HPE is the first major OEM to use Ampere computer arm chips


Hewlett Packard Enterprise is an early and enthusiastic supporter of alternative processor architectures beyond the standard Xeon X86 CPUs that make up the bulk of its revenues and shipments, especially with Arm server chips starting in 2011. Maybe now the fourth time is the charm because it adds Ampere Computing’s “Mystique” Altra Max Arm Server Chip as an engine in a standard ProLiant rack-mounted server.

The CPU vendor agreement with Ampere Computing is definitely a watershed moment for that particular Arm server chip manufacturer, which is landing its first major OEM server maker after having had some success to buy Baidu, Tencent, Alibaba, Oracle and now Microsoft its 80-core Altra. and 128-core Altra Arm CPUs in their custom server infrastructure—typically for inward-facing workloads, but increasingly for outward-facing workloads sold as raw infrastructure for customers of their clouds who want to learn more about Arm servers.

The announcement, made at HPE’s Discover 2022 customer and partner event, is quite modest, with only one machine – the single socket, 1U rack server known as the ProLiant RL300 – in the lineup. And given HPE’s history with Arm servers, you probably understand why the company is playing it a little more cautiously.

In November 2011, the company was just known as Hewlett-Packard and had the four-core EnergyCore ECX-1000 Arm server chip from Calxeda, which was basically a system-on-chip with a single DDR3 memory controller, an integrated management controller, four PCI-Express 2.0 controllers, a SATA 2.0 controller for flash and disk storage, and a fabric switch with eight 10 GB/sec Ethernet channels (three to connect to three other ECX-100 chips on a board and five to talk to the outside world, including other boards connected in a mesh). The “Redstone” server cards with the Calxeda processor were part of: the “Moonshot” hyperscale server design effort launched with much fanfare by Hewlett-Packard at the same time Calxeda unmasked. But the Calxeda chips only had 32-bit processing and memory addressing, and while 48-bit chips were on the roadmap, adding 64-bit chips would take years and that was just too long for Calxeda or Moonshot to get. Off the ground.

In 2017, Hewlett Packard Enterprise, as the company was called after the spin-out of its PC and printer business, touted Cavium’s ThunderX2 server chip as the next big Arm hope. (Well, it really was Broadcom’s recycled “Vulcan” Arm server chip.) The ThunderX2 was placed in HPE’s “Comanche” Apollo 70 dense, barebone servers. A lot of these systems sold, but nothing close to volumes, and a while after that Marvell bought Cavium for $6 billion and a few HPC systems were up and running and testing the ThunderX2, Marvell was about to launch the “Triton” ThunderX3 follow-on and then, as 2020 drew to a close, Marvell ThunderX3 peaked as Broadcom enriched ThunderX2 and this time, no one picked up the pieces. although there were rumors that Microsoft would do it around the same time

And then, a little over two years later, Microsoft tapped Ampere Computing as supplier of Arm server chips† We have thoroughly analyzed the Altra instances running on the Azure cloud and calculated that they offer anywhere from a quarter to a third better price/performance than instances based on the AMD “Milan” Epyc 7003 processors on Azure. And for all the reasons we’ve discussed in the past about the Altra chip line, and most importantly because it’s a chip that targets hyperscalers and clouds and cloud native workloads that perform more consistently on a core with a set clock speed and no simultaneous multithreading. HPE is adding the Altra chips to its regular ProLiant machines.

an aside: As we have discussed here in the past, HPE is also an avid supporter of Fujitsu’s A64FX Arm processordesigned with thick vector math units and used in the “Fugaku” supercomputer at the RIKEN Lab in Japan, which has been added to its Apollo 80 systems.)

It’s that deterministic performance and that price/performance that attracted HPE to the Altra CPUs, and the fact that HPE doesn’t sell this as a special Moonshot-type server, or any of its Cloudline ell cheap hyperscaler boxes, but in a standard ProLiant machine that comes with an iLO or OpenBMC baseboard management controller commonly used in enterprises, shows that HPE is serious.

“HPE and ProLiant have brought 30 years of leadership and innovation in computing to a sea of ​​followers,” explained Neil MacDonald, executive vice president and general manager of the Compute group at HPE. “We are deeply committed to innovation in compute, whether in automation management, performance, security or energy efficiency. And this is a very, very natural next step in our journey to bring cloud-native silicon computing solutions to support these emerging markets and address the evolving needs of customers. We are excited to partner with Ampere, who are delivering next-generation cloud-powered compute for the data center, and Ampere is modernizing that core compute for cloud software and cloud workloads.”

Not many feeds and speeds are available for the ProLiant RL300 server, but it is a single socket server in a 1U chassis with 16 DDR4 memory slots with up to 4 TB of main memory, which is a lot for a single socket vending machine. Both the 80-core Altra and 128-core Altra Max chips plug into the socket of the box, which has three PCI-Express 4.0 slots and two OCP 3.0 slots. The machine has up to two M.2 NVM-Express flash memory sticks and ten small form factor NVM-Express SSDs.

The ProLiant RL300 will be available in the third quarter. The price has not been disclosed, but we assume it will be priced aggressively against ProLiant machines that use Intel Xeon SP and AMD Epyc processors – and that in this case a larger share of the margin will be HPE’s profits. than has been done in the past with Xeon SPs.

We also anticipate that HPE will do a wider product portfolio, including a two-socket workhorse machine that competes with the flagship ProLiant DL380 with Xeon SPs and its rival, the ProLiant DL385 with Epycs. We also expect more compact sled servers for the Apollo line for those who want more density than these ProLiants offer. A full-width 1U box with only one server connector in it isn’t that compact, and it’s reasonable to expect a 2U chassis with four sleds in it at some point. Whether HPE takes the Ampere Arm server chips as seriously as MacDonald would have us believe. When asked about the delivery of such a full line, MacDonald smiled and said this:

“When we make decisions about technology, we take them very thoughtfully with a view to the future. And there are other activities already in the works that I’m not going to talk about today in terms of things we can do in the future. But we are very excited about the opportunity and the ecosystem that exists around this platform.”

As we said, when we see a ProLiant RL380 and an Apollo 2000 using Ampere Computing’s Altra CPUs, we know that HPE really means it.

Leave a Comment

Your email address will not be published. Required fields are marked *