Tuesday, September 18th 2018

AMD Readying a 10-core AM4 Processor to Thwart Core i9-9900K?

To sustain its meteoric rise at the stock markets, AMD needs to keep investors convinced it has a competitive edge over Intel, even if it means investing heavily on short-term roadmap changes. According to an Elchapuzas Informatico article, AMD could be working on a new 10-core/20-thread processor for the AM4 platform, to compete with the upcoming Core i9-9900K 8-core/16-thread processor from Intel. The said processor is being labeled "Ryzen 7 2800X" and plastered over CineBench nT screenshots, where due to the sheer weight of its 10 cores, it tops the nT test in comparison to Intel's mainstream-desktop processors, including the 2P Xeon X5650 12-core/24-thread.

The Forbes article that cites the Elchapuzas Informatico, however, is skeptical that AMD could make such a short-sighted product investment. It believes that development of a 10-core die on existing "Zen+" architecture could warrant a massive redesign of the CCX (Zen Compute Complex), and AMD would only get an opportunity to do so when working on "Zen 2," which AMD still expects to debut by late-2018 on its EPYC product line. We, however, don't discount the possibility of a 10-core "Zen+" silicon just yet. GlobalFoundries, AMD's principal foundry partner for CPUs, has given up on 7 nm, making the company fall back to TSMC to meet its 7 nm roadmap commitments. TSMC already has a long list of clientele for 7 nm, including high-volume contracts from Apple, Qualcomm, and NVIDIA. This could force AMD to bolster its existing lineup as a contingency for delays in 7 nm volume production.
Source: Forbes
Add your own comment

133 Comments on AMD Readying a 10-core AM4 Processor to Thwart Core i9-9900K?

#1
londiste
X5650 is not mainstream-desktop and it is from 2010.
From the leaks so far, 2100 nT Cinebench score is roughly at par with i9-9900K, not convincingly higher.

The silicon is a big question though. Zen2 this early?
Posted on Reply
#2
hat
Enthusiast
Begun, the core war has.
Posted on Reply
#3
Japie073
Xeon X5650 is 6C/12T. They used a dual CPU setup. Never have I ever saw a 12 Core 1366 CPU.
Posted on Reply
#4
R0H1T
The alleged deca core isn't possible, supposedly with the current 4 core per CCX arrangement, of course AMD might have an ace or two up their multiple sleeves.
We'll see, but I have serious doubts about the rumor!
Posted on Reply
#5
notb
londisteX5650 is not mainstream-desktop and it is from 2010.
From the leaks so far, 2100 nT Cinebench score is roughly at par with i9-9900K, not convincingly higher.
Exactly. Ryzen 8C/16T competes with i7 6C/12T.
10C Ryzen might not be enough for 9900K. They should go with 12 already.

Of course we would be once more flooded with simple multi-core benchmarks like rendering. :-)
The silicon is a big question though. Zen2 this early?
Surely not Zen1 on current node, since it's basically against the whole architecture.
We know from the start what can be done with CCX and IF.
It's very limiting for AMD. For example: they can't even make an 8-core APU (there's a lot of free space in the package).
IGP takes one CCX space away.

Of course AMD could design a 10-core CPU from scratch if they really wanted. Or 11 cores. Or whatever they fit.
But that would mean the whole brilliant Zen idea lasted just for 2 years.

The other issue is the mount. Would >8 cores work on AM4? How? AMD never told us.

This is also a big question in case of 7nm. People believe AMD will manage to squeeze a scaled down Threadripper into AM4, but how would it work?
In fact, have they ever said that AM4 will get Zen2? Maybe it's time for AM4+?
Posted on Reply
#6
The Quim Reaper
Not going to happen.

The best we can expect are some highly binned 2700x CPU's which can reach, and sustain, a 4.5Ghz boost.

Personally, I think AMD should just ignore the 9900K, it will be too expensive for most and won't take away (many) sales from the 2700X, and just carry on getting the mainstream Zen 2 ready for launch in Q1 2019.
Posted on Reply
#7
R0H1T
notbExactly. Ryzen 8C/16T competes with i7 6C/12T.
10C Ryzen might not be enough for 9900K. They should go with 12 already.

Of course we would be once more flooded with simple multi-core benchmarks like rendering. :)

Surely not Zen1 on current node, since it's basically against the whole architecture.
We know from the start what can be done with CCX and IF.
It's very limiting for AMD. For example: they can't even make an 8-core APU (there's a lot of free space in the package).
IGP takes one CCX space away.

Of course AMD could design a 10-core CPU from scratch if they really wanted. Or 11 cores. Or whatever they fit.
But that would mean the whole brilliant Zen idea lasted just for 2 years.

The other issue is the mount. Would >8 cores work on AM4? How? AMD never told us.

This is also a big question in case of 7nm. People believe AMD will manage to squeeze a scaled down Threadripper into AM4, but how would it work?
In fact, have they ever said that AM4 will get Zen2? Maybe it's time for AM4+?
Why would they tell you? AMD never said anything about mainstream Ryzen topping out at just 8 cores, did they? Do you also think zen2/3 will only have 4 cores per CCX, if not then next Ryzen could bring 12~16 cores for mainstream but that's still at least half a year away.
Posted on Reply
#8
silentbogo
notbSurely not Zen1 on current node, since it's basically against the whole architecture.
You've answered all of your following questions here. It's an early Zen2 prototype, since Zen1 only allows for 4 cores per module, and the package can only fit 2 modules.
Zen 2 (according to AMD slides), is projected to have 6-8 cores per CCX module.
With this in mind, we may expect something ridiculous like a 16c/32t CPU in a consumer segment next year.

Maybe I should wait for an upgrade one more year... this is getting quite interesting.
Posted on Reply
#9
kastriot
Price/perf is winner here so all this is shooting the breeze, and atm amd is winner for next 2-3 years, intel has got what it deserves by not pulling head from sand and milking his loyal customers.
Posted on Reply
#10
Prima.Vera
hatBegun, the core war has.
We are on Episode XXL already...
Posted on Reply
#12
notb
R0H1TWhy would they tell you?
Rather: why would they tell us?
Because they told us that AM4 will be in use until 2020. And people took the bait. Platform longevity and upgrade path is still one of the most popular arguments people mention when recommending Ryzen.
But at the same time people expect 7nm, Zen2 and 16 cores in consumer segment before 2020.
Wouldn't these people want some sort of confirmation? Or at least a clear info that it will still work on the same AM4 motherboards.
AMD never said anything about mainstream Ryzen topping out at just 8 cores, did they?
AFAIK they never said AM4 will go beyond 8 cores as well.
Do you also think zen2/3 will only have 4 cores per CCX, if not then next Ryzen could bring 12~16 cores for mainstream but that's still at least half a year away.
It doesn't matter what I think. What matters is the "could" in your sentence. :)
silentbogoZen 2 (according to AMD slides), is projected to have 6-8 cores per CCX module.
These slides are exactly what I'm missing. Could you give a link? I can't find anything.
Posted on Reply
#13
R0H1T
notbRather: why would they tell us?
Because they told us that AM4 will be in use until 2020. And people took the bait. Platform longevity and upgrade path is still one of the most popular arguments people mention when recommending Ryzen.
But at the same time people expect 7nm, Zen2 and 16 cores in consumer segment before 2020.
Wouldn't these people want some sort of confirmation? Or at least a clear info that it will still work on the same AM4 motherboards.

AFAIK they never said AM4 will go beyond 8 cores as well.

It doesn't matter what I think. What matters is the "could" in your sentence. :)

These slides are exactly what I'm missing. Could you give a link? I can't find anything.
You won't find anything because Zen2 isn't offcial yet, maybe we'll get a preview from AMD before the end of the year. AMD would also be wary to not leak too much unnecessary info before Zen2 lands, especially as EPYC ~ Intel is Serving Major Xeon Discounts to Combat AMD EPYC

We're just looking at a tree in the forest, the actual forest is the bigger deal.
vprem“Intellect Running-Amok” and “others” might call this “Chase-Me”, Competition, but it could also be caused by “Half-Baked” CPUs. You know, those “Brain-Deadened” CPUs that did not make it to being Da Bigge-Vun but are otherwise “bigger”/more-muskular that Dem Weeny-Vuns”Cheapos”.

When Sell, Sell, & Sell is Running Wild-&-Wanton, “Relatively”/Materially it could be seen as Buy, Buy, & Buy but Realistically/”Spiritually”, it could also be due to “Thought-Processing” needing “Its-Walkies”/to-Run-Amok. Especially when The Kitchen is getting Too-Hot due to all that Wild & Wanton Baking. You know, when “Baking” is Running Amok. When “Parts” are being “Binned”/Brain-Deadened/Financially-Justified.
Is this what you call klingon o_O
Posted on Reply
#14
silentbogo
notbThese slides are exactly what I'm missing. Could you give a link? I can't find anything.
Lol. This is gonna be embarrassing. I've decided to find those old slides for you, and out of all the sources the first one that came up in google was an article from WCFTech with an informative title "Fake AMD Ryzen 2800X 12 Core 5.1GHz Slide Sends Media Into Frenzy" ))))
So much for keeping up with news.... :banghead:

So, all we have to go on, is a now-taken-down and non-existent MSI promotional video for B450 motherboard that claimed "8-core and up CPU" support... All clues and hints have been meticulously erased.


Posted on Reply
#15
Mr.Origami696
I think everyone is missing an important point here.

AMD isn't interested in trivial competitions against Intel but in is own way to conceive the CPU technology keeping a solid path.

Why the logic cores number are higher, compared to the physical ones? Well, one reason might be...better compute performances and lower power consumption at a final cost market always competitive and catching for the customers (any category). It is almost obvious that GPU card are the most expansive compared on the many things the CPU's have to do on the office side. So, TDP reasonably increase on gaming rigs where a mid-higher GPU (mostly Nvidia) is the ideal...but AMD just should be keeping their excellent part, providing the best multi thread performances at lower power consumption (and prices).

That's AMD philosophy, after all. Just think about it.
Posted on Reply
#16
Vya Domus
Intel is the one late to the party, not AMD. They can simply wait this one out Zen 2 less than a year away, this 10 core part seems unlikely.
silentbogoSo, all we have to go on, is a now-taken-down and non-existent MSI promotional video for B450 motherboard that claimed "8-core and up CPU" support... All clues and hints have been meticulously erased.
I don't know why it is that hard for people to believe core count can increase without a change in socket/package.

Platform longevity is indeed something that people don't believe in thanks to Intel but it is in fact a thing.
Posted on Reply
#17
medi01
Intel has enjoyed (and still is) margins AMD could never dream of.
Intel's fab advantage has vanished, thank you, TSMC.
Intel is years behind on GPU front.
nVidia won't even be let to join the x86 party (so let's even skip the "how many decades would it take them to catch up").
In "moar coars" ward, AMD has architectural edge, CCX + infinity fabric, 16 core CPU for $699 anyone?

So, where was I: who says that AMD even needs to have symmetric or even "faster" answers to whatever Intel has?
Posted on Reply
#18
R0H1T
medi01Intel has enjoyed (and still is) margins AMD could never dream of.
Intel's fab advantage has vanished, thank you, TSMC.
Intel is years behind on GPU front.
nVidia won't even be let to join the x86 party (so let's even skip the "how many decades would it take them to catch up").
In "moar coars" ward, AMD has architectural edge, CCX + infinity fabric, 16 core CPU for $699 anyone?

So, where was I: who says that AMD even needs to have symmetric or even "faster" answers to whatever Intel has?
AMD needs to cover all their bases, or at least put viable alternatives in segments where Intel enoys a near monopoly. At this point in time AMD's probably doing the best they can, but they can do even better.
Posted on Reply
#19
notb
Vya DomusI don't know why it is that hard for people to believe core count can increase without a change in socket/package.
There's a thing called physics. And another thing called architecture limits.

I'm not saying AMD can't make a 10 core CPU for AM4. But I am skeptical about it being compatible with current AM4 mobos. If it will be - great for Ryzen owners. But why hasn't AMD told us that? It's not a technological secret or anything.
Wouldn't current AM4 owners like to know that "AM4 supported until 2020" means something more than just current Zen+ CPUs being in production for another 2 years?
Because if they switch to a new socket for new CCX modules, what happens to the mystical "upgrade path"?

And there's another issue as well. Next year we might see DDR5 in servers, but DDR4 will remain in consumer segment for another 1-2 years. How will AMD cover this? Will they still be able to build both consumer and server CPUs out of the same parts? :-)
medi01So, where was I: who says that AMD even needs to have symmetric or even "faster" answers to whatever Intel has?
Because their market share is still around 10% of mobile+desktop market and 1% of servers. And they need more.
Mr.Origami696I think everyone is missing an important point here. [cut]
Please read that post once again. Or maybe show it to your life partner or a friend. There's no shame in having a text edited before posting...
I've read it twice and I don't know what you wanted to say. :-/
Posted on Reply
#20
R0H1T
notbThere's a thing called physics. And another thing called architecture limits.

I'm not saying AMD can't make a 10 core CPU for AM4. But I am skeptical about it being compatible with current AM4 mobos. If it will be - great for Ryzen owners. But why hasn't AMD told us that? It's not a technological secret or anything.
Wouldn't current AM4 owners like to know that "AM4 supported until 2020" means something more than just current Zen+ CPUs being in production for another 2 years?
Because if they switch to a new socket for new CCX modules, what happens to the mystical "upgrade path"?

And there's another issue as well. Next year we might see DDR5 in servers, but DDR4 will remain in consumer segment for another 1-2 years. How will AMD cover this? Will they still be able to build both consumer and server CPUs out of the same parts? :)

Because their market share is still around 10% of mobile+desktop market and 1% of servers. And they need more.

Please read that post once again. Or maybe show it to your life partner or a friend. There's no shame in having a text edited before posting...
I've read it twice and I don't know what you wanted to say. :-/
There's this thing called speculation, what you're saying is beyond that. Where did Physics come into all of this, do 16 cores @7nm bore a quantum tunnel that 8 cores @14nm cannot?
Posted on Reply
#21
kings
Given that most 9900K buyers should be gamers (such as the 8700K), an AMD 10 core most likely forced to lower clock speeds than the 2700X, it would bring nothing new!

Furthermore, Intel is advertising this new CPU mostly for gaming, so even less sense makes AMD go that way, when they already know that in pure gaming adding more cores don't solve anything.

In most games the 10-core would be worse than the 2700X, unless they discover some magic to increase clocks significantly. So, I think it would be a release with no sense!
Posted on Reply
#22
hat
Enthusiast
notbI'm not saying AMD can't make a 10 core CPU for AM4. But I am skeptical about it being compatible with current AM4 mobos. If it will be - great for Ryzen owners. But why hasn't AMD told us that? It's not a technological secret or anything.
Wouldn't current AM4 owners like to know that "AM4 supported until 2020" means something more than just current Zen+ CPUs being in production for another 2 years?
Because if they switch to a new socket for new CCX modules, what happens to the mystical "upgrade path"?
Were you aware of AM3 processors that supported both DDR2 and DDR3? This meant that you could stick an AM3 chip in an AM2/AM2+ board. You could also use an AM2/AM2+ chip in an AM3 board, provided it had DDR2 slots. That's the kind of forwards (and backwards) compatibility AMD users have enjoyed previously, and there's no reason to expect this will come to an end with AM4. However, probably not all boards will do this. Support varied from board to board; it was up to the board makers to provide BIOS updates for support. I've taken advantage of this feature myself a number of times in the past.

These features exist for a reason. It's the same reason we had motherboards like the Asrock 775Dual-VSTA. That was a weird (but useful) board for sure. It had DDR and DDR2 slots, and it had an AGP slot as well as a PCI-E slot. Some of us only want to (or can only afford to) upgrade one component at a time once in a while, so these things really come in handy in such situations. How nice would it be if I could put one of the upcoming Whiskey Lake chips in my old, but still functioning socket 1155 board? I wouldn't need to buy a new motherboard and RAM just to upgrade the CPU.
Posted on Reply
#23
techy1
this (rumored) 10 core is as possible as 11 core - yea anything is possible, but not gonna happen. and I sure do hope amd is not preparing anything at all (vs next months intel moves). I do not want amdto spend time/money on Zen+ when Zen 2 is near (aprox 6+ months) and Zen2 will be the fatality move vs intel current and next few year lineup.
Posted on Reply
#24
Mr.Origami696
R0H1TThere's this thing called speculation, what you're saying is beyond that. Where did Physics come into all of this, do 16 cores @7nm bore a quantum tunnel that 8 cores @14nm cannot?
Hi R0H1T, you're perfectly right. Anyway, quantum physics isn't something really light to digest for everyone :) but it is astonishing to see (not only here of course, I'm new here) a slight useless conspiracy theories....even inside the IT environment! :/
Posted on Reply
#25
dj-electric
There comes a point where adding more cores isn't the solution to the problem, because the problem wasn't "not enough cores".
Posted on Reply
Add your own comment
Apr 28th, 2024 05:16 EDT change timezone

New Forum Posts

Popular Reviews

Controversial News Posts