Supercomputer can perform 2 trillion calculations per second

The US supercomputer Aurora has been successfully assembled and will be operational by the end of this year.

The installation of Aurora’s 10,624th and final blade marks a major milestone for the exascale supercomputer at the U.S. Department of Energy’s (DOE) Argonne National Laboratory. After years of planning and research, the system now includes all the hardware that makes it one of the world’s most powerful supercomputers for scientific use. Built by Intel and Hewlett Packard Enterprise (HPE), Aurora is capable of delivering more than two exaflops of computing power.

The Aurora team built the system piece by piece over a year and a half, assembling the compute blades and other components and shipping them to DOE’s Argonne Leadership Computing Facility (ALCF) in November 2021, said Susan Coghlan, Aurora project manager at ALCF. “While there is still much work to be done before we can deploy the system to the scientific community around the world , we are extremely excited to have the final hardware installed,” Coghlan said.

As the “backbone” of the system, Aurora’s compute blades are rectangular blocks that house processors, memory, networking, and cooling technology. The machine’s computing power comes from Intel’s central processing unit (CPU) and graphics processing unit (GPU). Each blade is equipped with two Intel Xeon CPU Max Series processors and six Intel Data Center GPU Max Series processors.

Because each computing blade weighs nearly 70 pounds, the team needed a specialized machine to mount them vertically on Aurora’s refrigerator-sized racks. The supercomputer has a total of 166 racks, each holding 64 computing blades. The racks are arranged in eight rows, occupying the space of two professional basketball courts at the ALCF data center.

Before the system could be installed, Argonne had to make several major facility upgrades, including adding new space to accommodate the supercomputer and building mechanical and instrumentation rooms to increase power and cooling capacity. Now that the machine is complete, a team from ALCF’s Aurora Early Science program and DOE’s Exascale Computing Project will begin scaling the full system.

An Khang (According to ANL )

Source link