| 10 | 10 | For the purpose of the visualization applications scenarios we have prepared an innovative HPC testbed setup where many-core and accelerated hardware (GPUs) can be used used in conjunction with high-speed networks. One of the key requirements of the modern visualization scenarios is an access to fast (10GBit or even faster up to 100GBit) optical networks that can effectively link computing resources with high-resolution visualization panels aiming at the 16K resolution (64 x Full HD). The setup can then be used to visualize online and offline results of various scientific application, including multimedia streaming, real-time many-core processing, 3D Internet, bio-chemistry, nanotechnology and many more. Moreover, the setup can be used to run various application benchmarks to identify potential bottlenecks for both software and hardware. A sample of architecture for the 16K visualization setup is presented in Figure 1. This architecture was actually deployed and tested at PSNC in 2011[??]. It was based on two 100Gb routers, each equipped with one 100Gb port and ten 10Gb ports. One of them was placed in data center where up to ten fast multicore Intel servers were connected with it by 10Gb links. The second router was placed near 16K display. Both routers were connected remotely with 100Gbit Ethernet optical link. The visualization 16K panel was driven by to up to ten pc computers connected with the second router with 10Gb links. Each of these pc's was operating up to 6 to 12(depending on the actual demo configuration) LCD tiles of the visualization tiled-display. |
