Thursday , 21 June 2018
Home >> C >> Chips and Components >> Nvidia licensing kills off Geforce for data centres

Nvidia licensing kills off Geforce for data centres

Nvidia has updated its End User Licence Agreement (EULA) to force data centres users away from its consumer-grade GPUs and towards the (much) more costly workstation and server products.

Floating point performance has long been the main differentiator between Nvidia’s Geforce, Quadro and Tesla families. While the latter two have been able to use full double-precision floating point capabilities, the consumer hardware was artificially limited to be of less use for general purpose GPU (GPGPU) accelerated workloads.

Shifts in GPGPU workloads towards uses like big data and machine learning, though, have made consumer hardware more usable for data centre owners, which has led to rooms full of Geforce and Titan GPUs being used for professional purposes.

Nvidia, of course, would rather its customers pay the full £10,000 for a Tesla V100 GPU rather than £800 for a GTX 1080 Ti, and is enforcing that through the new EULA. It states:

‘No data centre deployment: The software is not licensed for data centre deployment, except that blockchain processing in a data centre is permitted.’

Large-scale cryptocurrency miners can continue to use consumer hardware – Nvidia doesn’t want to lock itself out of this market – but any professional applications are now banned from doing so.

The new rules only apply to software, rather than explicitly invalidating the GPUs themselves from being used. Users who refuse to update to the newest drivers can continue to utilise Geforce and Titan cards, but will be sacrificing future updates and support.

Further reading

<!–

–>

  • <!–

  • Save this article

  • –>

close
==[ Click Here 1X ] [ Close ]==