You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Saying that the GTX 1080 > Maxwell Titan X is misleading. The metric should be the time of epoch (or convergence), not forward + backward. The +4GB in the Titan X Maxwell make it much faster than the GTX 1080 for training.
The text was updated successfully, but these errors were encountered:
Speed of a GPU mostly depends on program optimisation, GPU architecture, Memory clock, Type of Memory (Not in memory size), Memory Bandwidth ,PCIe bandwidth, Number of CUDA cores for parallel processing and system clock. Except for CUDA counts and program optimisation (most of the current Deep learning frameworks are more optimize for Titan X) any other department GTX 1080 is better than Maxwell Titan X though Titan X has 12 gb memory but it's slower than 1080's GDDR5X 8 gb memory. Yes in memory demanding task Titan X has advantage over 1080 but that's scenario is pretty rare because in present time 8 gb is far enough memory for graphic card.
The ~15% increase in forward+backward speed in the GTX 1080 is not better than the ~30% increase in batch size in the Titan X Maxwell. Deep Learning training is a memory demanding task.
Saying that the GTX 1080 > Maxwell Titan X is misleading. The metric should be the time of epoch (or convergence), not forward + backward. The +4GB in the Titan X Maxwell make it much faster than the GTX 1080 for training.
The text was updated successfully, but these errors were encountered: