Re:The end-to-end refresh of our server hardware fleet
by LeifCarrotson on Saturday March 11, 2017 @08:44AM
(#13845392)
The whole point of refreshing the hardware fleet from Mk 1 to Mk 2 is that it has a practical basis that they will benefit from.
> Big Basin can train models that are 30 percent larger because of the availability of greater arithmetic throughput and a memory size increase from 12 GB to 16 GB. In tests with popular image classification models like ResNet-50, we were able to reach almost 100 percent improvement in throughput compared with Big Sur
Mk 2 is better than the Mk 1 in several important ways. They're not creating Mk 2 for no reason!
> Big Basin can train models that are 30 percent larger because of the availability of greater arithmetic throughput and a memory size increase from 12 GB to 16 GB. In tests with popular image classification models like ResNet-50, we were able to reach almost 100 percent improvement in throughput compared with Big Sur
Mk 2 is better than the Mk 1 in several important ways. They're not creating Mk 2 for no reason!