AMD Ryzen SKU and Price Information/Speculation.

[jpiniero]

"Guaranteed Frequency 3.7Ghz" is fishy. Where did they get this information from? Guaranteed under what workloads?

As I mentioned, I'm assuming they are guaranteeing that all cores can hit 3.7 Ghz. That's it, but it would full core turbo if the TDP allows it to.

 

[frozentundra123456]

Wcc is being a rubbish bin again.

A guaranteed frequency literally means base clock, as turbo clock cannot be maintained for a prolonged period of time.

I would interpret it as the opposite-- guaranteed frequency on max turbo, single core.

 

[Arachnotronic]

These terms are turning me off.

Unlimited boost frequency...

AMD Ryzen: To Infinity, and Beyond!

 

[Doom2pro]

AMD Ryzen: To Infinity, and Beyond!

Thanks for the buzzkill (pun intended).

 

[Agent-47]

I would interpret it as the opposite-- guaranteed frequency on max turbo, single core.

Highly unlikely. Guaranteed frequency is a term that WCC cooked up in its "infinite wisdom". it means nothing.

AMD listed the product as a 3.7GHz, which is either base frequency or all core turbo (the latter being more likely). that is the norm in the industry. If AMD tries to break the norm, they can expect another lawsuit similar to the one they have w.r.t. BD,

 

[Khato]

Then how are they gonna sell? You can bet your dollars that these prices br based on focus groups, how much stock they have to sell and at what margins.
How are you gonna tap someone with a quad sandy or later for a new system? You want that with double cores sir? Same price, yes sir. I can see that strategy. Market penetration and they've got inventory to sell.

The top SKU is never for volume. It's for the halo effect. If AMD had something competitive with the i7-6900k there's absolutely no way they wouldn't price it at parity or slightly above. Even if it was a heavily binned SKU that they couldn't provide any meaningful volume on. Why? Because people frequently make decisions on which brand to purchase based off the high end SKU, but then buy the version that's in their price range. There's more of a reason for Intel's absurd top-end pricing than just to discourage Xeon customers from going to HEDT chips.

 

[LTC8K6]

As I mentioned, I'm assuming they are guaranteeing that all cores can hit 3.7 Ghz. That's it, but it would full core turbo if the TDP allows it to.

All cores will hit 3.7Ghz with what cooler?

Some of these come with a cooler, and some don't, right?

So is this 3.7 guaranteed with the provided cooler?

And how does the guarantee work when the cooler has to be provided by the user? Do they base it on using AMD's cooler, even though it's not provided?

 

[Lovec1990]

LTC8K6 1700 will probably have stock cooler so its based on it if 1700 has XFR then 3.7ghz boost is limit becouse of stock cooler

 

[jpiniero]

All cores will hit 3.7Ghz with what cooler?

Presumably it'd depend on how lucky you get with the binning.

 

[Agent-47]

All cores will hit 3.7Ghz with what cooler?

Some of these come with a cooler, and some don't, right?

So is this 3.7 guaranteed with the provided cooler?

yes, wraith cooler
http://www.hardwareschotte.de/preisvergleich/Amd-Ryzen-7-1700-Wraith-p22037535

the link above now redirects to some other page, but the text on the link says wraith

 

[waldoh]

Will Ryzen have a 4c8t competitor for the 7700k? in terms of clock speed?

I don't need 6-10 core CPU's for my current needs, but I just want the highest clock speed possible on a 4c8t CPU.

 

[Magic Hate Ball]

yes, wraith cooler
http://www.hardwareschotte.de/preisvergleich/Amd-Ryzen-7-1700-Wraith-p22037535

the link above now redirects to some other page, but the text on the link says wraith

So if we're getting the Wraith cooler that's good for cooling 125w processors currently, there will be enough cooling capacity for a mild overclock with stock cooling.

 

[norseamd]

Pakistan, the center of the technology world

Funny, I had thought they were from India.

 

[LTC8K6]

LTC8K6 1700 will probably have stock cooler so its based on it if 1700 has XFR then 3.7ghz boost is limit becouse of stock cooler

Wraith is a good cooler, though. I would think 3.7 is low for a Wraith. Maybe the 65w TDP is the culprit?

 

[Dresdenboy]

Problem is, CPC just means that it can indeed be called a 65W sku, not that it will operate at that tdp in reality.

The full load power consumption of produced dies varies a lot at higher frequencies. So going a bit lower in freq might yield enough parts meeting 65W SKU specs. The price will modulate demand in the end.

 

[Doom2pro]

The full load power consumption of produced dies varies a lot at higher frequencies. So going a bit lower in freq might yield enough parts meeting 65W SKU specs. The price will modulate demand in the end.

Yeah some people don't realize the TDPs are granular, an dies will fall anywhere between the two but they eventually get lumped in either one.

Slightly lowering the frequency of a die that is 74-80W could get it into the 65W bracket.

 

[PPB]

I don't think the context of my response was clear: If indeed 500mhz of difference at the 3 to 4ghz range makes a wild difference in power consumption (from 65w to 115w, say), this is either great news for the non-ocers/undervolters and the cheapos like me that will surely end up with the 1700, bad news to the overclockers and bad news for AMD into the futture since the design will suffer the Broadwell-C syndrom.

 

[cytg111]

Problem is, CPC just means that it can indeed be called a 65W sku, not that it will operate at that tdp in reality.

This is really expected, if 1700 real tdp was indeed 65w, we would be looking at an engineering marvel and a godly undervolting architecture, if such a low clock speed difference could indeeed imply almost 2x tdp difference (1700 at real 65w, 1800x at a little bit over 95w just like CPC tested their 3.6/4 sample).

If that would hold true. Then I would not fear for Intel's desktop and HEDT revenue, I would really start looking at future mobile market share because an architecture that loves to scale downwards in tdp is indeed a wonderful product for the sub 35w tdp enviorment.

PS: This would mean we are looking at another broadwell C type of product and anything above 4.5 would be really difficult, either by a vcore limit or by a power consumption-vrm amperage one.

That would fall in line with "4.4 for suicide runs" for early ES leaks. Intel speedsteps its desktop procs to 800MHz(mine do, haswell), do we know any characteristics of zen here?

 

[frozentundra123456]

Highly unlikely. Guaranteed frequency is a term that WCC cooked up in its "infinite wisdom". it means nothing.

AMD listed the product as a 3.7GHz, which is either base frequency or all core turbo (the latter being more likely). that is the norm in the industry. If AMD tries to break the norm, they can expect another lawsuit similar to the one they have w.r.t. BD,

You mean like they advertised the 9590 as the first "5 ghz processor"? I recall when Llano first came out and the clockspeeds were really low they advertised the clockspeed by the turbo frequency as well, and I dont think it was all core turbo. Technically, if one core reaches 3.7 ghz for even a short period of time for every processor, that is a "guaranteed" clockspeed. The statement does not say "for all cores" or for any specified period of time.

 

[Glo.]

Problem is, CPC just means that it can indeed be called a 65W sku, not that it will operate at that tdp in reality.

This is really expected, if 1700 real tdp was indeed 65w, we would be looking at an engineering marvel and a godly undervolting architecture, if such a low clock speed difference could indeeed imply almost 2x tdp difference (1700 at real 65w, 1800x at a little bit over 95w just like CPC tested their 3.6/4 sample).

If that would hold true. Then I would not fear for Intel's desktop and HEDT revenue, I would really start looking at future mobile market share because an architecture that loves to scale downwards in tdp is indeed a wonderful product for the sub 35w tdp enviorment.

PS: This would mean we are looking at another broadwell C type of product and anything above 4.5 would be really difficult, either by a vcore limit or by a power consumption-vrm amperage one.

Why do you have to stick with 95W TDP? Broadwell-E CPUs are 140W, but they draw around 110-115W under load, without OC.

95W is then maximum power target for the CPUs.

 

[Agent-47]

You mean like they advertised the 9590 as the first "5 ghz processor"? I recall when Llano first came out and the clockspeeds were really low they advertised the clockspeed by the turbo frequency as well, and I dont think it was all core turbo.

from what i recall the first lLano did not have turbo. the ones that had mentioned base clock as default.:

no turbo, AMD A8-3870K@3Ghz: https://www.newegg.com/Product/Product.aspx?Item=N82E16819106001
with turbo AMD A8-3820@2.5Ghz (as listed): https://www.newegg.com/Product/Product.aspx?Item=N82E16883220163

Technically, if one core reaches 3.7 ghz for even a short period of time for every processor, that is a "guaranteed" clockspeed. The statement does not say "for all cores" or for any specified period of time.

"technically" true, but at the same time it is not the same thing as guaranteed. like i said the statement you are quoting is from WCC. not AMD

 

[Mopetar]

Will Ryzen have a 4c8t competitor for the 7700k? in terms of clock speed?

I would imagine that there will be some that can hit at least 4 GHz since there's an 8c16t part that's supposed to boost that high. No idea if they'll clock quite as high as Intel's chip, but there have been some posts earlier suggesting Ryzen runs at .9V for the stock settings, which if true would suggest that it has sufficient room to grow above that and if the 4-core chips are salvage parts they should have plenty of headroom given they were designed to run with twice as many cores. I suspect that Intel's chips will probably be able to reach higher overclocks when all is said and done, but if AMD is selling their chips for $250 instead of $350, it's probably not a deal breaker if you can get a few hundred MHz more from an i7.

 

[Dresdenboy]

That would fall in line with "4.4 for suicide runs" for early ES leaks. Intel speedsteps its desktop procs to 800MHz(mine do, haswell), do we know any characteristics of zen here?

IIRC the leaks stated 550MHz here.

 

[Dresdenboy]

I don't think the context of my response was clear: If indeed 500mhz of difference at the 3 to 4ghz range makes a wild difference in power consumption (from 65w to 115w, say), this is either great news for the non-ocers/undervolters and the cheapos like me that will surely end up with the 1700, bad news to the overclockers and bad news for AMD into the futture since the design will suffer the Broadwell-C syndrom.

The power/performance distribution of dies looks similar to this plot, just with f on x-axis (higher on the left) and power on y-axis.

This (amongst other factors, i.e. yields of bigger dies) leads to a bending like this:

 

[otinane]