We recently bought an NVIDIA ARM Devkit with the BlueField 2 DPU, and I am trying to connect it into the rest of our network (once it arrives). The rest of our network is based on QSFP28 100GbE with a 32 * 100GbE switch at its core.
What would the best way be to connect the DPUs...
I am currently in the process of planning an Infiniband Network for our new GPU cluster. We want to be able to run simulation / AI workloads across the big GPU nodes and are hence looking to connect everything with a 100 Gb Infiniband-Network. At first, I was thinking of the...
I am just about to order some new GPU servers and am faced with the choice whether to get the 8 * A40, or 8 * A6000 in a 4U server chassis instead. Has anyone of you got any experience comparing the two for such 8 GPU builds? Is there an inherent advantage of one over the other in...
As @Agremlin has already mentioned, your GPUs will still be faster than the 9900k.
There is only one major difficulty. You need to find a way to get the ML-frameworks, which support your older version of CUDA. Keras can work with Tensorflow, Theano or CNTK, you just need to configure the...
I am currently thinking about adding SLI-Bridges to an 8 RTX2080 Ti build in a Supermicro SuperServer 4028GR-TR for deep learning applications. NVLink bridges available for the RTX series only come in a 3/4-slot spacing, which is too wide in a server setting.
Looking at the NVLink...
good luck for your dissertation! Have you also considered using a state-of-the-art setup such as GANs coupled with Reinforcement Learning?
We use it for different kind of spatial data and early results are looking pretty promising.