TFApprox: Towards a Fast Emulation of DNN Approximate Hardware Accelerators on GPU
This work addresses the problem of inefficient software emulation for hardware designers and researchers, enabling faster evaluation of approximate circuits without expensive prototyping, though it is incremental as it builds on existing GPU and TensorFlow frameworks.
The paper tackled the slow emulation of approximate hardware accelerators for deep neural networks on GPUs by proposing a method using look-up tables and texture memory, achieving a 200x speedup in inference time compared to an optimized CPU version on networks like ResNet.
Energy efficiency of hardware accelerators of deep neural networks (DNN) can be improved by introducing approximate arithmetic circuits. In order to quantify the error introduced by using these circuits and avoid the expensive hardware prototyping, a software emulator of the DNN accelerator is usually executed on CPU or GPU. However, this emulation is typically two or three orders of magnitude slower than a software DNN implementation running on CPU or GPU and operating with standard floating point arithmetic instructions and common DNN libraries. The reason is that there is no hardware support for approximate arithmetic operations on common CPUs and GPUs and these operations have to be expensively emulated. In order to address this issue, we propose an efficient emulation method for approximate circuits utilized in a given DNN accelerator which is emulated on GPU. All relevant approximate circuits are implemented as look-up tables and accessed through a texture memory mechanism of CUDA capable GPUs. We exploit the fact that the texture memory is optimized for irregular read-only access and in some GPU architectures is even implemented as a dedicated cache. This technique allowed us to reduce the inference time of the emulated DNN accelerator approximately 200 times with respect to an optimized CPU version on complex DNNs such as ResNet. The proposed approach extends the TensorFlow library and is available online at https://github.com/ehw-fit/tf-approximate.