Thursday, September 28, 2017

Nvidia's Tesla V100 GPU gets wide support as server merchants eye AI workloads

Nvidia's Tesla V100 GPU likens to 100 CPUs. That implies as far as possible is lifted for AI workloads.






The real server merchants are arranging behind Nvidia's Tesla V100 GPU quickening agent in a move that is required to make manmade brainpower and machine learning workloads more standard. 

Dell EMC, HPE, IBM and Supermicro sketched out servers on Nvidia's most recent GPU quickening agents, which depend on the Volta engineering from the illustrations chip creator. Nvidia's V100 GPUs have more than 120 teraflops of profound learning execution per GPU. That throughput successfully takes as far as possible off AI workloads. 

In a blog entry, IBM's Brad McCredie, VP of the Big Blue's intellectual framework improvement, noticed that Nvidia with the V100 and its NVLINK PCI-Express 4 and Memory Coherence innovation brings "extraordinary inside data transmission" to AI-enhanced frameworks. 

The V100-based frameworks include: 

  • Dell EMC's PowerEdge R740, which underpins up to three V100 GPUs for PCIe and two higher end frameworks with the R740XD, and C4130.
  • HPE's Apollo 65000, which will bolster up to 8 V100 GPUs for PCIe, and the ProLiant DL380 framework supporting up to three V100 GPUs.
  • IBM's Power System with the Power 9 processor will bolster numerous V100 GPUs. IBM will reveal its Power 9-based frameworks in the not so distant future.
  • Supermicro likewise has a progression of workstations and servers worked with the V100.
  • Inspur, Lenovo and Huawei likewise propelled frameworks in light of the V100.




No comments:

Post a Comment