AMD Unveils 64-Bit ARM-Based Opteron A1100 System On Chip With Integrated 10GbE (hothardware.com)

← Back to Stories (view on slashdot.org)

AMD Unveils 64-Bit ARM-Based Opteron A1100 System On Chip With Integrated 10GbE (hothardware.com)

Posted by timothy on Thursday January 14, 2016 @04:54AM from the paradigm-shift dept.

MojoKid writes: AMD is adding a new family of Opterons to its enterprise processor line-up today called the Opteron A1100 series. Unlike AMD's previous enterprise offerings, however, these new additions are packing ARM-based processor cores, not the X86 cores AMD has been producing for years. The Opteron A1100 series is designed for a variety of use cases and applications, including networking, storage, dense and power-efficient web serving, and 64-bit ARM software development. The new family was formerly codenamed "Seattle" and it represents the first 64-bit ARM Cortex-A57-based platform from AMD. AMD Opteron A1100 Series chips will pack up to eight 64-bit ARM Cortex-A57 cores with up to 4MB of shared Level 2 and 8MB of shared Level 3 cache. They offer two 64-bit DDR3/DDR4 memory channels supporting speeds up to 1866 MHz with ECC and capacities up to 128GB, dual integrated 10Gb Ethernet network connections, 8-lanes of PCI-Express Gen 3 connectivity, and 14 SATA III ports. AMD is shipping to a number of software and hardware partners now with development systems already available.

98 comments

Min score:

Reason:

Sort:

For the love of... by DivineKnight · 2016-01-14 05:00 · Score: 1

You'd think they could have at least upgraded some of their x86 stock offerings to PCIe 3.0, but no, that'll have to wait...
1. Re:For the love of... by ArmoredDragon · 2016-01-14 05:05 · Score: 1
  
  "Enterprise is a much bigger market" is what I'd otherwise say, but seeing as that (and datacenters) runs on vmware, I doubt this will see that much adoption. Unless of course the value prop is so good that open source virtualization (I.e. xen) starts running on it.
2. Re: For the love of... by Billly+Gates · 2016-01-14 05:27 · Score: 1
  
  These are great for ISPs, Google, and Amazon where power saving is important. Think nodes in clusters. Server 2012R2 has an arm port and has hyper-v too.
  Power costs are astronomical in these environment's
  
  --
  http://saveie6.com/
3. Re:For the love of... by Anonymous Coward · 2016-01-14 05:31 · Score: 1
  
  There is no Windows x86 support except for Wine support, which they say is slow, so I doubt that you will see any Windows VMs running on it
  However, there is support for Linux, Android and iOS, which will probably be the primary targets
  There are probably plenty of customer who would choose this over Xeon for datacenters which do not need to serve a win MS-centric market
4. Re: For the love of... by subk · 2016-01-14 05:50 · Score: 1
  
  These are great for ISPs, Google, and Amazon where power saving is important. Think nodes in clusters. Server 2012R2 has an arm port and has hyper-v too.
  Power costs are astronomical in these environment's
  It will also be great for more-or-less embedded applications like firewalls, storage arrays, de-duplicators, media server/transcoders, etc where you need a healthy CPU and great ethernet bandwidth. I think they on the right path here in terms of an efficient platform for SoC-based appliance manufacturers, where watt-hungry dual and quad-core Intel chips have been the de facto choice. Atom provided an option for those craving a more efficient chip, but IMHO, while it works in clusters, it never performed well enough to stand alone. It may seem like a bold maneuver to some, it is not like they are proposing a new (or even "dead") architecture--ARM is well established and it should be little sweat for developers to adopt. I would love to see this work out for AMD, it is a decent strategy seeing as how they haven't been able to catch up to Intel in the desktop market for quite some time.
  
  --
  Now, if you'll excuse me, I have backups to corrupt.
5. Re: For the love of... by Anonymous Coward · 2016-01-14 06:01 · Score: 0
  
  I don't know about these specific ones, but other 'server grade' arm systems I have messed with actually compare poorly performance per watt with the Intel stuff. Things may be different in the ultra low power segment, but despite using almost as much power as the Intel chips, the performance is disproportionately terrible for the workloads I tried.
6. Re: For the love of... by Billly+Gates · 2016-01-14 06:28 · Score: 2
  
  Still there is a common misconception that all servers are CPU bound.
  Not true for databases or Java servlet apps. It is latency based. Latency on networks and disk i/o as your SQL query gets processed and the CPU goes on massive WAIT cycles in the process.
  An ARM will be great for lower power with an SSD and lots of bandwidth.
  
  --
  http://saveie6.com/
7. Re: For the love of... by fuzzyfuzzyfungus · 2016-01-14 07:23 · Score: 1
  
  Could well be useful for memcache-type applications. Not likely to be terribly powerful; but quite possibly one of the cheaper ways to get 128GB of ECC RAM into a small box(Intel's C2000 Atom-for-server stuff currently tops out at 64GB, and their Xeons cost more).
8. Re:For the love of... by bored · 2016-01-14 08:47 · Score: 1
  
  As much as I think that this is probably the "best" ARM server at the moment, I 100% agree with you. Its not going to make AMD any real money, and the amount of R&D invested would have been better spent upgrading their existing product lines.
9. Re:For the love of... by ArmoredDragon · 2016-01-14 09:17 · Score: 1
  
  I don't think iOS would ever run on it. Apple would shit bricks and summon a demon hoard army of lawyers on anybody who runs it on anything except for their in-house designed ARM SoC's.
10. Re:For the love of... by hairyfeet · 2016-01-14 10:34 · Score: 0
  
  That is coming with Zen which will be coming along with a new socket which will allow them to combine the APU and CPU lines along with a completely new architecture designed by the same guy who made the Athlon64
  So there really is no point in releasing a new chip until the new socket boards by the OEMs are ready to go, especially not just for PCIe 3 which test after test has shown has a completely negligible effect on performance even with 3 way crossfire. The simple fact of the matter is we have yet to saturate PCIe 2 yet and with AAA titles costing so much to make AND the fact we have just started on a new console generation that came out of the gate with lower performance than midrange gaming PCs? I seriously doubt we'll see PCIe 3 used as anything but a bullet point on a box for the foreseeable future, its just not needed.
  
  --
  ACs don't waste your time replying, your posts are never seen by me.
11. Re: For the love of... by LinuxIsGarbage · 2016-01-14 11:13 · Score: 1
  
  Server 2012R2 has an arm port and has hyper-v too.
  Is there any proof of this outside of internal Microsoft test versions? As far as I can tell the failed tablet OS, Windows RT, is the closest.
12. Re:For the love of... by LinuxIsGarbage · 2016-01-14 11:26 · Score: 1
  
  However, there is support for Linux, Android and iOS, which will probably be the primary targets
  There are probably plenty of customer who would choose this over Xeon for datacenters which do not need to serve a win MS-centric market
  How many people use iOS in a datacenter? How many people use iOS on third party hardware?
  On Linux you'd still get better hardware/software support on x86, and better performance/ dollar, or performance per watt.
  I could certainly see this maybe in embedded applications, and maybe networking appliances (like a storage server, or maybe a web server). But I think it will be a few years still before you see a big foothold of ARM based CPUs in general purpose servers.
13. Re:For the love of... by Anonymous Coward · 2016-01-14 16:33 · Score: 0
  
  Why would they bother? Is anyone going to pair a fast video card with one of their slow CPUs? They should start by making their CPUs not suck first.
14. Re:For the love of... by Anonymous Coward · 2016-01-14 20:47 · Score: 0
  
  Why not not that many games are cpu bound and it would make sense to spend as little as possible on cpu/motherboard to buy a bigger gpu.
  Cut 100USD from cpu and put a 100USD more on gpu is generaly a better idea for good gaming performance.
15. Re:For the love of... by MoarSauce123 · 2016-01-16 00:42 · Score: 1
  
  Keep in mind that their x86 offerings are half the price compared to the equal Intel offerings, something gotta give.
16. Re: For the love of... by TheRealLifeboy · 2016-01-16 01:32 · Score: 1
  
  Just because MS is noisy and nosey (requiring registration of their products) doesn't mean that they're the fastest growing or most popular overall. Ubuntu Linux is running on more than a billion devices already and that's a conservative estimate. Also, the other virtual machines solutions are fast, growing and free. Think Proxmox, LCX and more. Hey, even MS have embraced a lot of this, so it's time to rethink your assumptions, like you have with your Xen closing remark.
  Below are some Ubuntu figures, check for the number of VM's fired up every minute. Seems AMD have a clue about what's happening in the market after all
  http://blog.dustinkirkland.com/2015/12/more-people-use-ubuntu-than-anyone.html
Only X8 pci-e? by Joe_Dragon · 2016-01-14 05:00 · Score: 3, Insightful

Only X8 pci-e?
At least they have dual 10-gige but come on give at least x16 pci-e even if you need to cut down the sata links.
1. Re:Only X8 pci-e? by CastrTroy · 2016-01-14 05:09 · Score: 2
  
  For what purpose? These are obviously meant for the server room. What need do you have for anything beyond pci-e X8 if you don't even have a video card?
  
  --
  
  Anthropic principle: We see the universe the way it is because if it were different we would not be here to see it.
2. Re:Only X8 pci-e? by Anonymous Coward · 2016-01-14 05:17 · Score: 2, Interesting
  
  I've got a hunch this will be more aimed at "lots and lots of dies in one small box" applications rather than as part of a large monolithic system. (Think virt host)
  You only need enough I/O to attach a few disks or a controller to attach to a SAN or something like infiniband and x8 is plenty. You could attach a single GPGPU as well but I don't think that would be a target application. (Does Nvidia even have Tesla support for ARM platforms?)
3. Re:Only X8 pci-e? by chuckymonkey · 2016-01-14 05:17 · Score: 2
  
  Some servers do have video cards. We have them in lots of our servers for GPU compute tasks.
  
  --
  "Some books contain the machinery required to create and sustain universes."-Tycho
4. Re:Only X8 pci-e? by Anonymous Coward · 2016-01-14 05:24 · Score: 0
  
  Using GPU's for compute tasks is a hacky workaround to having dedicated high speed CPU's available as first class devices on the main system bus. Can't blame AMD for not designing to a market that has no real growth potential.
5. Re:Only X8 pci-e? by DRJlaw · 2016-01-14 05:30 · Score: 1
  
  You only need enough I/O to attach a few disks or a controller to attach to a SAN or something like infiniband and x8 is plenty.
  Storage controllers tend to be x8 devices, meaning that you'd have to have 2 x8 expansion slots to enable redundant controllers in your small box.
  Rumor has it that redundant controllers is, indeed, a thing in that market.
6. Re:Only X8 pci-e? by Anonymous Coward · 2016-01-14 05:31 · Score: 0
  
  but but.. nvidia ceo said deep learning needs GPUs!!!! Deep learning!.... neeed! nooow! the economz of NVidia depends on it
7. Re:Only X8 pci-e? by mrchaotica · 2016-01-14 05:39 · Score: 2
  
  So would your GPU compute servers benefit from having an ARM instead of x86 CPU?
  Clearly, there is some market where the workload favors ARM CPUs, and some market where the workload favors GPU computing (and therefore PCIe x16). Combining both in the same product only makes sense if there's some market that needs both at the same time. It's not as if AMD were discontinuing x86 Opterons, after all.
  
  --
  "[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
8. Re: Only X8 pci-e? by Anonymous Coward · 2016-01-14 05:41 · Score: 0
  
  Last year I took a parallel processing university class. We reworked the same computation problem for desktop clusters, a multicore/multinode supercomputer and GPUs. For certain problem classes, the gpu can be superior.
9. Re:Only X8 pci-e? by Anonymous Coward · 2016-01-14 05:43 · Score: 1
  
  You'd need something like that for a dedicated storage node but I don't think that's the target application for this chip. This isn't a big monolithic server chip.
  If you look at modern server farms you have endless rows of boxes that are as simple and as cheap as possible. Usually just a power supply, board, cpu, memory, 2x ethernet, and 2 cheap commodity SATA hard drives.
  If anything on the node fails it's taken offline. The "cloud fabric" along with object based storage make s sure the data has already been duplicated elsewhere. You're not going to attach dual redundant storage controllers, each of which cost more than the host system, to a low cost SoC server chip.
  This SoC has most of what it needs built right. The x8 pci express is just some extra I/O so OEMs can build systems with some flexible options for slightly different applications. More Ethernet, more, storage controllers, maybe something to connect to a SAN, some other thing they want to stick on the motherboard like a custom system management controller. If the chip ends up being put on to a generic motherboard for OEMs or home servers/hobbiests the lanes are useful for peripheral slots and extra SATA controllers.
10. Re:Only X8 pci-e? by Anonymous Coward · 2016-01-14 05:56 · Score: 0
  
  What are you going to use them for?
  PCIe SSD boot drive.
  Additional NIC perhaps if you need more than two. Or you might want a fabric interconnect instead.
  SATA is not an issue for bulk storage uses (non-SSD).
11. Re:Only X8 pci-e? by Junta · 2016-01-14 06:04 · Score: 1
  
  While this is presumably a dig at AMD APUs, the fact is that GPUs greatly outclass even the most powerful CPUs for a lot of tasks. The next time the question is likely to be worth re-examining will be when Xeon's start getting AVX-512 extensions.
  
  --
  XML is like violence. If it doesn't solve the problem, use more.
12. Re:Only X8 pci-e? by Junta · 2016-01-14 06:06 · Score: 1
  
  Well, maybe if the ARM servers did NV-Link. POWER and ARM can do NV-Link CPU to GPU, Intel however was either not invited to the party or was disinterested in working to enable it. Of course I'm suspecting AMD and nVidia collaborating would be unlikely
  Again though, there'd be no room to complain about PCIe lanes.
  
  --
  XML is like violence. If it doesn't solve the problem, use more.
13. Re: Only X8 pci-e? by silas_moeckel · 2016-01-14 06:21 · Score: 1
  
  How may of those use cases also requires a lot of bandwidth to the GPU board itself? A 16 lane gen 3 has less bandwidth than a dingle stick of ddr3 1600 memory. So a GPU in a 1 lane 16 lane physical slot will probably works just as well for most of the GPU use cases. Mind you these chips are pretty anemic with only 2 memory channels and 128gb max ram so the big mem use cases would seem not a great fit anyways.
  From the layout this thing seems like a great small SAN/NAS controller, enough CPU grunt etc to get a 1 RU of SFF drives running ZFS etc. Enough RAM and network IO as well.
  
  --
  No sir I dont like it.
14. Re:Only X8 pci-e? by mrchaotica · 2016-01-14 06:21 · Score: 2
  
  NV-Link just sounds like an Nvidia copy of Hypertransport.
  
  --
  "[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
15. Re:Only X8 pci-e? by Junta · 2016-01-14 06:34 · Score: 1
  
  Roughly, it is. However that has never been used to connect a discrete GPU to a CPU, to my knowledge. AMD did it first, Intel copied it, and nVidia has joined the world of proprietary interconnects, but the first to do so for CPUGPU situation.
  
  --
  XML is like violence. If it doesn't solve the problem, use more.
16. Re:Only X8 pci-e? by DRJlaw · 2016-01-14 06:37 · Score: 1
  
  I don't disagree with what you're saying, but the type of node that you describe does not need dual integrated 10Gb Ethernet network connections and 14 SATA III ports. The dual 10Gb ethernet ports alone indicate a a combination of high I/O throughput and redundancy, not merely an application node.
17. Re:Only X8 pci-e? by TheRaven64 · 2016-01-14 06:50 · Score: 1
  
  Storage controllers tend to be x8 devices
  8x PCIe v3 gives you 7,880MB/s. Even just throwing the data out over the two 10GigE connectors, you'll only use a third of that. I doubt that 8 Cortex-A57 cores are going to find themselves data starved processing the rest.
  
  --
  I am TheRaven on Soylent News
18. Re:Only X8 pci-e? by wbr1 · 2016-01-14 07:01 · Score: 2
  
  GPUs for compute tasks do not need the bandwidth that GPUs pushing videos do. Look at and old mining rig using GPUs, you get 16x to 1x cables and stack your compute GPUs.
  
  --
  Silence is a state of mime.
19. Re:Only X8 pci-e? by Holi · 2016-01-14 07:23 · Score: 1
  
  Actually this sounds exactly like a storage node. Definitely not a VM host as it does not have nearly enough cores. It sounds like a perfect low cost nas for a vm cluster.
  
  --
  Sorry, teleporters just kill you and then make a copy. A perfect, soul-less copy.
20. Re:Only X8 pci-e? by mrchaotica · 2016-01-14 07:53 · Score: 1
  
  I think AMD uses Hypertransport in its APUs, but yeah, I've been wishing for years for an AMD FX (or Phenom II) + discrete Radeon setup that used it (maybe with the GPU in a motherboard socket instead of a daughter card, although I guess the hard part of that is the lack of standards for slotted graphics RAM).
  
  --
  "[Regarding the 'cloud,'] ownership was what made America different than Russia." -- Woz
21. Re:Only X8 pci-e? by afidel · 2016-01-14 08:26 · Score: 1
  
  LOL, GPU's are just the latest version of the vector coprocessor, something that's been at the heart of supercomputers pretty much since the start and which is absolutely needed to achieve good performance on certain classes of problems.
  
  --
  There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
22. Re:Only X8 pci-e? by amorsen · 2016-01-14 08:36 · Score: 1
  
  If you are doing GPU compute, you are unlikely to be interested in ARM. Those GPU's need to be fed somehow with data somehow, and the ARM won't keep up.
  We are still not at the point where GPU's can just fetch a bunch of data over the network by themselves, crunch it, and send it out again.
  
  --
  Finally! A year of moderation! Ready for 2019?
23. Re:Only X8 pci-e? by amorsen · 2016-01-14 08:38 · Score: 1
  
  Mining is very niche though. Very few workloads are so embarrassingly parallel, and it's usually worth it to use an FPGA or (ideally) an ASIC for those.
  
  --
  Finally! A year of moderation! Ready for 2019?
24. Re:Only X8 pci-e? by wbr1 · 2016-01-14 10:37 · Score: 2
  
  Still for rendering tasks, a large part of the bus is needed for moving large textures in and out of memory. While I am not an expert in this field, I do not think may of the other current GPU compute tasks require this memory bandwidth, so as large a bus is not needed.
  
  --
  Silence is a state of mime.
25. Re:Only X8 pci-e? by LinuxIsGarbage · 2016-01-14 11:33 · Score: 1
  
  I think more importantly they are a mass-marketed co-processor, which benefits from volume sales, so the unit cost is fairly low.
26. Re:Only X8 pci-e? by Anonymous Coward · 2016-01-14 12:00 · Score: 0
  
  and AMD copied the idea from... [long list of "high bandwidth interconnects" goes here]. i at least remember NUMAlink on SGI Origin machines in the mid-90s. that at least predates Hypertransport by 5+ years...? there's piggy-backing and leap-frogging in this industry all the time.
27. Re:Only X8 pci-e? by dbIII · 2016-01-14 15:30 · Score: 1
  
  They do if your task requires using a lot more memory than the cards have onboard. At that point as much bandwidth as you can get never seems like enough.
28. Re:Only X8 pci-e? by dbIII · 2016-01-14 15:33 · Score: 1
  
  I do not think may of the other current GPU compute tasks require this memory bandwidth
  Chicken/egg situation - there are compute tasks not considered for GPUs due to not having the memory bandwidth.
29. Re:Only X8 pci-e? by Perky_Goth · 2016-01-14 15:34 · Score: 1
  
  To be fair, it's a lot faster with a GPU.
30. Re:Only X8 pci-e? by thegarbz · 2016-01-14 20:01 · Score: 1
  
  Some people select their hardware for the application. This is not the server for you. Not even remotely.
31. Re:Only X8 pci-e? by drinkypoo · 2016-01-15 02:16 · Score: 1
  
  Combining both in the same product only makes sense if there's some market that needs both at the same time.
  While it might seem silly to try to save ~100W in a system that contains ~1000W of GPUs, that still adds up to another whole node relatively rapidly. If you're using pretty much all GPU and basically no CPU, it might make good sense. Isn't that a fairly common case now? With the 10GbE they are attractive cluster nodes. It might not have the crazy low latency of some of the more specialized interconnects, but it's still damned fast.
  
  --
  "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Price by Anonymous Coward · 2016-01-14 05:10 · Score: 0

I wonder how they'll be priced. Presently a dual port 10GE NIC runs over $300.
10GbE isn't that interesting by slaker · 2016-01-14 05:11 · Score: 2, Informative

10GbE Ethernet, (at least over copper, which is the only way I've gotten to mess with it), kinda sucks. Cost per port is really high and actually so are the power requirements per port. Infiniband was a lot easier and cheaper for me to deal with and having it implemented in relatively common hardware might improve its adoption.

--
-- I wanna decide who lives and who dies - Crow T. Robot, MST3K
1. Re:10GbE isn't that interesting by Anonymous Coward · 2016-01-14 05:22 · Score: 1
  
  I desperately want an affordable home switch/router with a single 10Gb port. My server can feed around 400MB/s from its raid array but the single 1Gb ethernet port is the limiting factor.
2. Re:10GbE isn't that interesting by Anonymous Coward · 2016-01-14 05:36 · Score: 1
  
  Unless you've got a 10GE client, it would be far more cost effective to add a few 1Gbps ports to your server and bond the interafces. There are lots of cheap Gbit switches capable of LAG/LACP and any server OS should be capable.
3. Re:10GbE isn't that interesting by PeeAitchPee · 2016-01-14 05:39 · Score: 2
  
  That's cuz you're not doing it right. Having dual 10GbE ports on each end will let you run SMB3 Multichannel for network transfer speeds that will outpace anything but the fastest RAID arrays. We're seeing real-world file transfer speeds of over 1.3 gigabytes (not gigabits) / second over copper Ethernet. I'm seldom a Microsoft advocate but it is awesome.
4. Re:10GbE isn't that interesting by Gravis+Zero · 2016-01-14 05:51 · Score: 1
  
  10GbE Ethernet, kinda sucks. Cost per port is really high and actually so are the power requirements per port.
  that's why it's built into the processor, you twit.
  
  --
  Anons need not reply. Questions end with a question mark.
5. Re:10GbE isn't that interesting by hattig · 2016-01-14 05:59 · Score: 2
  
  The A1170 high end Opteron has a 32W TDP, and two built-in 10GigE ports, for under $150 (expected SoC price).
  So how is that expensive per port, in terms of power consumption, etc?
6. Re:10GbE isn't that interesting by Junta · 2016-01-14 06:13 · Score: 2
  
  Cost per port on the switch side and power requirements on the switch side I presume he means. Particularly if he's talking about CAT6 rather than DAC, it is a pretty huge power hog. Note that in relatively recent developments a wave of PHYs have come about that significantly improve that, but it's still pretty big.
  Now using DACs should bring the power complaint in line with infiniband, though the runs can't be very long by comparison, but then again 'cheap' Infiniband cables can't be that long either (which are just common with 40 GbE assuming we are talking about QDR, FDR, or EDR IB).
  
  --
  XML is like violence. If it doesn't solve the problem, use more.
7. Re:10GbE isn't that interesting by Anonymous Coward · 2016-01-14 06:17 · Score: 0
  
  No one is going to bake infiniband in to their SoC.
  Ethernet is commodity, free of licensing costs, and cheap to implement. It'll shlep to the switch/concentrator/whatever on the rack (or the one on the end of the row) well enough. If you want infiniband they included x8 PCIE3 so you can attach a card (Or tie it to a controller soldiered on to the motherboard)
8. Re:10GbE isn't that interesting by slaker · 2016-01-14 06:26 · Score: 1
  
  I've read 5 - 7W per port. That's far from inconsequential.
  
  --
  -- I wanna decide who lives and who dies - Crow T. Robot, MST3K
9. Re:10GbE isn't that interesting by slaker · 2016-01-14 06:31 · Score: 1
  
  You are correct in your presumption. I was not aware there's been any real improvement WRT Cat6 PHYs. Thanks.
  
  --
  -- I wanna decide who lives and who dies - Crow T. Robot, MST3K
10. Re:10GbE isn't that interesting by jon3k · 2016-01-14 07:38 · Score: 1
  
  Look into 10Gb of direct attach copper SFP. Low power and low port cost (at least compared to Infiniband).
  
  10Gb port costs have come WAY down because of all the competition in the space (read: Arista) fighting with the established players (Cisco, Juniper). You can get a 48 port SFP based Nexus 9732PQ for around $13k (with multiple quad SFP ports). You don't need optics because you just use direct attached cables. Dual port 10Gb SFP based Intel NICs are around $300 now.
  
  Not sure what you mean by "implemented in relatively common hardware" because there are dozens of off the shelf options for NICs and switches. There's 10Gb Lan-on-motherboard, PCIe NICs, blade mezzanines, you name it. Plus inexpensive 10Gb switches from every major vendor (Cisco, Juniper, HP, Arista, etc).
11. Re:10GbE isn't that interesting by jon3k · 2016-01-14 07:41 · Score: 2
  
  There's lots of ways to aggregate multiple NICs (ie LACP). It's very easy to bond multiple 10Gb ports to increase bandwidth. And I've gotten comfortably over 1GB/s over a single 10Gb NIC.
12. Re:10GbE isn't that interesting by PeeAitchPee · 2016-01-14 07:51 · Score: 2
  
  SMB3 Multichannel isn't the same as link aggregation. It assigns CPU cores to process SMB transfers as they come across the wire(s), thereby handling one of the real-world bottlenecks (i.e., that the client typically chokes trying to process all of that inbound data coming off of the fast pipe).
13. Re:10GbE isn't that interesting by amorsen · 2016-01-14 08:42 · Score: 1
  
  Are they actually built in ports though? Do they have 10Gbase-T PHYs?
  Personally I prefer DAC, which is fairly cheap all around, but 10Gbase-T is winning in the market. Slowly. I haven't actually touched any 10Gbase-T equipment yet.
  
  --
  Finally! A year of moderation! Ready for 2019?
14. Re:10GbE isn't that interesting by cciechad · 2016-01-14 09:25 · Score: 1
  
  If you care about power per port you don't use the cat6 version you use the SFP one with twinax or 10GBaseSR. Cat6 is good because you can use your existing cable plant but you get increased latency and power use.
  
  --
  https://www.fsf.org/associate/support_freedom
15. Re:10GbE isn't that interesting by afidel · 2016-01-14 09:37 · Score: 1
  
  To put some numbers to it
  Cisco 3064 switches:
  3064-X 64 ports of DAC @ 143W = 2.2W per port
  3064-T 48 ports of GBaseT and 4 SR4 uplinks @362W = 7W per port
  Brocade 6740 switches:
  Brocade VDX 6740 48 ports of DAC and 4 ports of 40Gb QSFP @ 110W = 2.1W per port
  Brocade VDX 6740T 48 ports of 10GBaseT and 4 ports of 40Gb QSFP @ 460W = 8.8W per port
  24x7 operation at $.10/kw ~= $1/W/year so each port of 10GBaseT costs you ~$9-13/year (two sides to the link) over DAC, that would eat up the cable savings pretty fast. Add in reduced reliability and higher latency and it's not too surprising that 10GBaseT hasn't taken off.
  
  --
  There are 4 boxes to use in the defense of liberty: soap, ballot, jury, ammo. Use in that order. Starting now.
16. Re:10GbE isn't that interesting by dujardin · 2016-01-14 10:38 · Score: 1
  
  See Avago's ExpressFabric for an interesting alternative.
17. Re:10GbE isn't that interesting by Anonymous Coward · 2016-01-14 12:28 · Score: 0
  
  link aggregation requires cpu cores too. most of the time, however, that aggregation can be handed off to the network card. meaning SMB3 is a patch for fake servers. again, it's easy to reach 8gbps on 10GbE, however you usually are limited by other devices. the old saying is, spend the money and ye shall receive
18. Re:10GbE isn't that interesting by BitZtream · 2016-01-14 18:32 · Score: 1
  
  LACP doesn't actually increase bandwidth. Each host combination only talks to one port. At no point will any IP connection between those hosts go faster than 10gb.
  LACP will let multiple ports talk to multiple other machines with one IP and load splitting (its not balancing since its static mapping) across the LACP group. Its barely more useful than round robin DNS, and you'll lose any advantage from protocols that support multiple links, like iSCSI or SMB3.
  
  --
  Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
19. Re:10GbE isn't that interesting by BitZtream · 2016-01-14 19:05 · Score: 1
  
  Link aggregation results in fail over and load splitting. At no point will any single IP connection get more than the speed of any given port. The speed will match the port thats being used for that particular connection. If you just have two machines, even both with bonded NICs, you'll never get more than the speed of the fastest port in the group (assuming they aren't all the same speed)
  The point is not to get to 8gb/s on a 10gb connection between 2 hosts. The point is to get to 19.8gb/s on 2 10gb/s NICs between two hosts, which is possible with both SMB3 and iSCSI multi path. Neither of which is possible with 2 10gb ports in a LAG group of any sort.
  The best you can do with a 2 port 10gb lag group is 2 10gb/s transfers between the server and 2 different clients, assuming that your balancing method doesn't end up putting both hosts through the same NIC (its just a mod 2 on a hash of the ip or ip/port or mac or something depending on how you've configured the LAG group) but you'll never get a single connection over the speed of one port.
  
  --
  Persistent Volume manager for Kubernetes - https://github.com/dwimsey/openshift-pvmanager
Finally by Anonymous Coward · 2016-01-14 05:22 · Score: 1

Finally, an Intel Xeon killer.
1. Re:Finally by Anonymous Coward · 2016-01-14 08:30 · Score: 0
  
  It should at least nibble at Intel's sickening profits. With the rise of ARMv8, the future is finally looking brighter. This is the leading edge of a trend, and many look forward to escaping from Intel's years of monopoly abuse and x86 lock-in.
2. Re:Finally by Anonymous Coward · 2016-01-14 09:05 · Score: 0
  
  Intel's on top because their products are by far the best in terms of price/performance when you look at overall cost (Power, cooling, integration, reliability, etc). Datacenters are as commoditized and price sensitive as you can get.
  You can get some pretty damn high-end hardware for dirt cheap on ebay. Why? Because it was cheaper to buy new kit for the power savings alone.
  AMD has been, at best, stumbling for the past decade.
Perfect Use by 110010001000 · 2016-01-14 05:25 · Score: 1

This is a perfect way of speeding up the Cloud.
1. Re:Perfect Use by Anonymous Coward · 2016-01-14 06:37 · Score: 0
  
  This is a perfect way of speeding up the Cloud.
  So it can fail faster?
and few shits were given... by Anonymous Coward · 2016-01-14 05:26 · Score: 0

wasn't this originally supposed to launch around this time last year...?
Word On The Street by zenlessyank · 2016-01-14 05:48 · Score: 2

is that this chip is going be about US $150.
AMD still around? by Anonymous Coward · 2016-01-14 05:55 · Score: 1, Insightful

I thought they went bankrupt years ago.
Still not going to touch them with a bargepole :)
1. Re:AMD still around? by KiloByte · 2016-01-14 07:42 · Score: 2
  
  AMD is vital. All recent and semi-recent Intel CPUs include AMT which is a backdoor that can control any aspect of the running system without being detectable in any way by the operating system. It includes a completely separate sub-processor that has full control of the machine while being invisible to the main CPU.
  
  --
  The creatures outside looked from Alt-Right to Antifa; but already it was impossible to say which was which.
2. Re:AMD still around? by Anonymous Coward · 2016-01-14 08:04 · Score: 1
  
  Blah Blah Blah conspiracy theory drivel. Too bad AMD embeds and entire ARM core into it's own chips to mimic the same functionality of "AMT". AMD calls it "Trustzone" but it's the exact same thing.
  Oh, and while you wet yourself like a frightened preschooler over the NSA backdoor that was supposedly entirely implemented around the RDRAND instruction (which AMD is flat-out copying for Zen), AMD also put in black-box "crypto accelerators" for these server parts since the CPU power of an ARM chip is so anemic. Yeah, that's right kids: You want to encrypt any data at greater than dialup modem bit rates? You hand it off to a black-box crypto accelerator with no documentation or verification. Thanks AMD!
3. Re:AMD still around? by Anonymous Coward · 2016-01-14 16:49 · Score: 0
  
  False. AMD processors work fine without any firmware loaded on the TrustZone core. Doing that obviously loses the integrated TPM functionality, but the x86 CPU will happily operate.
  Now try running a intel CPU without loading the blessed ME and AMT blobs (hint: hard reset after 500ms).
4. Re:AMD still around? by drinkypoo · 2016-01-15 02:21 · Score: 1
  
  False. AMD processors work fine without any firmware loaded on the TrustZone core.
  As far as you know, sucker.
  Look, I have been using AMD processors since the K6/2, and have recommended them to many others over that time, but don't pretend that you know what's going on inside their CPU. You haven't the foggiest nucking fotion. If they slowed that prick way the hell down and ran their own code on it you'd be none the wiser.
  
  --
  "You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Decent specs.. but by wbr1 · 2016-01-14 05:55 · Score: 3, Funny

Hopefully the sales of this will be good enough for them to push the Zen core chips this year. If they come out as expected, they could be a great chip. I have several Bulldozer based machines, and when built they were a good balance of performance for price, but no more. They are really starting to show their age. They have always been spanked by i5 and above Intels, and as new intels continue to come out, prices on one or 2 generation back intels get better and the price difference ceases to be an issue.
So, please AMD, I want to be a fan, give us a good CPU again!

--
Silence is a state of mime.
Could be nice in routers, NASs, etc. by Doke · 2016-01-14 05:58 · Score: 1

These could be nice in appliances, ie routers, switches, NASs, etc. I can't see them being too useful in normal servers, since they're ARM, not intel, and have a relatively low clock rate (1.7 to 2.0 GHz).
Tell me when I can actually buy a system by erice · 2016-01-14 07:30 · Score: 1

And not from SoftIron. "Available today from SoftIron" actually means available to somebody soon, maybe. But you won't find a listing anywhere and they won't even respond to queries from individuals.
1. Re:Tell me when I can actually buy a system by Anonymous Coward · 2016-01-14 16:25 · Score: 0
  
  http://softiron.co.uk/order/
  I bought one today. They ship it tomorrow.
Chromebook! by Anonymous Coward · 2016-01-14 07:48 · Score: 0

Great. Now I want a Chromebook with this.
And a REALLY big battery.
Perfect for NAS if boards are reasonable... by Anonymous Coward · 2016-01-14 08:06 · Score: 0

This is exactly what I have been waiting for: a reasonable NAS board with decent CPU, ECC, and ample SATA, which isn't outrageously overpriced. Perfect for ZFS. ARM is a bonus for those looking to get away from x86 and its various "features". Assuming the 8x PCIe slot is open ended, it could make an inexpensive (ECC enabled) desktop too.
The Intel C2750 could have been that, but the $450 boards were not appealing; at that point, one may as well get a Xeon. Strangely enough, this wasn't entirely Intel's fault, but it doesn't change the fact that it was rip off either way. With a more integrated ARM SoC, there is at least the possibility of an inexpensive NAS board. Even if not, I expect that many will pay a premium to escape from Intel's crappy architecture and monopoly abuses.
AMD FX only had 1 HTX link used by the chipset by Joe_Dragon · 2016-01-14 08:55 · Score: 1

AMD FX only had 1 HTX link used by the chipset the cpu's with more where server ones.
Deja Vu by rakslice · 2016-01-14 09:22 · Score: 1

Didn't I read about this already like 2 years ago?
http://www.anandtech.com/show/...
Nintendo? by Anonymous Coward · 2016-01-14 09:38 · Score: 0

So, is this the 64 bit ARM processor, which will be appearing in the next generation Nintendo console? ARM wasn't ready in time for the Xbox One, but maybe it is for Nintendo.
Everything old is new again. by tlambert · 2016-01-14 10:54 · Score: 1

Everything old is new again.
It's basically a SiByte or Cavium network processor, only with ARM instead of MIPS.
It'll probably be useful for terminating SSL connections in a box to offload the cryptography, and for offload of packet reassembly and other less interesting things from your main compute cluster, but not much else.
The main problem is still that most ARM implementations memory bandwidth sucks; not knowing who was on the design team for the thing, until we get real numbers out of benchmarks, it won't be clear if the Hypertransport/DEC Alpha design people that AMD ate way back when were involved or not, and whether the I/O bus is crossbar or serial. I guess we'll get to find out.
Obviously, first gen is about getting product out; it'll be more interesting if/when they cram 8K of the processors into a single 1U box, or however they plan on 2nd and 3rd gen'ing the things, and whether they have/will address the abysmal ARM memory bandwidth issues (or not).
And now combine... by MoarSauce123 · 2016-01-16 00:46 · Score: 1

....x86 cores with ARM cores and provide drivers for OS so that work load can be switched seamlessly from power saving ARM to high performance x86 as well as run both and use x86 and ARM based apps simultaneously. Now THAT would be interesting!