AMD Ryzen Gen 2 Set For Q2 2018

Page 20 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
I think it's real.

Which is more likely:

1. In a fake this elaborate and time consuming to produce, the faker somehow missed that the embargo date was last year while he produced all of these slides?​

2. Some corporate drone made a typo?​

The faker seems more likely to pay attention to and catch such an obvious issue than the corporate drone is.
 
  • Like
Reactions: Jan Olšan

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
Nevermind, it's a fake. Almost certainty. Raven Ridge has 6MB cache L2 + L3, not 10MB. A typo is one thing, but getting technical specs totally wrong is another.

Combined with the date, the 105W TDP, lack of 2800X, and benchmarks showing the 8700K beating AMD in 1080p gaming (why not just show 4K?), it's all just a bit too much.

If it isn't a fake, someone at AMD should get fired or at least disciplined over how poorly these slides were put together.
 
  • Like
Reactions: CatMerc and raghu78

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Nevermind, it's a fake. Almost certainty. Raven Ridge has 6MB cache L2 + L3, not 10MB. A typo is one thing, but getting technical specs totally wrong is another.

Combined with the date, the 105W TDP, lack of 2800X, and benchmarks showing the 8700K beating AMD in 1080p gaming (why not just show 4K?), it's all just a bit too much.

If it isn't a fake, someone at AMD should get fired or at least disciplined over how poorly these slides were put together.

These are leaked slides, so they could be from before they were reviewed and had the typos fixed.

Having both a 1700X and 1800X never made much sense. Why two X models at 8 cores when you have an unlocked CPU anyway? The lineup in the slides are much more rationally segmented. If that is the work of a Faker, AMD should hire him.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
Why two X models at 8 cores when you have an unlocked CPU anyway?

Because binning and most people do not overclock.

But that's not the point I'm trying to make. I'm not making a value judgement on whether or not three 8-core SKUs are too many or too few, but that past predicts future and everything else being equal three SKUs is more likely than two.

To put it another way, if you had to put odds that the flagship LGA-1336 successor come Cannon Lake will be 9600K, 9700K, or 9800K, which would you deem most likely?
 

IEC

Elite Member
Super Moderator
Jun 10, 2004
14,323
4,904
136
There is also the possibility that besides saving the top 2-5% of dies for Threadripper 2950X, they will set some aside for a 2800X Black Edition... or similar chip.
 
  • Like
Reactions: scannall

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
Yes, of course that's possible. But again, not what I'm getting at. Lack of a 2800X SKU only makes this slightly less likely to be legit. Even the typo for the embargo date is plausible. The real big issue I see is getting the technical specs of Raven Ridge wrong. That's not a typo, it's ignorance, and a mistake that is far more likely to be made by a faker.
 
  • Like
Reactions: raghu78

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
Because binning and most people do not overclock.

But that's not the point I'm trying to make. I'm not making a value judgement on whether or not three 8-core SKUs are too many or too few, but that past predicts future and everything else being equal three SKUs is more likely than two.

To put it another way, if you had to put odds that the flagship LGA-1336 successor come Cannon Lake will be 9600K, 9700K, or 9800K, which would you deem most likely?

Intel segments by locked clock speed, so they can go segmentation crazy.

The Ryzens mainly sold as enthusiast parts. I saw a lot of sales on the X parts driving the price down right on top of the non X parts which is a strong indicator that overclocking or not, most people were skipping the X parts, so those SKUs seemed to be a mistake if you have to collapse the pricing to get rid of them. I expect a big factor in this was people overclocking the Non-X CPUs. There is no way I would have bought an X, all the reviews were essentially recommending to skip the X, etc...

So the future shouldn't repeat past mistakes and culling the SKUs and make them more attractive at each remaining SKU, makes more sense than simply repeating the way it was before.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
You're missing the point. I'm not saying that reducing the number of SKUs to two doesn't make sense. A lot of things make sense. Removing SMT from the 2700 and more aggressively binning the 2800X over the 2700X would also make sense for product stratification. But just because some random thing makes sense does not make it likely. Everything else being equal, the most likely thing looks a lot like the thing that came before it.
 

FIVR

Diamond Member
Jun 1, 2016
3,753
911
106
AMD is holding back on the 2800X because they know Intel is holding its own 8 core 9700k. Intel will release an 8 core chip to rain on AMD's parade and only then will AMD unleash their 2800X.


If AMD released their full lineup intel could just drop their 8 core and steal their thunder. You'd have 700 articles talking about how intel totally changed computing by offering 8 real cores for reasonable prices and nobody would care about 12nm Ryzen.
 
  • Like
Reactions: VirtualLarry

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
You're missing the point. I'm not saying that reducing the number of SKUs to two doesn't make sense. A lot of things make sense. Removing SMT from the 2700 and more aggressively binning the 2800X over the 2700X would also make sense for product stratification. But just because some random thing makes sense does not make it likely. Everything else being equal, the most likely thing looks a lot like the thing that came before it.

What exactly is your point? You think it's fake because it doesn't look enough like it did last time?

Noted. We will know soon enough whether this was legit.

I think it's real, because this is simply too much work to do a big chunk of tedious slide deck for a gag, and I don't see anything unreasonable outside of a few typos.

What is even the point of faking such a small increment? You would think if someone was going to fake it for shiggles, they would throw in something more juicy, like a 12 core Ryzen or 4.8GHz top clock speed. There is just too much boring work here, to bother faking it. There is no zinger.

So it rings true, and I will keep discussing it as if it is real with others who think so.
 
Last edited:
  • Like
Reactions: Space Tyrant

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
What exactly is your point? You think it's fake because it doesn't look enough like it did last time?

No. Did you read my previous posts at all?

The big red flag is getting the technical specs on Raven Ridge wrong. That's very uncharacteristic, but if that was the only thing out of place it might be possible to look past it. But then you combine that with the 2017 typo, showing off scenarios where the top SKU is beaten by the 8700K while the reverse are absent (and AMD's website makes the claim that the 1800X is only 2% slower than the 8700K @ 1080P, which makes it even weirder), the big jump from 65-105w TDP and the lack of the 2800X SKU. Each of these acts as a sort of force multiplier on the others, but of these the lack of 2800X is the least significant. I'd still think these slides are most likely fake if it had the 2800X. It's not that big of a deal, but it does contribute a bit.
 
Last edited:

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
No. Did you read my previous posts at all?

The big red flag is getting the technical specs on Raven Ridge wrong. That's very uncharacteristic, but if that was the only thing out of place it might be possible to look past it. But then you combine that with the 2017 typo, showing off scenarios where the top SKU is beaten by the 8700K while the reverse are absent, and the lack of the 2800X SKU. Each of these acts as a sort of force multiplier on the others, but of these the lack of 2800X is the least significant. I'd still think these slides are most likely fake if it had the 2800X. It's not that big of a deal, but it does contribute a bit.

It's an early version leak with an obviously incomplete slide deck, and some of these slides may have been meant for internal consumption.

I have been seeing these kinds of leaks for decades and the fakers stand out like a sore thumb, and this one does not have those warning flags. The fakers almost always aim for minimal effort (1 or 2 slides usually bad photoshops), with big payoff (some big step forward).

This is big effort, with small payoff. The opposite of past fakes. I would say very low odds this is faked.
 

HurleyBird

Platinum Member
Apr 22, 2003
2,670
1,250
136
This is big effort, with small payoff. The opposite of past fakes. I would say very low odds this is faked.

They do look polished and elaborate, and I agree most fakers are easy to spot, but there are also a ton of people out there with the skills to pull something like this off. I probably could if I put my mind to it, and I don't even identify as an artist.

If these slides are actually legit then AMD put the wrong year as the embargo date, messed up the specs of their own processors, and only chose to showcase the one scenario where they are weakest against their competition, and even then in a way where they've supposedly fallen further behind the 8700K than the processor that this one usurps.

The only way this is "internal" is if it's still a work in progress. And that's the only thing that would excuse what would otherwise be gross incompetence. Things are far more likely to leak after they've been distributed though.
 
Last edited:

Jan Olšan

Senior member
Jan 12, 2017
273
276
136
The wrong year is there because the employee (yeah I think it is legit) reused the presentation that unveiled Ryzen 5 last year. Look at one of the slides saying that Ryzen 5 launches 11th of April 2017 (true) and that NDA ends 15th of march (actual day when Ryzen 5 1xxx presentation went public).

There are often typos or mistakes in presentations shown under NDA and they send you a fixed version only closely before the NDA lift.

Edit: Plus the contents have lots of stuff that a random troll would be unlikely to come up with and the specs are fairly conservative (experienced gut feeling). And I agree that it would be too much work to fabricate this many slides, many with minor info. Faker would only make a couple.

Edit2:

WhyCry said:
For those saying slides are fake. I've reached out to El Chapuzas for confirmation, they are 100% sure they are legit. Plus, all their previous leaks from AMD were true.
A different source also confirmed to me that their sample is 105W TDP, aka 2700X.

https://videocardz.com/75194/amd-ry...pricing-performance-leaked#comment-3791852619
 
Last edited:
Feb 19, 2017
40
63
51
Slides are legit as far as I see.

I got some info regarding the Pinnacle Ridge from a friend who is working for Global Foundries. He said that as far as he knows the 12nm LP brings 8-10% Fmax boost to the existing Ryzen line. Also he had stated that a great design change is not going on, zero IPC gain is highly probable. As I expected Pinnacle Ridge and Summit Ridge would essentially be like the same core.

Five months ago, I said this. So it pretty much sums this leaks up. Seems like real Fmax boost is at around 10-12% not 8-10%
 

Asterox

Golden Member
May 15, 2012
1,026
1,775
136
I think it's real.

Which is more likely:

1. In a fake this elaborate and time consuming to produce, the faker somehow missed that the embargo date was last year while he produced all of these slides?​

2. Some corporate drone made a typo?​

The faker seems more likely to pay attention to and catch such an obvious issue than the corporate drone is.

As i wrote before, R5 2600X looks like very bad joke or someone was obviously drunk as a sponge.:rolleyes:

https://forums.anandtech.com/threads/amd-ryzen-gen-2-set-for-q2-2018.2530140/page-19#post-39335171
 

dnavas

Senior member
Feb 25, 2017
355
190
116
I would think that 1700 buyers might feel like this is a decent upgrade, while 1800X buyers might look at the base clocks and laugh (that's where I am, anyway). Did fmax move much if you really need 40W to go from 4.1 to 4.35 boost? Another 20W would hardly matter, really, if that's the case. And why 4.35 and not 4.4? If you're going to add 10W to the tdp, why not get the extra 50Mhz? Are existing boards going to be able to deliver 4.35 boost if they're not rated for the full tdp? 8700k looks pretty safe for Intel, but then there wasn't a chance that AMD was going to hit anything close to 5Ghz, so that's pretty much as expected. Doesn't make me feel less disappointed, though -- always better when we get alternating smack-downs.
ymmv
 

Asterox

Golden Member
May 15, 2012
1,026
1,775
136
It's a ~6% boost in top clock speed(along with improved XFR/PB modes), which is inline with reasonable expectation.

What were you expecting?

It is joke Monty Python style, or must by false specification no doubt.As yourself one question why AMD never lounched this red beast, and presented that model as the "last new or updated Phenom II X6 CPU".

Phenom II X6 1100T, 3.3-3.7ghz Turbocore

Phenom II X6 1200T, 3.3ghz-3.9ghz Turbocore

If by any chance R5 2600X 3.6-4.25ghz is real specification, this is simple just useless updated model.:cool:



 

PeterScott

Platinum Member
Jul 7, 2017
2,605
1,540
136
It is joke Monty Python style, or must by false specification no doubt.As yourself one question why AMD never lounched this red beast, and presented that model as the "last new or updated Phenom II X6 CPU".

Phenom II X6 1100T, 3.3-3.7ghz Turbocore

Phenom II X6 1200T, 3.3ghz-3.9ghz Turbocore

If by any chance R5 2600X 3.6-4.25ghz is real specification, this is simple just useless updated model.:cool:

You didn't answer the question, and it was a simple one:

What were you expecting?

Because all this really looks like, is that you had unrealistic expectations, for what was known to be a mild refresh, and are now in let down mode.
 

DrMrLordX

Lifer
Apr 27, 2000
21,582
10,785
136
AMD is holding back on the 2800X because they know Intel is holding its own 8 core 9700k. Intel will release an 8 core chip to rain on AMD's parade and only then will AMD unleash their 2800X.

Interesting theory, but we still have no indication if or when Intel would release such a chip. We can reasonably expect that it won't be in April 2018, and the longer they hold the chip in abeyance, the closer it gets to Icelake-S launch in 2019, making the launch less practical by the day.

I think the more likely event is that AMD had fine-grain binning/segmentation between the 1700X and 1800X due to die quality variance on the 14LPP process. If you look at Silicon Lottery's original data regarding max clocks for 1800X, 1700X, and 1700 dice, you can see why the segmentation was there - 1800X chips were the most likely to hit (or exceed) the magical 4.0 GHz barrier. With 12LP we aren't as process-limited. There isn't going to be a 2800X simply because there doesn't need to be one. Even if they could bin for a ~100 MHz higher max clock per die, it'll be pointless when 2700X owners will by-and-large be able to throw in more volts and cooling to achieve the same effect - something you couldn't necessarily do with a 1700X.
 

DisEnchantment

Golden Member
Mar 3, 2017
1,590
5,722
136
VCZ

AMD-Ryzen-Threadripper-2020.jpg




AMD-Ryzen-Threadripper-Monster-Truck.jpg



Dust hasn't yet settled on 2700X now this :p

Interesting..., "dominant leadership", 32C/64T @ 5GHz ? ... :p
 
Last edited:
  • Like
Reactions: lightmanek

CatMerc

Golden Member
Jul 16, 2016
1,114
1,149
136
Nevermind, it's a fake. Almost certainty. Raven Ridge has 6MB cache L2 + L3, not 10MB. A typo is one thing, but getting technical specs totally wrong is another.

Combined with the date, the 105W TDP, lack of 2800X, and benchmarks showing the 8700K beating AMD in 1080p gaming (why not just show 4K?), it's all just a bit too much.

If it isn't a fake, someone at AMD should get fired or at least disciplined over how poorly these slides were put together.
They're very poorly done, but I believe they're real. Just seems like AMD's MO.
 

FIVR

Diamond Member
Jun 1, 2016
3,753
911
106
Interesting theory, but we still have no indication if or when Intel would release such a chip. We can reasonably expect that it won't be in April 2018, and the longer they hold the chip in abeyance, the closer it gets to Icelake-S launch in 2019, making the launch less practical by the day.

I think the more likely event is that AMD had fine-grain binning/segmentation between the 1700X and 1800X due to die quality variance on the 14LPP process. If you look at Silicon Lottery's original data regarding max clocks for 1800X, 1700X, and 1700 dice, you can see why the segmentation was there - 1800X chips were the most likely to hit (or exceed) the magical 4.0 GHz barrier. With 12LP we aren't as process-limited. There isn't going to be a 2800X simply because there doesn't need to be one. Even if they could bin for a ~100 MHz higher max clock per die, it'll be pointless when 2700X owners will by-and-large be able to throw in more volts and cooling to achieve the same effect - something you couldn't necessarily do with a 1700X.

No, it's definitely that they are holding back. You don't create a nomenclature for the first generation of a product series that will exist for at least half a decade only to change that nomenclature with the second generation of products. If what you said was accurate, AMD would've simply labeled the 2700X as the 2800X and then further bifurcated the 2700 into two different skus with different clock speeds. It makes no sense to simply drop the 2800X nomenclature.

Also, I can recall other instances where AMD did this (during the Athlon-FX days) where they would release the "top" sku but keep the next sku secret until intel released its competitor.
 
  • Like
Reactions: VirtualLarry