The
fully operational Tianhe-1A, located at the National Supercomputer
Center in Tianjin, has been scored at 2.507 petaflops as measured by the
LINPACK benchmark. The result was announced Thursday at HPC 2010 China.
That score moves it past Cray's 2.3-petaflop Jaguar
located at Oak Ridge National Lab in Tennessee. The newest Tianhe,
which is Chinese for “River in the Sky” or “Milky Way”, achieved the
record using 7,168 NVIDIA Tesla M2050 GPUs and 14,336 Intel Xeon CPUs
consuming 4.04 megawatts. It operates with 262 TB of main memory and 2
PB of storage and boasts a proprietary interconnect technology developed
by China’s National University of Defense Technology. Tianhe-1A
epitomizes modern heterogeneous computing by coupling massively parallel
GPUs with multi-core CPUs.
According
to graphical processing unit manufacturer NVIDIA, which contributed
heavily to the project, it would require more than 50,000 CPUs and twice
as much floor space to deliver the same performance using CPUs alone.
More importantly, a 2.507 petaflop system built entirely with CPUs would
consume more than 12 MW. The use of GPUs in a heterogeneous computing
environment allows Tianhe-1A
to consume only 4.04 MW, making it three times more power efficient.
The difference in power consumption is enough to provide electricity to
over 5,000 homes for a year.
"The performance and efficiency of
Tianhe-1A was simply not possible without GPUs," said Guangming Liu,
chief of National Supercomputer Center in Tianjin.
কোন মন্তব্য নেই:
একটি মন্তব্য পোস্ট করুন