A year from now it will be two separate companies, but for now, it’s business as usual for HP, and that means rolling out new products and services like the latest additions to its Apollo family of servers. The company has announced new server trays for the HP Apollo 8000 and the HP Apollo 6000 with supporting application accelerators from Intel and Nvidia.
“We’re continuing a lot of the momentum we’ve had with the Apollo platform,” said John Gromala, Senior Director, Hyperscale Product Management, HP Servers. He told IT Trends & Analysis that they’re “definitely hitting the sweet spot of computing, not only in oil and gas, but even with a lot of enterprises”.
It’s still very early days for the new HPC platforms, with volume shipments having just started, he said. While 8000 series shipments are really just starting, the 6000 series is seeing stronger success. Gromala said he couldn’t give more detailed information because it was HP’s quiet period, but did note that the company has been growing at 13% over the last 5 years, double the growth rate of other vendors. He largely attributes this growth to HP’s new approach, optimizing systems for workloads.
Customers have been heavily involved with beta programs, although the majority with the air-cooled 6000, rather than the much more complex water-cooled 8000 series. Doing it on the bigger 8000 systems is more of a challenge, and they tend to be early deployments instead, he said. HP will begin to taking orders in December, and will be running them out in volume in the first quarter of 2015.
First unveiled back in June, Apollo consisted of two high-performance computing systems, the air-cooled Apollo 6000 and the Apollo 8000, which HP called the world’s first 100% liquid-cooled supercomputer. A month later, HP’s Ed Turkel, Group Manager, HPC Segment Product Management, said accelerators are one of the hot relatively new technologies as organizations grapple with growing demand for compute performance. He made the following claims for Apollo systems:
-6000 System – 35% greater performance for EDA; $3 Million savings per 1000 servers over 3 years; 4x density per rack per dollar; and,
-8000 System: 4x faster molecular simulations; $1 Million (up to) energy savings over 5 years; and, 3,800 tons removed of CO2 per year.
According to IDC, the HPC market declined slightly – down 2.6% year-over-year – for the second quarter, to $2.45 billion. At 32.4%, HP is the dominant HPC vendor, followed by IBM (24.6%) and third-place Dell (16.3%) Other key vendors include NEC, Bull, Dawning, and SGI. Shipments increased 6.6% year over year to 33,277 units, with the workgroup (systems priced below $100,000) and departmental ($100,000-$249,000) segments making up 54.5% of all HPC server systems revenue in the quarter.
The HPC market saw a much bigger dip in Q1, with factory revenues down 9.6%, but shipments – at 33,577 units – were ‘relatively steady’, growing 0.4% YoY and the market expected to average 7.4% yearly growth to $14.7 billion by 2018. “HPC technical server revenues are expected to grow at a healthy rate because of the crucial role they play in economic competitiveness as well as scientific progress,” said Earl Joseph, Program Vice President for Technical Computing at IDC. “As the global race toward exascale computing fuels the high end of the market, more small and medium-sized businesses and research organizations are exploiting HPC servers for advanced simulations and high performance data analysis.”
Like IDC, Gromala wasn’t overly concerned about the last two quarters HPC trends. It’s not uncommon when new products like Apollow come out, or new Intel processors, for customers to want to wait to maximize the benefits for their dollars, he said. HP has gotten “tons of interest” in the Apollo technology as HPC spreads beyond its traditional oil and gas, scientific and R&D roots.
The number of enterprises looking at these high-performance benefits are growing, he said. HP calls them high-performance enterprises, and they will use HPC as a “backbone to how they run their businesses”.
In other news from HP, the company announced the general availability of HP Vertica for SQL on Hadoop, an analytics platform that enables customers to access and explore data residing in Hortonworks, MapR, Cloudera, Apache based data lakes, or any combination of these leading Hadoop distributions.
“With HP Vertica for SQL on Hadoop, we’re combining the best of both worlds – the industry’s enterprise-proven HP Vertica Analytics Platform with the broadly accessible Hadoop ecosystem,” says Shilpa Lawande, GM Platform, HP Software Big Data Business Unit. “Now, organizations can store data in HP Vertica or any Hadoop distribution, explore data in place using HP Vertica for SQL for Hadoop, and serve the highest-performance analytic needs for mixed workloads with HP Vertica Enterprise.”
The Fiddly Bits (& Bytes)
Based on Nvidia’s CUDA architecture, Nvidia accelerators enable seamless integration of GPU computing with HP ProLiant servers for high-performance computing, large data center graphics, and virtual desktop deployments:
-Nvidia Tesla K10 Dual GPU 8 GB and K80 Dual GPU 24 GB, for double-precision algorithms; -Nvidia Tesla K20 5 GB, K20X 6 GB, K40 12 GB, K80 Dual GPU 24 GB, K40 XL dense form factor;
-Nvidia Quadro K2200 and NK4200, K5200 and K6000; and,
-Nvidia GRID K1 Quad GPU, K2 Dual GPU.
Based on the Intel Many Integrated Core (MIC) architecture, the Intel Xeon Phi coprocessor 5100 series is optimized for high-density computing and is well-suited for workloads that are memory-bandwidth bound, such as STREAM; memorycapacity bound, such as ray tracing; or both, such as reverse time migration:
-Intel Xeon Phi 5110P; and,
-Intel Xeon Phi 7120P and I7120D.
HP Insight Cluster Management Utility (CMU) eases the burden of managing thousands of compute nodes—including both CPUs and accelerators. HP iLO Management with Integrated Lifecycle Management provides new levels of performance and quality of service with HP Active Health and Agentless Management.