3 ms·
Artificial Learning's largest designs are 180nm tech so some way from Moore's Law limits. Efficiency advantage over GPGPU implementation calculated as better t
by ArtificialLearn 12y ago
Artificial Learning's largest designs are 180nm tech so some way from Moore's Law limits. Efficiency advantage over GPGPU implementation calculated as better than 10,000 fold in terms of learning updates per second per watt per kg.