NVIDIA Titan RTX Deep Learning Benchmarks
As we continue to innovate on our review format, we are now adding deep learning benchmarks. In future reviews, we will add more results to this data set.
ResNet-50 Inferencing using Tensor Cores
ImageNet is an image classification database launched in 2007 designed for use in visual object recognition research. Organized by the WordNet hierarchy, each node (or category of specific nouns) are represented by hundreds of image examples.
In our benchmarks, we use batches of 500 and vary the batch size by 16, 32, 64, and 128 respectively.
We start with Turing’s new INT8 mode.
In all cases, the Titan RTX is close to double the results compared to the RTX 2070 which is more of a budget example. Larger batch sizes improve performance if the GPU has the memory to handle that. In the case of the NVIDIA Titan RTX which has three times the amount of memory versus the 8GB found on the RTX 2070. Our next test will show why that increased memory footprint is important.
Training using OpenSeq2Seq (GNMT)
While Resnet-50 is a Convolutional Neural Network (CNN) that is typically used for image classification, Recurrent Neural Networks (RNN) such as Google Neural Machine Translation (GNMT) are used for applications such as real-time language translations.
We should note that other GPU’s we used to like the RTX2060 and RTX2070 could not complete this benchmark due to the lack of installed memory.
As we continue to expand our Deep Learning benchmarks, we will add more data points to this result set.
Next, we are going to look at the NVIDIA Titan RTX power and temperature tests, and then give our final words.