Monday, April 13th 2015

AMD "Zen" A Monolithic Core Design

AMD's upcoming "Zen" architecture will see a major change in the way the company designs its CPU cores. It will be a departure from the "module" core design introduced with "Bulldozer," in which two cores with shared resources constitute the indivisible unit of a multi-core processor. A "Zen" core will have dedicated resources in a way things used to be before "Bulldozer," and only the last-level cache (L3 cache), will be shared between cores. "Zen" will also implement SMT, much in the same way as Intel processors do, with HyperThreading Technology.

The first implementation of "Zen" will be an insanely powerful APU (on paper anyway), featuring 16 physical "Zen" CPU cores, 32 logical CPUs enabled with SMT, 512 KB dedicated L2 cache per core, and 32 MB of shared L3 cache. The CPU's ISA instruction set will see a spring-cleaning, with the removal of underused instruction-sets, and the introduction of new ones. Other features on this APU are equally surprising - a quad-channel DDR4 integrated memory controller, a separate HBM (high-bandwidth memory) controller dedicated to the integrated graphics, with up to 512 GB/s bandwidth, and an integrated graphics core featuring "Greenland-class" stream processors. Given that AMD is able to build 7-billion transistor GPUs on existing 28 nm processes, building an APU with these chops doesn't sound far-fetched. The company could still have to rely on a newer fab.
Source: FudZilla
Add your own comment

102 Comments on AMD "Zen" A Monolithic Core Design

#51
Aquinus
Resident Wat-man
Gentleman, brace yourself. This is going to be a new socket and with 64 PCI-E lanes (over 40 on skt2011) and quad-channel memory, we're looking at a massive size for this processor just to accommodate the pinning of such a CPU, which means AMD will have a lot of area to work with. The picture shows SATA and GBe as interfaces on the CPU which would also technically make this CPU a SoC, like the AM1 (except on crack). This may push AMD to adopt LGA sockets for consumer CPUs which I'm 100% for due to the number of pins on such a CPU.

I find this very interesting in general that it's an SoC and has a lot of functionality. Board layouts with a performance-level SoC should be *very* interesting to see. (Take a peak at Asus' C2750 workstation/nas board.) Since there would be no north bridge or pch-like chips, it allows for more area to do other things directly off PCI-E and not use the PCH/(chipset/SB) for PCI-E lanes which introduces latency.

All in all, I think this is great. I would love to see where this goes.
Posted on Reply
#52
erocker
*
HumanSmoke...and just like that it is like CMT never was.
Well.. they tried. lol.
Posted on Reply
#53
TheinsanegamerN
AquinusGentleman, brace yourself. This is going to be a new socket and with 64 PCI-E lanes (over 40 on skt2011) and quad-channel memory, we're looking at a massive size for this processor just to accommodate the pinning of such a CPU, which means AMD will have a lot of area to work with. The picture shows SATA and GBe as interfaces on the CPU which would also technically make this CPU a SoC, like the AM1 (except on crack). This may push AMD to adopt LGA sockets for consumer CPUs which I'm 100% for due to the number of pins on such a CPU.

I find this very interesting in general that it's an SoC and has a lot of functionality. Board layouts with a performance-level SoC should be *very* interesting to see. (Take a peak at Asus' C2750 workstation/nas board.) Since there would be no north bridge or pch-like chips, it allows for more area to do other things directly off PCI-E and not use the PCH/(chipset/SB) for PCI-E lanes which introduces latency.

All in all, I think this is great. I would love to see where this goes.
consumer cpus have been lga for over 20 years. im assuming you meant BGA, which would never happen. remember the backlash intel got from that? AMD wouldnt weather such backlash nearly as well.
Posted on Reply
#54
librin.so.1
TheinsanegamerNconsumer cpus have been lga for over 20 years. im assuming you meant BGA, which would never happen. remember the backlash intel got from that? AMD wouldnt weather such backlash nearly as well.
He did mean LGA. Since AMD CPUs, unlike Intel's, are/were mostly PGA till now.
Posted on Reply
#55
TheinsanegamerN
VinskaHe did mean LGA. Since AMD CPUs, unlike Intel's, are/were mostly PGA till now.
oh yeah. whoops. forgot PGA was different from LGA.
Posted on Reply
#56
rruff
Freedom4556Not to get anybody's hopes dashed, but the instant I read "16 cores" and "current process" I immediately assumed we're looking at a game console/server-esque sub-2GHz bulk process affair. This just sounds like a double-sized, updated PS4 APU.
I'm underwhelmed by massive numbers of threads also. If it's a "current process" and reasonable power consumption, then it will be slow.

I guess that is where we are headed though, now that progress in single core speed has stalled. It sucks for computations that can't be done in parallel.
Posted on Reply
#57
MikeMurphy
Petey PlaneAMD been selling 16 core chips for at least 3 years already
For servers. The huge cache and HBM also suggest servers. No doubt they included a small GPU for functionality, and therefore called it an "APU".

I'm not sure why everyone is getting their panties in a twist over server chips.
Posted on Reply
#58
FordGT90Concept
"I go fast!1!11!1!"
Because a server chip wouldn't have up to 16 GB of dedicated VRAM.
Posted on Reply
#59
HumanSmoke
Petey PlaneAMD been selling 16 core chips for at least 3 years already
Factually, 16 core Opteron is no more a single chip than Magny Cours, Core 2 Quad, or Pentium D.
It is two chips bound on a common substrate, under a common heatsink, in an MCM package.
AquinusGentleman, brace yourself. This is going to be a new socket and with 64 PCI-E lanes (over 40 on skt2011) and quad-channel memory, we're looking at a massive size for this processor just to accommodate the pinning of such a CPU
With the pin count and the fact that it has HBM on-package, I would sincerely doubt that this would be a conventional socketed processor. More likely the CPU + HBM would be an interposer module, just as the GPU+HBM are expected to be.
MikeMurphyFor servers. The huge cache and HBM also suggest servers. No doubt they included a small GPU for functionality, and therefore called it an "APU".
That would be my estimation also. The GPU would undoubtedly be for GPGPU ( co-processor) duty - a fairly conventional answer to Xeon Phi or Nvidia's planned Volta + ARM product lines.
Posted on Reply
#60
Renald
giving the nature of the source, and how powerful this proc seems to be (server-grade crearly), I think that it could be possible that this supposed leaked is a fake.

It could be :laugh:
Posted on Reply
#61
newtekie1
Semi-Retired Folder
People really need to realize that Zen is an architecture. It will be used for BOTH server and desktop CPUs.
Posted on Reply
#62
JMccovery
The features and rumored TDP lead me to believe that this will be a BGA design, similar to Xeon D. I wonder exactly how many 'cores' the GPU will have. 20/16nm could make for something with a shader count that could be close to a 7850 or 7870/XT.

Hmm... 50-100% more shaders than the PS4's GPU, using GCN 1.3+ tech with half-rate DP, 16GB HBM with a 512GB/s interface, 16 CPU cores/32 threads, 4 of these on a liquid-cooled blade with 64/128GB of DDR4-3200 per chip, 16 blades per 4U cabinet, 6-8 cabinets per rack ... Holy HPC servers, Batman!

I doubt this will be built on anything older than GF or Samsung 20/16nm process.
Posted on Reply
#63
Aquinus
Resident Wat-man
HumanSmokeWith the pin count and the fact that it has HBM on-package, I would sincerely doubt that this would be a conventional socketed processor. More likely the CPU + HBM would be an interposer module, just as the GPU+HBM are expected to be.
I was expecting this to be a multi-die CPU, that's not my line of reasoning. If you have skt2011 where the vast majority are used simply for quad-channel memory and 40 PCI-E lanes. I suspect even after the CPU itself is complete, the socket is going to at least be as massive as skt2011 is now. Don't forget the connections for the onboard SoC parts such as SATA and GBe.

Consider AM1 with 722 contacts. That's only for 8? PCI-E lanes, Gigabit, 2 sata ports, graphics output, and only 1 memory channel. Now compare that is a beefy SoC. 60 PCI-E lanes, 2 GBe maybe, ~6 SATA, 4 DDR4 memory channels, and maybe a little more for display output. Seriously, I think 2000 contacts is going to be optimistic. I wouldn't be surprised if it was closer to 3k if this is legit.

With that said, I'll sit back, wait, and see. Anything to add some competition in the market.

Edit: It does seem that PCI-E and SATA share the same circuitry, so more SATA means less PCI-E and more PCI-E means less SATA. I could live with that.
Posted on Reply
#64
Captain_Tom
newtekie1Yep, and they've been doing it on 32nm too...with 115w TDPs...
The bulldozer series really was made for servers where it was right at home. Unfortunately this was at the expense of EVERY other market.
Posted on Reply
#65
Captain_Tom
RealNeilI remember the advertising hype just before Bulldozer came out.

AMD's Ad guys had everyone worked up into a frenzy with a lot of fantastic, dubious claims about its performance.
Then, It came out and everyone was underwhelmed and pissed off about it.
They took a huge PR hit then, and not long afterwards, they fired most of the Ad guys responsible.

I think that they learned from that and have toned down the rhetoric since then.

Me, I'm now cautious about new releases. I want to see reviews, and not just one of them either.
I want to read about other's experiences with brand new tech before I spend money getting into it.
I don't care if I'm the first one to have something anymore.

Zen ~sounds~ really good, but show me the reviews first.
As excited as I am for the possibilities of Zen I think almost all of us feel this way deep down. Zen sounds amazing, but no true enthusiast buys blindly.
Posted on Reply
#66
RejZoR
Well, the fact they are aware of the Bulldozer modules and problems it had explains a lot.

Btw, if I understand this correctly, AMD will have their own version of hyperthreading this time around? I love this thing with Intel CPU's. It's the reson why my ancient Core i7 920 is still so competitive. It churns out 8 threads and that still works incredibly well with file compression, video coding, audio conversion, large image processing etc. I know it's not like a full core on it's own but it certainly makes a huge difference.
Posted on Reply
#67
Relayer
qubitThis sounds great and I'm not completely surprised, since the new guy at the top apparently has turned around other businesses.

I'd love to see an AMD CPU/APU kick Intel's butt once more like they did a decade ago.

Let's hope this really delivers this time.
The "Guy" at the top is named Lisa. :pimp:
wikipediaCareer[edit]
In 1994 Su was a member of the technical staff at Texas Instruments, in its Semiconductor Process and Device Center (SPDC).[6][7]

She spent 1995 to 2007 at IBM in engineering and business management positions, including vice president of the Semiconductor Research and Development Center responsible for the strategic direction of IBM’s silicon technologies, joint development alliances, and semiconductor R&D operations.[6]

Su joined Freescale Semiconductor in 2007 as chief technology officer, and she led the company’s technology roadmap and research and development efforts. From September 2008 she was the senior vice president and general manager of Freescale's Networking and Multimedia group, and was responsible for global strategy, marketing, and engineering for the company’s embedded communications and applications processor business.[6][7]

Su became senior vice president and corporate director at AMD in January 2012, and chief operating officer in June 2014.[8][6] She was appointed president and CEO of AMD in October 2014, replacing Rory Read.[5]
The "GUY" that was CEO before her was Rory Read. While he was at AMD everything he touched turned to salt. :shadedshu:
Posted on Reply
#68
Kyrios74
I wish Jim Keller to do a good job and to bring back the days of AMD k7 and k8
Posted on Reply
#69
Dent1
Kyrios74I wish Jim Keller to do a good job and to bring back the days of AMD k7 and k8
How do you suggest he does that?

K7 was over 15 years ago and things change, the economy was different, banks were lending more, interest rates were higher, tech companies were showered with sponsorship money and loan money, there was the big technology bubble, there was no big internet tech websites, you had to buy specialist magazines to learn about upcoming products and technologies. Desktop machines were selling like hotcakes and nobody anticipated mobile marketing taking off. Keller can't snap his fingers and rewind time because the world was different back then.
Posted on Reply
#70
Kyrios74
hai ragione. Il mio era un modo per esorcizzare i tempi.
Posted on Reply
#71
HumanSmoke
Kyrios74I wish Jim Keller to do a good job and to bring back the days of AMD k7 and k8
K7 and K8 owed much to AMD's purchased IP from DEC ( just as K6 before them was the result of purchased IP from NexGen) and the people that worked on the DEC Alpha processors (including a certain Dirk Meyer) and Fred Weber. AMD doesn't have that luxury this time around.
Dent1Desktop machines were selling like hotcakes and nobody anticipated mobile marketing taking off.
I wouldn't say that. Intel was pushing mobile computing and looking ahead to smart mobile marketseven while nailing the coffin closed on NetBurst back before the turn of the new century. IIRC, Intel unilaterally scrapped the NetBurst based Timna and a few other projects and set the design teams to design low-power mobile processing (Banias/Pentium M)
Posted on Reply
#72
cyneater
I wonder if this will be like the Optron Arm processors that where meant to come out in 2014 ....

Aka its just some marketing hype and there will be no product that is made.
Posted on Reply
#73
HumanSmoke
cyneaterI wonder if this will be like the Optron Arm processors that where meant to come out in 2014 ....
Aka its just some marketing hype and there will be no product that is made.
AMD have already confirmed that the Zen architecture is x86 based. The ARM (ARMv8-A) architecture is K12.
As for marketing hype, I doubt this came from AMD - more likely clickbait from the usual suspects - and bearing in mind Faud's (Fudzilla) strike rate, I'd be more than a little wary of its veracity.
Posted on Reply
#74
Dent1
Kyrios74hai ragione. Il mio era un modo per esorcizzare i tempi.
Va bene mio fratello itallian lol
Posted on Reply
#75
RealNeil
Captain_TomAs excited as I am for the possibilities of Zen I think almost all of us feel this way deep down. Zen sounds amazing, but no true enthusiast buys blindly.
I build PCs for friends. I try to match what I build to the needs that they express to me before I do their builds. As often as I can, I make AMD CPUs a part of the build. I believe that AMD needs to survive, just to keep Intel's prices from running amuck.
The choices that we have, help us out in the long run. I also like to use AMD GPUs.

AMD is necessary in this market.
Posted on Reply
Add your own comment
Dec 24th, 2024 09:16 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts