Habana separates machine learning training from inference
By building different processors to address machine learning training and inference, Habana has been able to achieve impressive results as per its benchmarks
1 of 3 Habana
Architecture for machine learning training and inference
Different machine learning workloads are better suited for different processors with different architectures, according to Habana
2 of 3 The Microprocessor report
Habana smokes inference incumbents
Habana's Goya compared to Nvidia's Tesla and other processors on inference workloads
3 of 3 Habana
Habana's Goya processor compared to CPU and GPU
Results based for the ResNet-50 inference benchmark. Note results also include power efficiency
Related Galleries
Holiday wallpaper for your phone: Christmas, Hanukkah, New Year's, and winter scenes
Related Galleries
Holiday wallpaper for your phone: Christmas, Hanukkah, New Year's, and winter scenes
21 Photos
Winter backgrounds for your next virtual meeting
Related Galleries
Winter backgrounds for your next virtual meeting
21 Photos
Holiday backgrounds for Zoom: Christmas cheer, New Year's Eve, Hanukkah and winter scenes
Related Galleries
Holiday backgrounds for Zoom: Christmas cheer, New Year's Eve, Hanukkah and winter scenes
21 Photos
Hyundai Ioniq 5 and Kia EV6: Electric vehicle extravaganza
Related Galleries
Hyundai Ioniq 5 and Kia EV6: Electric vehicle extravaganza
26 Photos
A weekend with Google's Chrome OS Flex
Related Galleries
A weekend with Google's Chrome OS Flex
22 Photos
Cybersecurity flaws, customer experiences, smartphone losses, and more: ZDNet's research roundup
Related Galleries
Cybersecurity flaws, customer experiences, smartphone losses, and more: ZDNet's research roundup
8 Photos
Inside a fake $20 '16TB external M.2 SSD'