still aren't in the greatest shape, but considering where they were back in February, they are really in phenomenal shape--again, it's all relative, heh. You're right--I think that 8 core really needs to be performance/quality priced appropriately. Considering that AMD got there with roughly 1/20th??? the R&D budget of Intel--and this includes R&D in GPUs? wtf--they are at a remarkable advantage of winning back significant value in terms of their current near-0 market share.

Intel doesn't really want to get into a price war at this point. They probably have more to lose than AMD has to gain, so it behooves Intel to let AMD regain 5%+ of that market without much response. Intel made this bed with their current pricing, and now AMD has a lot of leverage in pricing their lineup. Kabylake looks to be a rather unexciting update to SL, but I'm not sure if AMD can really capitalize on that--Intel marketing just too strong. The majority of the market just needs to hear "New Intel = more better for reasons!" and they are sold, heh.

Right, so if Intel is willing to give up some market share, why not charge a higher price for what look to be good CPU's?

Shit, if they price this top-binned 8c CPU at $500, I'll buy two, bronze the second one and mount in on my wall.

...and I don't think I've ever spent more than $250 on a CPU.

There are others out there that will spend more. The most expensive I bought was probably the Athlon 64 X2 3800+ for about $350. Others surely bought CPU's closer to $1000. Maybe those days are over, but a quality product can demand a premium price.

I can see AMD having some ~$1K flagship that is very well binned, they've done so before, but I'd put money on the lowest end 8c16t Zen chip being priced at or under $450

That would be fine. Let the early adopters pay a ton for it. Those who can wait can save a good amount by waiting a few months. AMD can always adjust prices. Better for them to start higher and drop if they aren't selling.

I hope AMD goes the core unlock route again. Will instantly change all my gear to AMD.

That's just bad for AMD though. They need money desperately.

If they price the chip too high, then this thing is pointless. They're biggest problem is that people have forgotten about them. They need new customers to put an AMD chip in their rig, and people need to be excited about it. If the chip costs $800, then 5 people will buy it and good for them, but that's all that will happen.
AMD used to charge 1k for their flagships, but that was when the AMD name meant high performance. Right now they need to get consumer confidence back. Its a risk buying a Zen CPU because of AMD's embarrassing history. They need to price it low enough so people actually buy the damn thing.

I disagree. There are plenty of people who have been around long enough to know AMD has been on top before.

 

[AtenRa]

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)

Also Core i5 is the CPU most used by gamers, attacking this segment ($200-300) will also increase AMDs margins by 2-3x over what they sell today.

They will use the 8C 16T Black Edition or something at $600 to lure people at the $200-300 with faster (more threads) and lower price SKUs than Intel.

Increasing their marketshare by 20% (currently at 0%) at the $200-300 price range will make them tons of profit.

So selling a 4C 8T at the $200 to $230 range is not low margins for them, its actually 2x the margins they currently have.

Selling a 160-200mm2 die at $600 is higher margins than what NVIDIA sells its TITAN XP (470mm2 at $1300). 160-200mm2 has higher yields and TITAN XP also has higher COG/BOM (PCB, Memory, Heat-Sink etc).

 

[coercitiv]

So selling a 4C 8T at the $200 to $230 range is not low margins for them, its actually 2x the margins they currently have.

In other words, there's a huge difference between margins for making ends meat and margins for the nice profits Intel makes, and AMD will feel very comfortable somewhere in the middle.

 

[bjt2]

Did anyone get the vibe that AMD made a big deal out of the Neural Network Smart Branch predictor? Do we know how long Ryzen's pipeline is? The way I see it, Lisa mentioned 25% of the improvement to the IPC uplift comes directly from the branch prediction improvement. It makes me think that they seriously invested in the branch prediction to mitigate a long pipeline. Which could mean that this thing might clock pretty well.

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...

 

[bjt2]

I understand why you are confident, but I don't necessarily agree with you. Sorry if you have already been asked, but what kind of base/turbo frequencies would you expect for 4/6/8C Zen, assuming 95W TDP?

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)...

EDIT: for the 6/4 cores. AFAIK a 65W 4c will be produced (no news on a 6C, nor a 95W 4C), and i hope 4/4.5 65W at least...

 

[coercitiv]

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

You weren't hoping, you were repeatedly predicting this on the forums. I've seen more FO4 on this thread than a priests mention Jesus in a ceremony. When some here (positively) talked about base clocks similar to what Intel is offering you were touting higher than that. Now people talk 4Ghz turbo and 4Ghz+ overclocks, you go straight for 4.5Ghz. By the time all this is done and product is revealed I can already foresee a switch towards "next Zen iteration will clock as predicted".

 

[bjt2]

You weren't hoping, you were repeatedly predicting this on the forums. I've seen more FO4 on this thread than a priests mention Jesus in a ceremony. When some here (positively) talked about base clocks similar to what Intel is offering you were touting higher than that. Now people talk 4Ghz turbo and 4Ghz+ overclocks, you go straight for 4.5Ghz. By the time all this is done and product is revealed I can already foresee a switch towards "next Zen iteration will clock as predicted".

Next batch, not next iteration. Anyway the 4.5Ghz max turbo i think that can be realized even on january. The graph without label is strangely hinting at 4GHz "Programmed" turbo, with peaks of 4.5Ghz...

 

[Gideon]

Selling a 160-200mm2 die at $600 is higher margins than what NVIDIA sells its TITAN XP (470mm2 at $1300). 160-200mm2 has higher yields and TITAN XP also has higher COG/BOM (PCB, Memory, Heat-Sink etc).

While I agree with the overall sentiment (AMD doesn't need anywhere near Intel-level crazy margins), this is also oversimplifying things. One has to keep in mind, that CPUs tend to be developed for much longer than GPUs (and they require more thorough validation). Titan XP was maybe in the works for 18 months, Zen was 4-5 years. You also wan't to capitalize on that investment.

 

[HurleyBird]

While I agree with the overall sentiment (AMD doesn't need anywhere near Intel-level crazy margins), this is also oversimplifying things. One has to keep in mind, that CPUs tend to be developed for much longer than GPUs (and they require more thorough validation). Titan XP was maybe in the works for 18 months, Zen was 4-5 years. You also wan't to capitalize on that investment.

Given current trends Titan XP will also go obsolete 4-5x faster, and the work on Zen will stay relevant longer, with new iterations being small evolutionary steps combined with better process technologies.

 

[unseenmorbidity]

Bad idea all around. AMD needs the money. If it performs like an $1100 Intel CPU price it at $900. A slower locked 8C version could come in around $5-600.

I don't understand why people think AMD needs to sell these chips at such a bargain. Even if it performs better than higher priced part people may question why it is priced lower, and think it is inferior somehow. Price it appropriately!

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.

They need sustainable cash flow. $500 for their flagship will put them way ahead of ~$225 for the 9590. And it will help them build brand loyalty for Zen+ and future iterations of AM4 (or AM5 or whatever).

I expect that if Summit Ridge is a success, we should expect higher prices on Zen+ whenever that comes out.

That is actually another good point. AM4 is a brand new socket, and every person they can get onto their side will be essentially locked in to AMD.

 

[AtenRa]

While I agree with the overall sentiment (AMD doesn't need anywhere near Intel-level crazy margins), this is also oversimplifying things. One has to keep in mind, that CPUs tend to be developed for much longer than GPUs (and they require more thorough validation). Titan XP was maybe in the works for 18 months, Zen was 4-5 years. You also wan't to capitalize on that investment.

I was only trying to illustrate that AMD doesnt need to sell at the same price/margins as Intel and have a huge profit. Even selling ZEN SKUs at $600 is higher margins than what NVIDIAs Titan XP has at $1300. And we all know how profitable NVIDIAs Q3 2016 was selling GPUs that have way lower volume per quarter than CPUs.

Also dont forget that AMD will use the same 8C 16T die in the server market at equal or higher prices than desktop SKUs.

Also, people should remember that Intel has to sell at higher margins because they have to fuel R&D for manufacturing as well. So when they sell a CPU they have to make enough profit to amortize the cost of R&D they spend on 14nm for example. AMD doesnt need to have the same margins because they dont spend for manufacturing R&D.