HP, however, counters that by highlighting the performance per watt achievable with GPUs. Hamilton talks up the company’s use of the latest PCI and networking technology to solve any bottlenecks.
“Four out of the top 10 use the Nvidia GPU, and AMD and Intel are coming out with their own approaches to acceleration,” said Hamilton.
Interestingly, HP is as keen to talk about system 117 on the list as its No.4 entrant. This is Georgia Tech’s Keeneland System, which consists of 120 HP ProLiant SL390s servers and 360 Nvidia Tesla M2070 accelerators to deliver 64 Tflop.
“This newest offering from Nvidia provides 6GB of memory per chip,” said Hamilton.
Further notes from the Top500 release:
- No. 5 is Hopper, a Cray XE6 system at DOE’s National Energy Research Scientific Computing (NERSC) Center in California with 1.05 petaflop/s
- Seven of the top 10 achieved performance at or above 1 petaflop/s, and five of these systems are newcomers.
- Five of the top 10 are in the United States; the others are in China, Japan, France, and Germany
- The U.S accounts for 275 of the 500 systems; Europe has 124 and Asia has 84
- China now has 42 systems on the list
- HP is still ahead of Cray measured in the number of systems, and both are trailing IBM
- Quad-core processors are used in 73 percent (365) of the systems, while 19 percent (95 systems) are already using processors with six or more cores.
Drew Robb is a freelance writer specializing in technology and engineering. Currently living in California, he is originally from Scotland, where he received a degree in geology and geography from the University of Strathclyde. He is the author of Server Disk Management in a Windows Environment (CRC Press).