Super Micro, a server and computing providerhas launched a line of GPU accelerated servers and workstations based on NVIDIA’s Tesla GPUs. The new line combines Supermicro’s server design, thermal management, and energy-efficient power management technologies with NVIDIA’s Tesla K10 and upcoming K20 GPUs.
The 2U SuperServer features a cooling design that supports up to six double-width GPUs, dual Intel Xeon E5-2600 processors, up to 256GB, 1600MHz memory and high-speed interconnectivity through 10GbE or InfiniBand. The 4U SuperWorkstation runs an N-Body 2 gravitational simulation utilizing hybrid computing power of four passively cooled K10 GPUs combined with dual Intel Xeon E5-2600 processors. The 4U supports up to 16 DIMMs for 512GB 1600MHz memory, seven PCI-E slots, and eight hot-swap 3.5″ HDDs. Supermicro has also launched a smaller footprint SuperWorkstation with dual actively-cooled double-width GPUs, dual Intel Xeon E5-2600 processors, up to 512GB, 1600MHz memory in 16 DIMM slots, four internal 3.5″ HDDs, dual IEEE1394 Firewire ports, and build-in 7.1 HD audio.
Supermicro’s GPU platforms also include the scalable high-density 1U four GPU, dual processor SYS-1027GR-TQF featuring non-blocking GPU connections, the value-oriented 1U, two K10 GPU, uni-processor (UP) solutions SYS-1017GR-TF/SYS-5017GR-TF, 3U two GPU DP SYS-6037R-72RFT+ supporting up to 768GB 1600MHz memory in 24 DIMM slots and high compute-density GPU SuperBlade (SBI-7127RG) supporting 20x GPUs/20x CPUs in 7U. Supermicro’s GPU computing solutions support Intel Xeon E5-2600 series processors and offer high performance I/O and connectivity with PCI-E 3.0, 10GbE and 4x QDR/FDR InfiniBand.
For more information, visit Supermicro’s GPU.
