This morning’s AMD press event had two highlights. The first was the announcement of new x86 and ARM cores, which Scott already covered earlier. The second was the first public demonstration of Seattle, an upcoming server chip based on ARM Cortex-A57 CPU cores.
The demo involved web hosting. AMD showed the entire LAMP (Linux, Apache, MySQL, PHP) stack running on top of a Seattle-powered development server. You probably can’t read the little spec sheet in the picture above, so here’s what it says:
AMD Opteron A1100 series development platform
- AMD Opteron A1100 Series processor
- 4 registered DIMM slots for up to 128GB of DDR3 DRAM
- PCI Express connectors configurable as single x8 or dual x4 ports
- 8 Serial ATA connectors
- Compatibility with standard power supplies
- Linux environment based on Fedora
The demo kicked off with this system spitting out web pages with information about the various software installed: Apache 2.4.6, MySQL 5.5.35, and PHP 5.4.16. The operating system was identified as “Red Hat Server for ARM 2.0 (development preview),” although I don’t see that release referenced on the Red Hat website.
Next, AMD showed the machine running a WordPress blog and serving a video. WordPress compatibility is a particularly big deal, according to the chipmaker, because “more than 60 million” websites and blogs are powered by WordPress. As for the benefits of video streaming capabilities in a server, those need no emphasis.
AMD announced a few weeks ago that Seattle is sampling and on track for a release in the fourth quarter of this year. For more details on the chip, don’t miss our previous coverage.
I’ve just discovered something odd. There is a small country in Europe, called Belgium (although there are a lot of countries like that in Europe, with 10 million inhabitants). In that country there seems to be no official agent for AMD anymore. That means that all Belgians HAVE TO buy Intel chips! No wonder AMD has bad figures.
I wonder how the situation is in those other little countries, like Portugal, Greece, the Netherlands… If they lost their agents there too, they might have lost half a continent.
This chip seems to be undermined by their “Never Seattle” campaign.
Meadows beat you by 10 hours. :p
The shame of thinking like Meadows, only slower!
I need to stop posting after midnight.
I hope you jest. Thinking like me makes you a genius!
Even if it’s done slower…?
EDIT: oops… did I just bump Saber..?
This is odd. It’s like a full-size server board as you would expect with a traditional x86 server instead of ARM micro server. Unless this is competitive in performance with x86 then I don’t see the point of it.
Note the hand sized PCB?
And look under for more info on uses
[url<]http://www.opencompute.org/projects/server/[/url<]
It’s funny how my brain works sometimes: I saw “serving video” and immediately went on the tangent of redeploying this as a MythTV HTPC… and then thought “wait, no GPU… but there’s an x8 slot” and then lamented the lack of IGP, and then thought of the irony of this being AMD and how they could… oh wait, I just reinvented Tegra.
I could see the Tegra board doing well running mythfrontend or XBMC. But both the backend and the frontend of MythTV? That doesn’t seem like a good idea. Actually the Tegra board doesn’t really make sense (at the current price) to be honest.
You would have less headache (and save a whole lot of money) just going with Bay Trail or Kabini ($80 and $90 vs $192).
That and the fact that it will take you roughly 5 years to recoup the price difference in power savings.
I wasn’t thinking in terms of building one for my own use. I was thinking in terms of AMD designing something so that OEMs could produce cheap set-top boxes. But I suspect they’d have to be so cheap that it wouldn’t really be worth it for AMD (especially compared to the server market if they can actually establish that niche).
You need a GPU only for Live streaming where Video needs to be encoded on board. for the rest of the file based video servers do not really deal with the video encoding, they only identify frame boundaries and serve the raw video data.
Yes, but “video streaming” prompted me to jump to the related but completely different application of HTPC, from source to receiver, so to speak.
Microsoft announcing an ARM build of Windows Server in 3… 2… 1…
and it’ll be released in a year or more, but they can’t be seen to ignore this market.
One thing is true, Seattle is very close to Redmond.
Never Seattle Forever
Here’s a bump before Saber Cherry gets all the attention.
Don’t talk trash about Seattle!
Ooooooo…. RedHat support for ARM is a big news. Other distributions already support ARM in some fashion but that’s a big player in the enterprise market due to support. That’ll help spur adoption among the low end server market (ie the exact LAMP stack demoed).
The available IO also looks interesting. 8 SATA lends itself to the NAS market and two 4x PCIe slots could be used for high speed networking or NVMe SSD’s. As a single PCI-e 8x slot, that’d be good enough for a single GPU.
The only thing left on my wish list for a low end server product would be some out-of-band management and that’s something that OEM’s (HP, Dell, Supermicro) typically add on their own accord.
I can read it.
The only way I see this being useful is if either the Price or the power consumption is significantly less (for the whole system) than x86 for the same performance. I don’t see either of those being more than 15-20% less, so I predict it will fail.
The appropriate form factor for this would seem to be blades, to get the most serving capacity per cubic foot (and make the best use of the low power and cooling requirements). I realize this is just a demo, but the 1U form factor just seems inefficient — and sets it up to fail when compared with 1U x86 systems.
The dev kit is to validate software. But yes, this will be leveraged by seamicro in their fabric solution.
Then its up to customers to decide. do I want a Xeon, Atom, Opteron based solution.
And soon, AMD want to make the ARM/x86 choice a non issu.
Run your x86 code, side by side with ARM.
Same infrastructure top to bottom, maybe even including the processor.
If AMD can unify the underlying RISC architecture and include two decoders one x86, one ARM ,
then you could software switch a blade to be in ARM or X86 mode for your client needs.
(An x86 decoder in modern CPU use about, what 5% of the transistor budget?)
“You probably can’t read the little spec sheet in the picture above,”
Cyril, was that a smart phone picture?? 😉
Where are the Beema A6 notebooks?!?!?! Wait a minute, where are the Kaveri notebooks?!?!?!
And where’s that famous A8 7600?
China, if you believe AMD…though I’m not sure that I do.