Picture for Canberk Sönmez

Canberk Sönmez

Accuracy Boosters: Epoch-Driven Mixed-Mantissa Block Floating-Point for DNN Training

Add code
Nov 22, 2022
Figure 1 for Accuracy Boosters: Epoch-Driven Mixed-Mantissa Block Floating-Point for DNN Training
Figure 2 for Accuracy Boosters: Epoch-Driven Mixed-Mantissa Block Floating-Point for DNN Training
Figure 3 for Accuracy Boosters: Epoch-Driven Mixed-Mantissa Block Floating-Point for DNN Training
Figure 4 for Accuracy Boosters: Epoch-Driven Mixed-Mantissa Block Floating-Point for DNN Training
Viaarxiv icon

Scale-out Systolic Arrays

Add code
Mar 22, 2022
Figure 1 for Scale-out Systolic Arrays
Figure 2 for Scale-out Systolic Arrays
Figure 3 for Scale-out Systolic Arrays
Figure 4 for Scale-out Systolic Arrays
Viaarxiv icon