rm 服务器芯片,NVIDIARTeslaRM级集成式服务器GPU概述-TESLAM-.PDF

NVIDIARTeslaRM级集成式服务器GPU概述-TESLAM-.PDF

™

TESLA M-C LASS

GPU CoMPUTinG Mod ULES

ACCELErATinG SCiEnCE

™ ™

Based on the cUDA architecture codenamed “Fermi,” the Tesla

M-class GPU computing Modules are the world’s fastest parallel

computing processors for high performance computing (HPc). Tesla

GPU’s high performance makes them ideal for seismic processing,

biochemistry simulations, weather and climate modeling, signal

processing, computational finance, cAE, cFDand data analytics.

Accelerate your science with NVIDIA® The Tesla M-class GPU modules are

Tesla™ 20-series GPUs. A companion integrated into GPU-cPU servers

processor to the cPU in the server, Tesla from oEMs. This gives data center IT

GPUs speed up HPc applications by 10x. staff much greater choice in how they

Based on the Fermi architecture, these deploy GPUs, with a wide variety of oIL & GAS

GPUs feature up to 665 gigaflops of double rackmount and blade systems and with

precision performance, 1 teraflop of single remote monitoring and management

precision performance, Ecc memory capabilities – enabling large data

error protection, and L1 and L2 caches. center, scale-out deployments.

TEcHNIcAL SPEcIFIcATIoNS

Tesla M2090 MoLEcULAr DyNAMIcS

Peak double precision floating 665 Gigaflops Tesla M2070 / M2075

point performance Tesla M2050

515 Gigaflops

Peak single precision floating 1331 Gigaflops 515 Gigaflops

point performance