IBM launched its first POWER9 system, which positioned as “AI and cognitive workloads of a game-changing power station” – The benefits of artificial intelligence and cognitive computing and based on the same architecture as the Department of Energy’s Coral supercomputer summit and Sylar. The new system also supports the Nvlink 2.0 and OPENCAPI protocols, which provide up to 10x of PCI 3.0 maximum bandwidth based on x86 systems. According to Sumit Gupta, vice president of of AI and HPC within IBM’s Cognitive Systems business unit think tha The design concept the POWER9 is that it will work as a peer processor for a peer or other processor. Whether it’s a GPU accelerator, an FPGA, or any other accelerator on the market, our goal is to provide links and hooks so that all these accelerators can be developed equally on the server.
The new AC922 server is on two Power9 CPUs between four and six Nvidia Tesla V100 Nvlink GPU. The AC922 extends many of the design elements introduced in the Power8 “Minsky” box and focuses on enabling connections to a range of accelerators (Nvidia GPU, ASIC, FPGA, and PCIe connected devices)-using a series of interfaces. Power9 the consistency introduced by CAPI and Nvlink 2.0 is another key enabler code. AC922 solves these problems, allowing accelerated applications to use system memory as GPU memory. This reduces latency and simplifies programming by eliminating data movement and location requirements. The AC922 server can be configured with either four or six Nvidia Volta V100 GPUs. According to IBM, a four GPU air-cooled version will be available December 22 and both four- and six-GPU water-cooled options are expected to follow in the second quarter of 2018.
Thousands of AC922 nodes, along with storage and network connectivity, drive about 200 in Oak Ridge and 120 1.271 of 1.271. As King pointed out in our interview, only one of the original CORAL contractors is fulfilling its mission to deliver a “pre-exascale” supercomputer to the collaboration of US labs. Sano praised “IBM’s progress in the development of the latest energy technologies” and said that “the POWER9 Opencapi Bus and high-capacity memory feature allow for further innovation opportunities in Google data center.”
IBM is also being leveraged by Google, and it’s working with partner Rackspace to build a server called Zaius Power9 processor. IBM do see their strong appeal, but there are many other factors that are working. They have solutions that, after a very fast-growing AI space, our solution will be done after the open source database NoSQL the data center and Have announced cooperation with Nutanix to hyperconverged space which see the HyperScale market as “a good capacity opportunity”, and clearly aware of the impact of bulk pricing on the traditional server market. So we have a lot of different elements that drive the number and opportunities of power servers on our Linux, including of course SAP’s Hannah.
IBM will also sell POWER9 chips through its openpower ecosystem, which now contains 300 members. Three versions of the POWER9 chip will be deployed in 2019. There will be an amplified processor, which is a traditional chip for AIX and high-end space, and then another that will provide more acceleration with enhanced memory and built-in other features; we’re working with other memory providers.
Before the Power8 began to fall in the second year, Capabilities like Nutanix and building Powerai and other software-based solutions have led to some bounce, which is still negative. The pricing or at least price perception is a problem with Power8. For traditional market, I think pricing is competitive; for some of the new markets we’re trying to get into, such as the Hyperscaler data center, I think we have some work to do. It’s really a TCO and a price-performance competitiveness versus price only. And we think we’re going to have a much better price performance competitiveness with Power9 in the hyperscalers and some of the low-end Linux spaces that are really the new market.
We are very confident that we are building a lot of workloads on top of this architecture, we will see more growth, Power9 positive growth, powerai with Nutanix some other workloads we’ve put it there, it won’t be the only reason for a hardware. This will be the bulk of the software features we build on the platform and support more new workloads. In the next two or three years–it’s a huge hockey stick–we’ve built and designed POWER9 markets, especially for that market Over the next few months and years, more power9-based servers will follow IBM and its ecosystem partners.