Comparison of processing performance and architectural efficiency metrics for FPGAs and GPUs in 3D Ultrasound Computer Tomography

Abstract

With the rise of heterogeneous computing architectures, application developers are confronted with a multitude of hardware platforms and the challenge of identifying the most suitable processing platform for their application. Strong competitors for the acceleration of 3D Ultrasound Computer Tomography, a medical imaging method for early breast cancer diagnosis, are GPU and FPGA devices. In this work, we evaluate processing performance and efficiency metrics for current FPGA and GPU devices. We compare topnotch devices from the 40 nm generation as well as FPGA and GPU devices, which draw the same amount of power. For our two benchmark algorithms, the results show that if power consumption is not considered the GPU and the FPGA give both, a similar processing performance and processing efficiency per transistor. However, if the power budget is limited to a similar value, the FPGA performs between six and eight times better than the GPU.

Software-based high-level synthesis design of FPGA beamformers for synthetic aperture imaging

“…as such, it is becoming possible to realize fast porting of high-level c-based parallel kernels, typically designed for execution on cPU and GPU, into FPGa reconfigurable substrates without the use of highly specialized domain-specific languages. at the same time, these software-designed rTl descriptions are capable of achieving real-time performance on FPGas, while presenting one order of magnitude superior power efficiency levels when compared with GPU-based implementations [20].…”

A Survey of Methods for Analyzing and Improving GPU Energy Efficiency

“…They observe that the GPU outperforms the ASIC and the CPU in energy efficiency (performance per watt); however, its energy efficiency is an order of magnitude less than that of the FPGA. Birk et al [2012] compare the performance and energy efficiency of a GPU and an FPGA for 3D ultrasound computer tomography which is used for medical imaging. They observe that the performance of the GPU is comparable with that of the FPGA, however, the FPGA offer much better energy efficiency.…”

“…Similarly, some authors have found FPGAs to be more energy-efficient than GPUs [Kestur et al 2010;Hefenbrock et al 2010;Baker et al 2007;Thomas et al 2009;Pauwels et al 2012;Birk et al 2012;Hussain et al 2011;Hamada et al 2009;Gohringer et al 2011;Zou et al 2012;Benkrid et al 2012;De Schryver et al 2011;Lange et al 2009;Richardson et al 2010;Van Essen et al 2012;Brodtkorb et al 2010;Chau et al 2013;Cong and Zou 2009;Llamocca et al 2011;Cong et al 2011;Waidyasooriya et al 2012;Chow et al 2012;Struyf et al 2014], while others have found GPUs to be more energy efficient [Duan et al 2011]. Similarly, some researchers observe other computing systems such as Cell, DSP (digital signal processor) or ASIC to be more energy efficient than GPUs [Chung et al 2010;Baker et al 2007;Benkrid et al 2012;Mu et al 2011;Pedram et al 2012].…”

Untitled

“…Similarly, some authors have found FPGAs to be more energy efficient than GPUs [Kestur et al 2010;Hefenbrock et al 2010;Baker et al 2007;Thomas et al 2009;Pauwels et al 2012;Birk et al 2012;Hussain et al 2011;Hamada et al 2009;Gohringer et al 2011;Zou et al 2012;Benkrid et al 2012;De Schryver et al 2011;Lange et al 2009;Richardson et al 2010;Van Essen et al 2012;Brodtkorb et al 2010;Chau et al 2013;Cong and Zou 2009;Llamocca et al 2011;Cong et al 2011;Waidyasooriya et al 2012;Chow et al 2012;Struyf et al 2014], while others have found GPUs to be more energy efficient [Duan et al 2011]. Similarly, some researchers observe other computing systems such as Cell, DSP (digital signal processor), or ASIC to be more energy efficient than GPUs [Chung et al 2010;Baker et al 2007;Benkrid et al 2012;Mu et al 2011;Pedram et al 2012].…”