Nvidia has just fired the first shots against Intel in a blog post on its official site.
The blog post, titled Correcting Intel’s Deep Learning Benchmark Mistakes claims that Intel was using outdated benchmarks when showing off its new Knights Landing Xeon Phi processors.
In particular Nvidia took issue with Intel’s claims that:
- Xeon Phi is 2.3 times faster in training than GPUs.
- Xeon Phi offers 38 per cent better scaling than GPUs across nodes.
- Xeon Phi delivers 50 times scaling on 128 nodes.
Nvidia did finish the post by welcoming the Intel to the deep learning market but saying that they “should get their facts straight”.
More gaming news
How much should we be paying for games?
South Africa’s first VR arcade is now open
Overwatch could be getting a free-to-play mode
Forum discussion
Join the conversation