80 citations · 136 across the 7 of their papers we have counts for
8 papers
Schrödinger's FP: Dynamic Adaptation of Floating-Point Containers for Deep Learning Training
Miloš Nikolić, Enrique Torres Sanchez, Jiahui Wang +5
The transfer of tensors from/to memory during neural network training dominates time and energy. To improve energy efficiency and performance, research has been exploring ways to u…
Mokey: Enabling Narrow Fixed-Point Inference for Out-of-the-Box Floating-Point Transformer Models
Ali Hadi Zadeh, Mostafa Mahmoud, Ameer Abdelhadi +1
Increasingly larger and better Transformer models keep advancing state-of-the-art accuracy and capability for Natural Language Processing applications. These models demand more com…
APack: Off-Chip, Lossless Data Compression for Efficient Deep Learning Inference
Alberto Delmas Lascorz, Mostafa Mahmoud, Andreas Moshovos
Data accesses between on- and off-chip memories account for a large fraction of overall energy consumption during inference with deep learning networks. We present APack, a simple…
FPRaker: A Processing Element For Accelerating Neural Network Training
Omar Mohamed Awad, Mostafa Mahmoud, Isak Edo +5
We present FPRaker, a processing element for composing training accelerators. FPRaker processes several floating-point multiply-accumulation operations concurrently and accumulates…
TensorDash: Exploiting Sparsity to Accelerate Deep Neural Network Training and Inference
Mostafa Mahmoud, Isak Edo, Ali Hadi Zadeh +4
TensorDash is a hardware level technique for enabling data-parallel MAC units to take advantage of sparsity in their input operand streams. When used to compose a hardware accelera…
Laconic Deep Learning Computing
Sayeh Sharify, Mostafa Mahmoud, Alberto Delmas Lascorz +2
We motivate a method for transparently identifying ineffectual computations in unmodified Deep Learning models and without affecting accuracy. Specifically, we show that if we deco…
Bit-Tactical: Exploiting Ineffectual Computations in Convolutional Neural Networks: Which, Why, and How
Alberto Delmas, Patrick Judd, Dylan Malone Stuart +5
We show that, during inference with Convolutional Neural Networks (CNNs), more than 2x to $8x ineffectual work can be exposed if instead of targeting those weights and activations…
Memory Controller Design Under Cloud Workloads
Mostafa Mahmoud, Andreas Moshovos
This work studies the behavior of state-of-the-art memory controller designs when executing scale-out workloads. It considers memory scheduling techniques, memory page management p…