Deep aluminum alloy die casting shell learning and computing server

★1。 Processor: Die Casting communication shell with more than ten cores; 2. Memory: 16gbddr42400mhz or above; 3. System hard disk: 400gpci-essd solid-state disk or above; 4. Data hard disk: 3.5-inch 4t7200rpm or above enterprise hard disk; 5.gpu card: more than 6 GPU cards (CUDA core number 5120, 12g video memory). Optical drive: 16xdvd or above. Power supply: 1200 high efficiency power supply / pmbus or above 8. Display: 24 inch 2K resolution display or above 9 inch. Supports keyboard and mouse 10. The equipment can support no less than two SD cards of aluminum die-casting communication shell, support 16g, 32g, 64g and 128G, and support system fault through the equipment. 11. Matrix multiplication test report, 72 hours full load GPU test report and other related test materials, and provide the copy of test report; the operating environment must provide all the following software, installation and debugging: GNU compiler, including C / C + + / Fortran Intel compiler, including C / C + + / Fortran compiler, MKL, MPI, etc. Parallel environment: configure OpenMP parallel environment; configure openmpi, MPICH and other MPI parallel environments; GPU development environment: configure new CUDA driver, cast aluminum communication chassis compiler, debugger, SDKs, sample file and deep learning framework Caffe, torch, tensorflow and anano. Provide the copy of software copyright registration certificate of server running environment monitoring system and server computing architecture exchange control system.