Your browser is not supported. Please update it.

30 November 2018

Amazon launches Intel offensive, spins up ARM server CPUs

Amazon’s AWS cloud computing business, which is still fending off calls for investors for it to be spun out of the retail monolith, has just unveiled the spanner that it plans to throw into Intel’s work – a tool that it has been teasing for some time now. It’s a big win for ARM, but Intel’s x86 portfolio has a stranglehold on the application processing field, and so it will require a monumental effort to dethrone the current CPU king.

AWS is now offering its A1 Graviton chips to customers looking to buy compute resources. As an alternative to Intel’s Xeon line, the new Graviton has to compete on both the upfront purchase and integration prices and then the operational costs for data center use – which is mostly just the power they use.

Now, ARM has pedigree in low-power designs, and Intel knows how difficult it is to try and enter that space. Intel aborting its mobile initiative was a great win for ARM, but while ARM-based CPUs have been threatening to encroach on the data center for the past two years, this is one of the strongest signs that ARM might finally start taking chunks out of Intel in its core market.

Amazon bought Annapurna Labs back in 2015, an IoT-focused ARM SoC designer. In time, Annapurna designed the Nitro SoC, which AWS used for networking and storage functions in its EC2 cloud systems – offloading work from the x86 chips in the servers. Now it seems that Annapurna’s designs have evolved sufficiently to offload the x86 chips entirely.

This isn’t good news for Intel, as Microsoft has said that it would like to eventually have 50% of its Azure servers to be ARM-powered. Marvell recently acquired Cavium, another prominent ARM data center designer, and will be pushing the Cavium ThunderX2 designs as well.

Intel’s x86-rival, AMD, is also pushing forward, with a new generation called Epyc that has leaped ahead of Intel’s designs – which have floundered as Intel struggles to get a foothold on the 10nm production process. AWS does offer Epyc configurations in EC2, although it’s not clear how the A1 chips compare to the AMD Epyc performance. Apparently, AMD was in talks with AWS to build the ARM-based Graviton line, although that partnership fell apart.

Currently, A1 Graviton instances are priced cheaper at AWS than comparable x86 ones, but good luck getting AWS to confirm that it isn’t pricing below cost to encourage early adoption. Of course, AWS spends a lot of money on Intel parts, and so the Graviton experiment could help it negotiate lower prices from Intel – or conversely, enable it to move completely away from Intel if it chose or needed to.

One of Intel’s saving graces is the difficulties that many companies run into when emulating x86 and Windows on ARM chips. There’s also the potentially threat of IP lawsuits, should Intel want to throw some weight around that way, but it seems likely that judges and juries would take kindly to the monopoly power trying to crush a rival out like that.

For now, the A1 is available in AWS’ EC2 instances, and can be bought ranging from the small 1 virtual CPU (vCPU) with 2GB of RAM, all the way up to the 16 vCPU with 32GB of RAM. The pricing ranges from $0.0255 per hour for the smallest configuration, called ‘a1.medium,’ to $0.408 per hour for the big ‘a1.4xlarge’ setup – which seems to be four quad-core chips in a single instance, rather than a 16-core chip. At those prices, AWS says that this makes them up to 45% cheaper than x86 VMs.

Of course, your actual costs are going to depend on your workload, so if the software runs faster on x86, you may still have a smaller bill than the cheaper-per-hour options. This means that the Intel line has an early advantage, but if software companies begin optimizing for ARM, then that advantage could be quickly eroded.

Modern code and software should be able to handle the shift to the ARM chips sufficiently well. Amazon is providing the A1 with Amazon Linux 2, Ubuntu, and Red Hat Enterprise Linux 7.6. If you are planning on using Linux in the cloud, then it seems like you’ll be well supported, but if you absolutely need a Windows environment, then things might not be so smooth.

The A1 chips appear to be based on the Cortex-A72 core. They support the ARMv8-A architecture, which is running the AArch64 instruction set. That should mean that you can port most modern programs around, but there’s also support for legacy 32-bit applications inside AArch64. If you need that legacy support, then maybe a bleeding-edge VM instance isn’t for you.

One thing to bear in mind is that the physical chip currently appears to be just a quad-core design. That means, unless there’s something pretty revolutionary happening in the motherboards, the Graviton line is currently limited to four physical cores per physical socket.

In comparison, Intel and AMD both have options for 16 physical cores, offering 32 processing threads thanks to hyperthreading, and the more extreme designs have 32 physical cores, for 64 threads (32C/64T). Without knowing more about how AWS is actually racking them in its data centers, it’s not yet clear if this is a potential roadblock moving forward.

There are potential benefits for networking resources, when it comes to linking things to PCI lanes in the motherboards. The ARM designs appear to enable much more PCI options, which means you could fit more peripherals into a board and better optimize its usage – rather than having much of a beefy x86 core idling due to running out of PCI lanes, you could get by with a smaller ARM chip like this. Again, we’ll be doing some digging to confirm this.

If this was Amazon simply trying to sell a chip to customers (much like Intel, AMD, or Qualcomm), we would be pointing out that corporate inertia is a massive obstacle to overcome, and convincing CTOs and engineering heads to jump ship from the established x86 ecosystem in ARM would be very difficult in the short term. You would need a few major success stories to sway opinion here, especially in such mission critical roles, but for ARM at least, there’s a good chance that AWS itself can become one of those flagship successes.

As for ARM itself, the AWS announcement comes not long after the SoftBank-owned chip-designer unveiled Neoverse – the new brand for its data center offerings, segmented out to prevent confusion with the Cortex CPU line. Neoverse is currently on the 16nm Cosmos design, but ARM says it will be pushing forward with a new generation each year, moving to Ares (7nm), Zeus (7nm+), and then Poseidon (5nm).

ARM says that a million Neoverse servers will ship in 2018, and that they are proving popular in ‘a new class of cloud servers that manage the networking, storage, and security workloads.’ That will sound familiar, as it’s a similar function as to what those AWS Nitro chips were doing. Now that the A1 Graviton is on the scene, the Neoverse platform can begin to pursue the application processor workloads too.

Currently, the AWS EC2 A1 instances are being targeted at scale-out workloads, which include containerized microservices (buzzwords ahoy!) as well as web servers, development environments, and caching. In time, we expect to see IoT workloads cropping up as standalone elements too.

In the meantime, it’s going to be interesting to see how the AWS announcement alters the cloud computing marketplace. Intel and ARM are partners, working together in ARM’s Pelion PaaS, and while Intel might be playing nicely, the competition between the two could get very bitter very quickly.