Naples CPU Xeon competitor!

Now it starts to make sense. As with Bulldozer, AMD chose to target the server market and to do that you need a design that handles multithreaded workloads with a reasonable TDP. AMD's implementation seems to come at the cost of desktop application / game performance which is what all the benchmarks have shown. The difference is that Ryzen isn't as weak on the desktop side as Bulldozer (compared to Intel CPUs at the time) was and it's a more compelling server product if the PC Perspective information is accurate. (I have no reason to doubt that it is.) Naples is looking damn good so far.

In other words AMD chose to build a server chip that does OK at playing games. They took the same gamble that Intel did and to be honest it makes sense if they can get the sales they need for business use in datacenters and even smaller server rooms.
 
I don't know about paradigm this and that - but 32 cores all covered by a single VMware license is a GOOD thing. 8 channel RAM too? Nice. And Intel's got a HUGE margin built into their enterprise market so AMD should have a lot of room to operate. They need major Software and Hardware manufacturers to be on board with this though. VMWare's got to WORK. Hyper V's got to WORK. They've got to be able to put this into Cisco UCS, Dell and HP 1-2U and blades.
 
CombatChrisNC said:
I don't know about paradigm this and that - but 32 cores all covered by a single VMware license is a GOOD thing. 8 channel RAM too? Nice. And Intel's got a HUGE margin built into their enterprise market so AMD should have a lot of room to operate. They need major Software and Hardware manufacturers to be on board with this though. VMWare's got to WORK. Hyper V's got to WORK. They've got to be able to put this into Cisco UCS, Dell and HP 1-2U and blades.
Click to expand...

I'm actually far more excited about Naples and what it means than I am about Ryzen. Naples is the product we need to shake Intel out of its apathetic state and further drive technology forward. The Opteron 64 did the same thing back in the day. It shamed the Xeon in a lot of areas and that's what we need again. For the enthusiast and gamer there is a side benefit. The first Athlon FX was basically a single socket capable Opteron and the CPU to beat when it was released. Naples has a much more compelling platform going with it which could be adapted to something we can use in a high end multi-GPU gaming system. Think about it, we could have a motherboard with 64 PCIe lanes, over 8 cores and 16 threads, and potentially a lot of freaking memory bandwidth if they set it up right. We could get at least four channels on a standard EATX board if not more. We could get a more modern feature set without DMI 2.0 limitations. In essence, this could spawn a very viable alternative and very worthy upgrade over X99 and Intel's dated HEDT platform.
 
It would be nice if AMD can get in with the storage providers too. All the SANs I've seen recently- all the heads/storage processor/controllers/ etc, whatever you want to call them, are just Intel chips and mobos. If you want BIG and FAST, and I mean REAL BIG and REAL fast storage with NVMe drives by the petabyte, PCIE lanes is going to be a bottleneck. With an assload more lanes than Intel for what I assume will be a reasonable price (ASS u ME), this could be very interesting to EMC/DELL, Netapp, Pure, Hitachi, etc.
 
Naples looks like an absolute beast. Well done, AMD!

EDIT: And, yes, I'd love to have one (or more) of these at work for video rendering...
 
Last edited:
Ryzen is a server CPU. LPP is the best entry for server. Wait til GloFo switches to EUVL high power process next year for APU and gaming desktop.
 
This is where the money is...I hope it's where AMD's focus has been for the last couple months with OEMs. I will forgive them for a teething issues on the desktop side if they've been busy getting their server architecture properly validated.

If they can hit the ground running with Naples from the big names it's a Big Deal.
 
From TheTechReport:
naplesdemosys.jpg
 
http://www.eetimes.com/document.asp?doc_id=1331444

In an interview, Forrest Norrod, a former Dell server executive who now heads AMD’s server and semi-custom group, set low expectations for Naples. He noted that the company has negligible market share in x86 server CPUs today.

“We will fire rifle shots, not shotgun blasts. We have a good general-purpose product, but our strategy is to identify workloads for which we are clearly best,” Norrod said, including workloads in both public and private cloud computing.

He noted that even the web giants at the OC event, such as Google, have many different workloads, such as search indexing, serving up search results, and serving adds. “Some of those workloads, I’m not going to break in,” he said.
Click to expand...
 
No matter how good Naples is, there is a lot of brand recognition for Intel and the Xeon to overcome. Additionally, AMD has to overcome its own negative stereo type in the industry as "that other CPU maker" or a "knock off" manufacturer of CPUs. In the industry, Intel has a reputation that isn't as much about performance as it is reliability. If you want performance from Intel you simply buy a bigger server. If you want something reliable, virtually any server with "Intel" inside will do. Over the last 20 years these thought processes haven't always been rooted in reality but this is how the company is still perceived. I think word of mouth and the regurgitation of information without fact checking is why these stereo types about AMD hold true even when it has a good product on their hands.

It's the same type of brand bias that GM and Ford have to overcome in a lot of markets with their products. In fairness they did it to themselves but that's another topic.
 
I remember Intel having an issue with convincing their customers to upgrade their systems. Big data can use more processing power but I think a majority have hit IO limitations and are using highly customized solutions.

I think AMD might be barking up the wrong tree, sure they will take some sales away from Intel but I thin Intel is banking on Optane taking off which will them further block the AMD platform.

In the end I think AMD has a winner (more hope than anything) and they will get some server business back but I'm not sure if they will dethrone anything Intel has.
 
Trimlock said:
I remember Intel having an issue with convincing their customers to upgrade their systems. Big data can use more processing power but I think a majority have hit IO limitations and are using highly customized solutions.
Click to expand...

That couldn't be more wrong. However enterprise customers are starting to fade away for cloud.
 
Dan_D said:
No matter how good Naples is, there is a lot of brand recognition for Intel and the Xeon to overcome. Additionally, AMD has to overcome its own negative stereo type in the industry as "that other CPU maker" or a "knock off" manufacturer of CPUs. In the industry, Intel has a reputation that isn't as much about performance as it is reliability. If you want performance from Intel you simply buy a bigger server. If you want something reliable, virtually any server with "Intel" inside will do. Over the last 20 years these thought processes haven't always been rooted in reality but this is how the company is still perceived. I think word of mouth and the regurgitation of information without fact checking is why these stereo types about AMD hold true even when it has a good product on their hands.

It's the same type of brand bias that GM and Ford have to overcome in a lot of markets with their products. In fairness they did it to themselves but that's another topic.
Click to expand...

Are you trying to say that extremely informed and well run companies are buying Intel because its Intel? And not because they have beaten anyone else so much that its pointless and directly a bad business decision to buy anything else?

Pretty much nobody is doing Naples servers for the same reason, its a direct disaster. Not because of some sort of brand loyalty.
 
Shintai said:
Are you trying to say that extremely informed and well run companies are buying Intel because its Intel? And not because they have beaten anyone else so much that its pointless and directly a bad business decision to buy anything else?

Pretty much nobody is doing Naples servers for the same reason, its a direct disaster. Not because of some sort of brand loyalty.
Click to expand...

First off, don't ever assume that the people making purchasing decision are well informed. That's not always the case. In my experience, it seldom is the case. In the IT world there is a huge bias against AMD. It stems from the days when AMD made shitty knock offs of Intel's 386 and 486 CPUs at bargain basement prices. Old school IT guys with 20 and 30 years in the business are now the guys in the management positions who make decisions. They may not even be current on technology so they'll hold to any old thinking or habits they have. Intel has a solid reputation for reliability in its CPUs, chipsets, network solutions, SSD's you name it. People will buy other shit that's cheaper. That's how Broadcrap manages to survive in the server world. That said, Intel has a foot hold in the CPU world and part of that isn't as much brand loyalty as recognition. People know what it is.

AMD does have to fight against brand bias. There are organizations like Match.com who had a good sized HP / AMD Opteron based infrastructure but knowing the guys who made those decisions it was usually about cost and bang for the buck when the company was new. This changed later on but at one time that's how those decisions were made in the earlier days. Other companies I've worked at won't ever entertain the idea of buying non-Intel systems anytime soon. One reason for this is some companies with complex infrastructure have a long validation period before standardizing on new hardware. Some companies will continue to buy servers that have been supplanted by newer models simply because it is there standard. I've seen hundreds of thousands of dollars spent on older models for this reason.

I've worked in datacenters and large infrastructures. They are slow to change and the people in charge of the final purchasing decisions are less technically inclined than you might imagine.
 
Shintai said:
Pretty much nobody is doing Naples servers for the same reason, its a direct disaster. Not because of some sort of brand loyalty.
Click to expand...

So why is Naples a "direct disaster" exactly? And I think it would be pretty hard for anyone to use something that hasn't launched yet.
 
SighTurtle said:
Tom's Hardware has a section where they run Ryzen through "Scientific & Engineering Computations and HPC" tests.
Click to expand...
I've not looked at what Tom's has up, but honestly... after seeing what Stilt had done in his very-technical analysis of Ryzen, I think major review sites will be very [h]ard pressed to provide any kind of similar review that is as in depth. In my opinion the only way to would be Stilt including a few more chips (a Carrizo APU perhap) and perhaps a bit more on the power consumption, but otherwise it compares Excavator to Zen to Hasewell to Kaby Lake. I think it, being a Desktop part, performed amazingly well against Intel's consumer grade and extreme grade parts! However, I suspect most people won't recognize/realize that unless his work gets passed around :) (As I had been turned to it from a different thread here on [H], but it pointed to a direct post much later in the thread)


Dan_D said:
I'm actually far more excited about Naples and what it means than I am about Ryzen. Naples is the product we need to shake Intel out of its apathetic state and further drive technology forward. The Opteron 64 did the same thing back in the day. It shamed the Xeon in a lot of areas and that's what we need again. For the enthusiast and gamer there is a side benefit. The first Athlon FX was basically a single socket capable Opteron and the CPU to beat when it was released. Naples has a much more compelling platform going with it which could be adapted to something we can use in a high end multi-GPU gaming system. Think about it, we could have a motherboard with 64 PCIe lanes, over 8 cores and 16 threads, and potentially a lot of freaking memory bandwidth if they set it up right. We could get at least four channels on a standard EATX board if not more. We could get a more modern feature set without DMI 2.0 limitations. In essence, this could spawn a very viable alternative and very worthy upgrade over X99 and Intel's dated HEDT platform.
Click to expand...
I'm right there with ya! Given that AMD built it in 4-core modules I suspect they could make a 12-core variant as well, but I also wonder exactly how feasible that'd be given the way they have the two quad-modules laid out...
This isn't the entire die package, just the cores package and their , but as it sits this is how they're laid out:
Ryzen-8C16T.png
As it sits, and as we're finding out due to how everything (software, BIOS, whatever) is coded, things appear to be treated as two "CPUs", or very much "SMP on a single package". Just like back when the first Dual Core rolled out and the OS wasn't quite sure how to treat the cores in order to properly utilize things.
Being the armchair CPU architect that I am... lol This is what I would've figured to be the more ideal layout:
Ryzen-12C24T.png

It'd have results in a much more square die structure instead of the current extremely-rectangle one, not that I figure that impacts thermal performance too much. Regardless, it also allows for much easier addition of more core-modules. So who knows how the Naples will be laid out in terms of the core modules, seeing as it'll no doubt be MCM again, as well as various other things.

It'll be interesting to see how they approach it from a socket standpoint. They did have 2 tiers in the past, with G34 and C32, but I know that there weren't a helluva lot of boards of either that targeted Workstations, particularly for Single CPU. I really wanted to rock a 12C Magny-Cours, but it was lack of motherboards that put the brakes on that idea. :p I'm really hoping they push for G34-level of single CPU boards, with at least quad-channel. I don't need it, I won't use it, but... me want! haha Which I think is the point, that's kind of the definition of "Enthusiast", be it computing or otherwise. We may not use something to its fullest capacity, but that won't stop us from wanting, and owning, just to still use it for our day to day tasks. :)


SimianRob said:
So why is Naples a "direct disaster" exactly? And I think it would be pretty hard for anyone to use something that hasn't launched yet.
Click to expand...
While you've been on the forums for a few years, you may not know, but it won't take you long to find this out. Took me only 4 or 5 days of continual reading in the AMD section that Shintai is, at the end of the day, someone that has some deep disdain for AMD. As I said in another thread, the accusations and things he says generally defy logic or are so wild, that I can't quite nail down if he genuinely thinks/feels that way or if he's your run of the mill troll that finds entertainment in starting flame wars. Regardless, the best advice I can give is to read what he says, roll your eyes if you need to, but not respond to him unless you plan to speak in facts. Never ask him a question though, unless it is to have him cite his source, as all it will do is cause him to spew more of the same. :(
DL;DR... "Don't engage the trolls." lol
 
Shintai said:
Are you trying to say that extremely informed and well run companies are buying Intel because its Intel? And not because they have beaten anyone else so much that its pointless and directly a bad business decision to buy anything else?

Pretty much nobody is doing Naples servers for the same reason, its a direct disaster. Not because of some sort of brand loyalty.
Click to expand...

That is something I want to ask. IBM and Intel have won several designs both private and goverment. Even ARM newcomers are getting designs and contracts. Does someone know a similar announcement for Naples?
 
Formula.350 said:
I'm right there with ya! Given that AMD built it in 4-core modules I suspect they could make a 12-core variant as well, but I also wonder exactly how feasible that'd be given the way they have the two quad-modules laid out...
This isn't the entire die package, just the cores package and their , but as it sits this is how they're laid out:
View attachment 18758
As it sits, and as we're finding out due to how everything (software, BIOS, whatever) is coded, things appear to be treated as two "CPUs", or very much "SMP on a single package". Just like back when the first Dual Core rolled out and the OS wasn't quite sure how to treat the cores in order to properly utilize things.
Being the armchair CPU architect that I am... lol This is what I would've figured to be the more ideal layout:
View attachment 18757
Click to expand...

It would be simpler to design a 6-core CCX and then join two to make a 12-core die... but leave that for Zen+.
 
I'd be interested in seeing a Naples based HEDT variant. If the memory bandwidth scaling have merit, there's more performance beyond dual channel DDR4 3200 to be had. A couple more memory channels sounds like what Zen needs... besides clock speed.
 
juanrga said:
That is something I want to ask. IBM and Intel have won several designs both private and goverment. Even ARM newcomers are getting designs and contracts. Does someone know a similar announcement for Naples?
Click to expand...

Back in 2016, AMD concluded a deal with China's THATIC rumored to involve Zen tech. This PCWorld article on Naples's launch includes this semi-quote.

AMD has already licensed its new server architecture to THATIC (Tianjin Haiguang Advanced Technology Investment Co. Ltd.), a joint venture in China that is making surrogate Zen chips for the local market. That doesn't mean AMD will hold its Naples chips from the China market, Norrod said.
Click to expand...
Norrod being Forrest Norrod, the guy presenting in the first post's video. But the statement does not say much what so ever, but thats the most I've seen on CPU partnerships recently. GPU server side, AMD touted a Google and Alibaba deal some time ago. The OCP Summit opening tomorrow will have two presentations by AMD, perhaps we can attain more tidbits from the conference.
 
Ehren8879 said:
I'd be interested in seeing a Naples based HEDT variant. If the memory bandwidth scaling have merit, there's more performance beyond dual channel DDR4 3200 to be had. A couple more memory channels sounds like what Zen needs... besides clock speed.
Click to expand...
Willing to bet that the server version won't operate that way. My thinking behind that is due to memory speeds and the fabric speed. It was probably done that way on the desktop for the reason of 1) Desktop memory runs faster 2) Limitation on the overclocks which would've resulted in further performance woes.

Servers will probably run with DDR4-2133 (or is FBDIMM only 1866? I don't pay attention to that), and the fabric will then be full speed.

But that leads me to an interesting thought. What if that IS the reason why Ryzen hits that overclock wall where it does, and it's not so much internal thermals or voltage (as Stilt had shown with his "Critical 1" and Critical 2" graph points), but because the Fabric.... wait nevermind.
I'm going to leave that just as food for thought, but I just realized the limit reach is on the CPU side, whereas the Fabric is based on the Memory speed. Now unless it's similar to the K10's ""Northbridge"" (memory controller) multiplier that increases mem bandwidth as that speed is increased, but it is now a multi that is locked with the CPU multi, that's the only way I can see that being the case. However, that's quite doubtful.

SighTurtle said:
Back in 2016, AMD concluded a deal with China's THATIC rumored to involve Zen tech. This PCWorld article on Naples's launch includes this semi-quote.

Norrod being Forrest Norrod, the guy presenting in the first post's video. But the statement does not say much what so ever, but thats the most I've seen on CPU partnerships recently. GPU server side, AMD touted a Google and Alibaba deal some time ago. The OCP Summit opening tomorrow will have two presentations by AMD, perhaps we can attain more tidbits from the conference.
Click to expand...
In before someone else makes the "lawl... Chinese knockoffs" joke. :cool:
Note how they opted to say "surrogate"... heh I suspect that was a last minute switch by the Editor haha

Will be interesting to see if someone can get their hands on one to compare to the variant made under AMD's supervision.
Unless these won't be clones, so much as the 'licensing' to let them create whatever they want with the architecture (ala ARM).
 
Last edited:
SimianRob said:
So why is Naples a "direct disaster" exactly? And I think it would be pretty hard for anyone to use something that hasn't launched yet.
Click to expand...
I read that as "if they end up not going with Naples, it'll be because it is a disaster rather than...", it didn't really feel like he was saying "Naples is a disaster, thus nobody will go with it."

...but I couldn't tell you for sure what he meant.
 
Also, 32 cores with hyperthreading (so, 64 threads)?! That's some serious shit, and I doubt it'll be cheap. Maybe less than Intels closest competing product, but I highly doubt it.
 
Ehren8879 said:
I'd be interested in seeing a Naples based HEDT variant. If the memory bandwidth scaling have merit, there's more performance beyond dual channel DDR4 3200 to be had. A couple more memory channels sounds like what Zen needs... besides clock speed.
Click to expand...

imagine the look on intels face if AMD releases a 16/32 cpu on a HEDT platform for less than a 6900k.

haha!

but then 1080p gaming...
 
Intel concluded a deal with Tsinghua University, and Montage Technology to develop data center infrastructure in China in January 2016. The THATIC(A investment arm of the Chinese Academy of Sciences)-AMD deal seems to be along the same lines as the earlier Intel deal. Anandtech has a good article on the AMD deal.
 
Master_shake_ said:
imagine the look on intels face if AMD releases a 16/32 cpu on a HEDT platform for less than a 6900k.

haha!

but then 1080p gaming...
Click to expand...


That would be great if it were unlocked, giving it another advantage over the xeons etc.
 
Clocks aren't that big of a deal in server space. You want cores running 'slow' so they consume less power and then you jack the shit out of the core/thread count. All the apparent weaknesses Ryzen has as a desktop chip (which may or may not be mitigated by updates) are basically gone in the server sector. Not saying that Naples will completely blow Intel away as we don't have any real silicon to look at yet to compare but I think AMD will be a lot closer to Intel in server than they are on the desktop.
 
  • Like
Reactions: atom
like this
There is alot to be had. If Naples is anything as good as it sounds. It will be a serious hit in server space. Think about it, Intel has been the only player with power and performance, with absolutely no competition from AMD. Naples will bring that, looking at zen core performance and how it matches up with intel in multithreaded test and you jack that shit up with some 8 channel memory this thing takes the same per core performance and blows it up. Hopefully this puts cash in to AMDs pocket, because this area is where the serious money is.
 
You know people are saying Naples is based on zen. At the core yes, but to me it seems like they started with a monster server chip and cut it down to make desktop chip. If naples has all those lanes and 8 channel memory, you can bet your end that slowly amd will be upping the desktop side with more lanes and 4 channel memory may be. This thing looks like a beast..
 
Intel might actually be in trouble with that thing. Much will depend on power and clocks, but I expect them to do well with it in the server space.
 
Gideon said:
Intel might actually be in trouble with that thing. Much will depend on power and clocks, but I expect them to do well with it in the server space.
Click to expand...

i don't think they're in trouble per say but there are definitely specialty markets where naples could potentially push intel out of which is what it sounds like AMD is targeting instead of the entire server market.. it'll just depend on if the performance to cost ratio is worth it for them to replace all their xeon based systems.
 
NKD said:
You know people are saying Naples is based on zen. At the core yes, but to me it seems like they started with a monster server chip and cut it down to make desktop chip. If naples has all those lanes and 8 channel memory, you can bet your end that slowly amd will be upping the desktop side with more lanes and 4 channel memory may be. This thing looks like a beast..
Click to expand...

It's the same chip as Summit Ridge.

Naples is four chips on on package. Each one is an 8-core Zen, with 2 memory channels and 32 PCIe lanes. That's how they get 8 memory channels and 128 PCIe lanes, using NUMA.

Desktop Zen has 20 PCIe lanes visible because that's all your average motherboard needs, WITHOUT BEING TOO EXPENSIVE TO ALSO HANDLE 8-LANE Bristol Ridge, and not having to switch to LGA to bring-up pin density.

But you can be sure those extra 12 leads are there on the die, just not connected to anything :D

Seriously, why would they do a separate core revision just to add more PCIe lanes, when NOT CONNECTING EXCESS I/O IN THE PACKAGE is one of the easiest things you can do with an existing CPU?
 
Last edited:
You must log in or register to reply here.
Back
Top