AMD rumoured to be creating consumer oriented 16 Core/32 Thread Ryzen CPU

Trimlock said:
Yes, the op is talking about a personal device which will most likely have a consumer chipset which limits what type of CPU and how many you have.

If it is just 2 8cores then it won't be competitive against the new Intel chipset coming out at the end of the year.
Click to expand...

Personal device? WTF is that? Why wouldn't a MCM 16 core with quad channel memory be competitive.


For the kind of loads that apply to 16 cores, there should be no issues.
 
Trimlock said:
Yes, the op is talking about a personal device which will most likely have a consumer chipset which limits what type of CPU and how many you have.

If it is just 2 8cores then it won't be competitive against the new Intel chipset coming out at the end of the year.
Click to expand...

it's not going to be.. the 16/32's just going to be the low end server platform.. it's basically going to be the return of the C32/G34 socket days they used with the phenom II x4 and x6 based opteron's and magny-cour(G34). so with C32 you had the option of running a single processor or dual processors with dual channel DDR3 and both C32/G34 used the same chipset. why i think this will be the case is the fact that because the 32/64 used quad dies it uses 8 channel DDR4 where as the 16/32 uses 2 dies meaning it'll just be 4 channel DDR4 thus not being compatible with the larger socket. i have no clue what the sockets will be called or what their pin layout will be but that's my best educated guess on what will happen. but it'll be up the AIB's to decide on single or dual socket board designs.
 
sirmonkey1985 said:
it's not going to be.. the 16/32's just going to be the low end server platform.. it's basically going to be the return of the C32/G34 socket days they used with the phenom II x4 and x6 based opteron's and magny-cour(G34). so with C32 you had the option of running a single processor or dual processors with dual channel DDR3 and both C32/G34 used the same chipset. why i think this will be the case is the fact that because the 32/64 used quad dies it uses 8 channel DDR4 where as the 16/32 uses 2 dies meaning it'll just be 4 channel DDR4 thus not being compatible with the larger socket. i have no clue what the sockets will be called or what their pin layout will be but that's my best educated guess on what will happen. but it'll be up the AIB's to decide on single or dual socket board designs.
Click to expand...
I already understand that, the op is talking about a rumored platform to compete with Intels HEDT's. So, of it is a normal consumer release I can see AMD disabling 2P and restrict use to 16 (maybe 20core?) CPU's.
 
Trimlock said:
I already understand that, the op is talking about a rumored platform to compete with Intels HEDT's. So, of it is a normal consumer release I can see AMD disabling 2P and restrict use to 16 (maybe 20core?) CPU's.
Click to expand...

What are you talking about? What 20 cores? Here is the original rumor text:

"
Public knowledge by now but AMD has a new HEDT platform coming out in a couple of months. You’ll see more of it at Computex I believe. It’s a 16 core /32 Thread, quad channel behemoth. And it is insanely quick in the tests that Ryzen is already excelling at. So Cinebench, and all other related productivity programs. The gaming issues that were causing the Ryzen AM4 CPUs to behave erratically to say the least have been ironed out. It’s akin to a newer revision on a newer platform. This should be competing with the Xeon and of course 6950X Intel offers for $1700~$1800USD, but at about $1,000 USD if not less for some Skews. Coming soon.

CPSs are pretty big physically, about twice the size of surrent 6950X CPUs and a bit more perhaps. And if you were hoping for pins, nope it’s strictly LGA! It’s NOT 8 channel, but Quad.

Will be a splendid competition between X299 and this AMD platform. Skylake-X is pretty good, not revolutionary but a meaningful step up in IPC and the clocks are pretty high as well. If Intel will have a 32 core part to compete on X299 remains to be seen, but the HEDT platform is going to change quite a bit in the next 4 to 6 months.?

"

Nothing about "consumer" model.

This can be a high end render box or server. There is no magic difference between the CPU in either roll.

99%+ chance, this is a Muliti-chip module with two Ryzen 8 cores the same as the ones in the 1800X.

For the kinds of loads you want to throw 16 cores at (rendering), a MCM is all you really need.
 
OrangeKhrush said:
This was actually quite comical, you managed to take something completely out of context and spin it off like you do best, creating your own fiction as you go along.

1) Like i7 7700K ie: AM4 = 1151/1150, mainstream platform.

2) I don't know how you derived from this that SP4 is a failed platform and that they are spining off server chips as HEDT, in case you have forgotton Intel do the same with all their LGA 2011 chips, recycling server chips and selling them as HEDT, but of course it is unlike you to paint two things with the same brush.

3) I am prepared to wait for performance determinations, if anything learned not to take your "opinion" to seriously, most of it is after the fact, the before the fact "predictions" missed so badly, I would be surprised if you hit water falling out a boat.
Click to expand...

Fiction? I am not the one has been saying in forums that "SMU" is AMD's version of hyperthreadng neither I am who is saying that all the current commercial Ryzen chips on stores are "rebranded engineering samples". LOL

1) You said that the new X399 platfform "is a genuine HEDT platform", and that "1800X was really just the i7 7700K equivilant (sic)". But AMD advertised 1800X on AM4 as a competitor for 6900k, not as competitor for 7770k. Do you need me to post AMD slides again?

2) For the first part just connect the dots. The second part, we know that both AMD and Intel recycle server dies for desktop. Bulldozer FX desktop used the dies rejected for Opteron. We know all this and that it is unrelated to the point made.

3) It seems it still hurts that most of my predictions were correct. It is not my fault that some people believed the hype.
 
Formula.350 said:
In the end I hope we do get a Workstation oriented Naples. Or if anything that the Naples server chip kicks enough ass that motherboard makers will just say "Eff this, we're making some badass single-socket Naples motherboards!!" :D
Click to expand...

It is not Naples. Naples is MCM4 and SP3.

This supposed desktop platform is based in Snowy Owl (MCM2 and SP4).
 
NKD said:
Oh and I am tired of this crap about shit running in GPUs. If everythjg was running on GPU what's the point of intel having a 10 core desktop chip? Like I said that statement goes both ways so let's not forget that. If there is a place for Intel $1700 dollar chip there will be a place for amd 16 core chip for $1000.
Click to expand...

The argument wasn't about number of cores but about LCU vs TCU and what kind of workloads run better in each.
 
juanrga said:
Fiction? I am not the one has been saying in forums that "SMU" is AMD's version of hyperthreadng neither I am who is saying that all the current commercial Ryzen chips on stores are "rebranded engineering samples". LOL

1) You said that the new X399 platfform "is a genuine HEDT platform", and that "1800X was really just the i7 7700K equivilant (sic)". But AMD advertised 1800X on AM4 as a competitor for 6900k, not as competitor for 7770k. Do you need me to post AMD slides again?

2) For the first part just connect the dots. The second part, we know that both AMD and Intel recycle server dies for desktop. Bulldozer FX desktop used the dies rejected for Opteron. We know all this and that it is unrelated to the point made.

3) It seems it still hurts that most of my predictions were correct. It is not my fault that some people believed the hype.
Click to expand...
Dammit, is it that hard to read? He wasn't talking about competitor but being the top chip for the board aka: 7700k (as a general reference not a fact of comparison in performance).
 
Trimlock said:
I already understand that, the op is talking about a rumored platform to compete with Intels HEDT's. So, of it is a normal consumer release I can see AMD disabling 2P and restrict use to 16 (maybe 20core?) CPU's.
Click to expand...

We already have a AMD platform to compete with Intel HEDT

6bv9kJn.jpg


This new 16C/32T X399 CPU is just a rebranded 16C/32T SP4 CPU. It is not a CPU optimized for desktop usage, just as a 2.2GHz 20C Xeon is not. Yes, I know a pair of people purchase Xeon and uses it as HEDT. But that is all.
 
Last edited:
OrangeKhrush said:
I don't see it being a dual socket.
Click to expand...

Yeah the wording makes it kinda confusing.
It could be interpreted as a single 16C die that also supports a 2 socket motherboard, while others are reading it as 8C in a 2 socket implementation for 16 cores.
Considering the size of the die (if some reports are correct), I am with you that it is a single die that can support both single and dual socket motherboards, but its spec rating is going to be important.
Cheers
 
juanrga said:
It is not Naples. Naples is MCM4 and SP3.

This supposed desktop platform is based in Snowy Owl (MCM2 and SP4).
Click to expand...
You're right, I am indeed throwing around the incorrect codenames, but technically Snowy Owl is slated to be an APU along with the rest of the predatory bird-named chips; however, the rumored difference with Snowy Owl is that it'll be using the... Naples cores.

Splitting hairs aside, I honestly would rather there be a non-APU version available, but if ALL of the Naples parts will be APUs then I totally understand why AMD wouldn't have a CPU-only HEDT/Workstation part. Furthermore, if the Naples are APUs and thus all Snowy Owl is, is a cut down version, then I can't really see the flaw in continuing to call it "Naples". The reason we differentiate Naples and Ryzen is due to the very large differences in their die structure/layout, which makes them very different animals (no pun intended). That wouldn't be the case in terms of Snowy Owl, though.

Granted, this is just my opinion, and was purely put out there for the sake of discussion. heh
 
Formula.350 said:
You're right, I am indeed throwing around the incorrect codenames, but technically Snowy Owl is slated to be an APU along with the rest of the predatory bird-named chips; however, the rumored difference with Snowy Owl is that it'll be using the... Naples cores.

Splitting hairs aside, I honestly would rather there be a non-APU version available, but if ALL of the Naples parts will be APUs then I totally understand why AMD wouldn't have a CPU-only HEDT/Workstation part. Furthermore, if the Naples are APUs and thus all Snowy Owl is, is a cut down version, then I can't really see the flaw in continuing to call it "Naples". The reason we differentiate Naples and Ryzen is due to the very large differences in their die structure/layout, which makes them very different animals (no pun intended). That wouldn't be the case in terms of Snowy Owl, though.

Granted, this is just my opinion, and was purely put out there for the sake of discussion. heh
Click to expand...

AMD only got one die currently and its Zeppelin. Naples, Ryzen (And the future Pineapple Ridge) etc all use the same. The only new die AMD is coming up with anytime soon is the quadcore APU for mobile and desktop called Raven Ridge.
 
OrangeKhrush said:
1) The first release steppings are from the same wafers used by the retail parts, in fact most review kits recieved were actually used prior with thermal grease residue reported by many with review samples.

2) No you said it was rebranded Naples chips because somehow to you SP4 failed wherever you got that from.

3) In workstation tasks it was compared to the 6900K and in Gaming notably the 1700 was compared to the 7700K, AMD never expressly stated that AM4 was their only performance PC socket, hence why this is very surprising because it wasnt' expected. Intrisicaly it makes the Ryzen 7 the AM4 equivalant to what Intel's I7 11XX parts are, the highest end of the mainstream or enthusiast gamer parts.

4) No you did not, and predictions after the fact are immaterial to me, you stated on THG that it was cast in stone the IPC per Cinebench was SB or less, but it is about 12+ % which is a major miss, You also made it cock sure that there were not 6 core SKU's i guess you got that wrong to, then you got the clock speed predictions very wrong, you only amended that once there was sway of expected speeds, frankly that is not a prediction, it is like me saying Spain got thumped by Netherlands in the last world cup, oh well yeah they did. Your prediction page is really just recycled information from other popular sources of insider knowlege, don't be a poser
Click to expand...

1) Irrelevant and unrelated to your nonsensical claim that Ryzen chips in stores are "rebranded engineering samples"

2) Again you accuse me of saying things that I didn't say. Naples is MCM4 and uses SP3 socket. I am talking about MCM2 chips using SP4 socket.

3) Another attempt to rewrite history. On the launch slides AMD said

R7 1800X ~ i7 6900k

R7 1700 ~ i7 7700k

for both compute and games. Almost any site has reproduced the slides from AMD. Your former claim that 1800X is the equivalent of the 7700k was and continues being false.

4) Really? Do you insist on attacking me? Do you want me to mention here what really happened at THG forums? Think well the answer.
 
juanrga said:
1) Irrelevant and unrelated to your nonsensical claim that Ryzen chips in stores are "rebranded engineering samples"

2) Again you accuse me of saying things that I didn't say. Naples is MCM4 and uses SP3 socket. I am talking about MCM2 chips using SP4 socket.

3) Another attempt to rewrite history. On the launch slides AMD said

R7 1800X ~ i7 6900k

R7 1700 ~ i7 7700k

for both compute and games. Almost any site has reproduced the slides from AMD. Your former claim that 1800X is the equivalent of the 7700k was and continues being false.

4) Really? Do you insist on attacking me? Do you want me to mention here what really happened at THG forums? Think well the answer.
Click to expand...
#3 is wrong again in the context he gave: (soory other members but this is now the 3rd time I am typing this with no response)

1800x to the 370X platform as the 7700k is to its platform.

Not as a direct competitor but a relationship to their respective platforms.

How hard is that to understand? He is talking about platforms as his focus and you keep harping on CPUs and their comparisons.
 
Formula.350 said:
You're right, I am indeed throwing around the incorrect codenames, but technically Snowy Owl is slated to be an APU
Click to expand...

Some sources claim that Snowy Owl is a server CPU, but yes, the codename is really irrelevant. Change "Snowy Owl" by XYZ if you prefer.


Naples --> SP3 socket --> MCM4 --> 8 memory channels

Snowy Owl --> SP4 socket --> MCM2 --> 4 memory channels

Summit Ridge --> AM4 socket --> Single die --> 2 memory channels

This rumored X399 consumer platform is derived from the SP4 platform
 
Last edited:
juanrga said:
1) Irrelevant and unrelated to your nonsensical claim that Ryzen chips in stores are "rebranded engineering samples"

2) Again you accuse me of saying things that I didn't say. Naples is MCM4 and uses SP3 socket. I am talking about MCM2 chips using SP4 socket.

3) Another attempt to rewrite history. On the launch slides AMD said

R7 1800X ~ i7 6900k

R7 1700 ~ i7 7700k

for both compute and games. Almost any site has reproduced the slides from AMD. Your former claim that 1800X is the equivalent of the 7700k was and continues being false.

4) Really? Do you insist on attacking me? Do you want me to mention here what really happened at THG forums? Think well the answer.
Click to expand...

1) never did say that, just said they were all of the same stepping and silicon, learn to english bruh, huge difference between stipulation and likening, likening doesn't mean it is.


2) again you are having trouble with comprehension it was again posted by just reason, if you are having trouble with this concept you could try asking. Seems you're the only one that doesn't get it.

3) the initial thread on THG closed but in there is full of the complete misses you made with CB, when that was the order of your fixation, not only did you miss the multi thread bus,. But missed singlethread,. The same crap on semi accurate where you got it wrong, failed to acknowledge that and then tried spinning off.

Here is free advice for you, don't make predictions, act like a WUM in the process then cry when you got it wrong. Did you ever go back to semi accurate? There are plenty that have a bone with you there
 
OrangeKhrush said:
1) never did say that, just said they were all of the same stepping and silicon, learn to english bruh, huge difference between stipulation and likening, likening doesn't mean it is.


2) again you are having trouble with comprehension it was again posted by just reason, if you are having trouble with this concept you could try asking. Seems you're the only one that doesn't get it.

3) the initial thread on THG closed but in there is full of the complete misses you made with CB, when that was the order of your fixation, not only did you miss the multi thread bus,. But missed singlethread,. The same crap on semi accurate where you got it wrong, failed to acknowledge that and then tried spinning off.

Here is free advice for you, don't make predictions, act like a WUM in the process then cry when you got it wrong. Did you ever go back to semi accurate? There are plenty that have a bone with you there
Click to expand...

1) You said the nonsense that current RyZens in stores are rebranded engineering samples

https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-29#post-38793748


3) I am happy with my dozen of predictions about Zen. I did a pair of mistakes, but the rest was fine.

Funny that you continue mentioning a closed thread at THG in the hope that no one here knows the truth, but mods remember very well you and all the BS that you posted about Zen, despite you have changed your nickname:

OrangeKhrush is a former member here who went by a different name, I would take any information from him with a grain of salt. I wouldn't waste any effort on information from him.
Click to expand...

http://www.tomshardware.co.uk/forum...egathread-faq-resources/page-24.html#19449834

I had enough with your off-topic rants. Bye.
 
juanrga said:
1) You said the nonsense that current RyZens in stores are rebranded engineering samples

https://forums.anandtech.com/threads/ryzen-strictly-technical.2500572/page-29#post-38793748


3) I am happy with my dozen of predictions about Zen. I did a pair of mistakes, but the rest was fine.

Funny that you continue mentioning a closed thread at THG in the hope that no one here knows the truth, but mods remember very well you and all the BS that you posted about Zen, despite you have changed your nickname:



http://www.tomshardware.co.uk/forum...egathread-faq-resources/page-24.html#19449834

I had enough with your off-topic rants. Bye.
Click to expand...
Pot calling Kettle Black? off topic rants? Seriously you haven't answered one question asked of you. You keep parroting lies and half truths.

You have yet to speak to the multitude of testimonials of Ryzen GREAT performance, you just keeping reiterating synthetics and benches out-of-context as if that is what people will see. Most have come from Intel , NOT AMD FANBOIS, and they speak of outstanding performance over the Intel they had previously.
 
lolfail9001 said:
Testimony compilation, got any?
Click to expand...
just the guys here and the few times I checked at OCN. This far I haven't seen one complaint on performance, most saying it was very fast. The only common complaint is the inability to run their ram at purchased speeds, but that will come in due time. The occasional issue here and there in regards to little gremlins that range from minor to annoying but nothing catastrophic thus far.

Honestly it gets old and tiresome seeing people post on a product they don't have nor have any desire to purchase constantly commenting in direct contrast to such testimonials. Benchmarks and reviews can only tell one so much, sometimes experience first hand tells us more.
 
lolfail9001 said:
For quite a few, even FX is fast enough.

Quite the opposite from my experience, past certain individual threshold you cannot tell a difference without measuring.
Click to expand...

Well I used a simple benchmark, I used Fallout 4 and played through a section that my old FX chip used to struggle with on my 290x video card. Same video card with a Ryzen at stock clocks didnt chug at all stayed nice and smooth, despite the fact that Ryzen is not as powerful at 1080p resolution and with SMT enabled and Fallout 4 does not make use of it very well. To me that was a noticeable upgrade despite the fact my FX was clocked at 4.7. Not all of us Benchmark for forum cred, we just want a reasonably fast system without spending a stupid amount of money and Ryzen fits that perfectly. Also nice to be able to encode much faster now. Now if you want to compare 310 fps to 285 fps then yeah benchmarks are needed as no one can tell the difference anymore.
 
lolfail9001 said:
For quite a few, even FX is fast enough.

Quite the opposite from my experience, past certain individual threshold you cannot tell a difference without measuring.
Click to expand...

It depends what your workload is. When it comes to video encoding (which I do a lot of) I know exactly how long something takes and I can definitely feel the difference. Furthermore I usually recompile encoders to further enhance performance and even test various compile settings to figure out what the best setting is and how it can affect outcomes.

Reviews just serve as a baseline. I use it as the worst case not the best case because I usually get better scores across the board than most.
 
Anecdotal evidence is just that and it's subjective to boot. Some of it you can immediately discount thanks to posting history showing bias.

For me, looking at my workload (video rendering) current Ryzen can't match my seven-year and four-year old systems' rendering times as Ryzen is 8c/16t and I have two 12c/24t systems, even though Ryzen is clocked higher by about 1 GHz. This is why a professional or even consumer-oriented 16c/32t system looks interesting to me. A small bump in core clocks and a 4c/8t bump in core count would definitely reduce rendering times and finally be a worthwhile upgrade.
 
Recent info about the X399 platform coming from CanardPC. Clocks in second gen engineering sample are 3.1GHz base and 3.6GHz turbo. But those engineering samples are using B1-grade silicon; therefore don't expect improvements from better silicon.

The marketing TDP is 180W. The real TDP is expected to be higher than 200W

 
lolfail9001 said:
Check your math, then, stock 1700 is like ~75W. This thing so far is basically 2 1700s glued on a package.
Click to expand...

Tricky part is working out how much power is required by the mesh CCXs for 16C variant and also the fabric, but still that would not be anywhere close to that figure from CPC, but maybe they are adjusting the TDP to correlate with their own findings with R7 *shrug*.
Cheers
 
CSI_PC said:
Tricky part is working out how much power is required by the mesh CCXs for 16C variant and also the fabric, but still that would not be anywhere close to that figure from CPC, but maybe they are adjusting the TDP to correlate with their own findings with R7 *shrug*.
Cheers
Click to expand...

Given the target frequency for a system like this it will be competitive with Intel... power is a number one priority for AMD right now.

Wait till it launches and then bitch.
 
juanrga said:
We already have a AMD platform to compete with Intel HEDT
Click to expand...
I think AMD is anticipating Intel's new up coming HEDT for late this year or early next, Purley, here. You're right that their current high end CPU are targeted at Intel's current HEDT of course.

juanrga said:
This new 16C/32T X399 CPU is just a rebranded 16C/32T SP4 CPU. It is not a CPU optimized for desktop usage, just as a 2.2GHz 20C Xeon is not.
Click to expand...
The arch won't change but I'd expect some cherry picked high OC'ing dies to go into it. Yeah not many will have use for a 16C/32T ~4Ghz (when OC'd, I'd expect stock clocks to stay around 3.4-3.6Ghz tops) CPU that uses upwards of 250W but the high end has always been more about ePeen size rather than rational use cases and value.
 
mesyn191 said:
I think AMD is anticipating Intel's new up coming HEDT for late this year or early next, Purley, here. You're right that their current high end CPU are targeted at Intel's current HEDT of course.


The arch won't change but I'd expect some cherry picked high OC'ing dies to go into it. Yeah not many will have use for a 16C/32T ~4Ghz (when OC'd, I'd expect stock clocks to stay around 3.4-3.6Ghz tops) CPU that uses upwards of 250W but the high end has always been more about ePeen size rather than rational use cases and value.
Click to expand...
Epeen for gaming or just boutique builds but for professional use they are worth their price.

Sadly we see a lot of these types of systems for purely gaming.
 
Trimlock said:
Epeen for gaming or just boutique builds but for professional use they are worth their price.
Click to expand...
Nearly all actual pros will just buy a single socket server platform from AMD instead. X399 isn't targeted at pros. But then neither is Intel's HEDT really. Anything that offers, much less emphasizes, overclocking isn't really for professional use at all.
 
mesyn191 said:
Nearly all actual pros will just buy a single socket server platform from AMD instead. X399 isn't targeted at pros. But then neither is Intel's HEDT really. Anything that offers, much less emphasizes, overclocking isn't really for professional use at all.
Click to expand...
There certainly features aimed at enthusiasts, Intel isn't stupid, it's nearly a free add on that opens up your market but to say that the HEDT has no market outside enthusiasts is short sighted. There are plenty of boards and set ups that aim at workstations and make for great work computers that keep potential buyers out of some of the more costly Xeon markets. One of the reasons AMD would want in on this market is because it's quite big, the platform is really what sells it. Outside of that though the smaller sockets are where the money's at.
 
You must log in or register to reply here.
Back
Top