This site may earn affiliate commissions from the links on this folio. Terms of employ.

AMD-Opteron-A1100

Today AMD is formally launching its starting time ARM processor cadre, the Opteron A1100. AMD start announced its plans to enter the ARM marketplace in 2013, with the chip expected to transport by mid-2014. The visitor apparently began early sampling effectually that fourth dimension frame, but is simply now launching the processor.

The new Opteron A1100 is pretty much what AMD promised in its early previews of the device. It packs viii Cortex-A57 CPU cores, with each pair of cores sharing a 1MB L2 (512K effectively allocated to each chip). An 8MB L3 cache backs the entire CPU cluster, and the CPU supports both DDR3 and DDR4. ECC support is also provided.

AMD-Opteron1

The new A1100 supports up to 64GB per channel (128GB total) if registered DDR4 DIMMs are used, and 64GB of total retentiveness when using standard DDR4. The board also includes dual 10GbE ports, 14 SATA3 controllers, and i PCIe three.0 slot with x8 support.

The exact SKUs are shown beneath:

Opteron-SKU

The quad-cadre A1120 is a 25W chip at i.7GHz, the A1150 is a 32W cadre at 2GHz, and the A1170 is a 32W fleck at 2GHz. The last two flavors are both 8-core processors, but all three SKUs share the same memory capabilities and L3 enshroud. If AMD follows its past history on evaluating TDP, the 25-32W figures volition represent a worst-case scenario for the chip, rather than its performance in a full general workload. AMD historically rates TDP in the first manner, while Intel uses the second — and that'southward why you tin can't usually directly compare TDPs between the ii companies.

Will it sell?

AMD has already lined up launch partners, with back up from SoftIron — though the imprint on the visitor'southward website states that systems are only bachelor in "limited quantities."

There are some major questions regarding AMD's conclusion to push into ARM, however. For one, the company told journalists that while it expected its new A1100 to compete well confronting Intel's Avoton line of Atom processors that first debuted in 2013. If AMD had kept its initial launch schedule and shipped the Cortex-A57 in 2014, information technology may well have had a leg up in the nascent dense server marketplace.

Betwixt the 2014 expected launch and the actual shipments today, Intel had time to bring another line of server products to market — the 14nm "Xeon-D." That bit has a very potent value statement — Anandtech labeled it "the Xeon D is probably the most awesome production Intel has delivered in years." It'southward not specially suitable for HPC or big-memory requirements, only it's nifty for everything else.

AMD has stated that it expects A1100 chips to start around $150, which would price them competitively against Intel's Cantlet-based hardware, just doesn't exactly offering the company a lot of profit margin.

AMD'due south delay may have limited the A1100'due south sales market, but information technology may turn out to be a smart move in the long run. It's hard to say that AMD is tardily when no other ARM server vendor has managed to field an Intel-compatible product. The ecosystem and back up network that ARM servers need to be competitive with their x86 counterparts is still in its infancy, and it ultimately doesn't matter how proficient your hardware is if the software stack isn't there to back up it. Rory Read, AMD's previous CEO, justified buying SeaMicro and investing in ARM every bit a long-term program considering he claimed the server market would be at least 15% ARM past 2018. This was always a dubious argument, and it's now conspicuously false — ARM will non shoot from 0% to 15% of a market in less than 24 months.

From this viewpoint, the A1100 is a commercial proof-of-concept production that may not offer competitive functioning across the lath, simply gives both AMD and its vendor partners an opportunity to earn a small ROI while ramping upwards second-generation designs more than chop-chop. In the past, AMD claimed that the A1100 would be its first ARM production, with K12 following quickly behind. The company hasn't talked about K12 at all since Jim Keller left the company, but information technology's supposedly on track for a 2017 deployment.

I tend to think that AMD did the right thing past easing Seattle out the door rather than drib-kicking it in 2014. Server customers care merely every bit much about stability and robust software solutions than they exercise about hardware functioning. I wouldn't look to the A1100 to do much to reverse AMD's lost server market share or overall earnings, yet — the company has nonetheless to announce any major purchasing wins, and it hasn't shown performance data.

At best, the Opteron A1100 will wedge a foot in the door of the ARM server market, with a few small-scale wins and enough competitive force to fuel customer involvement in future markets. If AMD or Qualcomm desire to bust this marketplace open, they'll probably need custom designs and meliorate hardware to exercise it.