Gabriel Marcano

Olivia Weng, Gabriel Marcano, Vladimir Loncar et al.

Deep neural networks use skip connections to improve training convergence. However, these skip connections are costly in hardware, requiring extra buffers and increasing on- and off-chip memory utilization and bandwidth requirements. In this paper, we show that skip connections can be optimized for hardware when tackled with a hardware-software codesign approach. We argue that while a network's skip connections are needed for the network to learn, they can later be removed or shortened to provide a more hardware efficient implementation with minimal to no accuracy loss. We introduce Tailor, a codesign tool whose hardware-aware training algorithm gradually removes or shortens a fully trained network's skip connections to lower their hardware cost. Tailor improves resource utilization by up to 34% for BRAMs, 13% for FFs, and 16% for LUTs for on-chip, dataflow-style architectures. Tailor increases performance by 30% and reduces memory bandwidth by 45% for a 2D processing element array architecture.

Matthew McClintic, Devon Maloney, Michael Scires et al.

We demonstrate an attack on the secure bootchain of the Nintendo 3DS in order to gain early code execution. The attack utilizes the block shuffling vulnerability of the ECB cipher mode to rearrange keys in the Nintendo 3DS's encrypted keystore. Because the shuffled keys will deterministically decrypt the encrypted firmware binary to incorrect plaintext data and execute it, and because the device's memory contents are kept between hard reboots, it is possible to reliably reach a branching instruction to a payload in memory. This payload, due to its execution by a privileged processor and its early execution, is able to extract the hash of hardware secrets necessary to decrypt the device's encrypted keystore and set up a persistent exploit of the system.