High-Performance 3D Compressive Sensing MRI Reconstruction Using Many-Core Architectures

Table 2

Summary of performance across different hardware.

	Execution time	Speedup
Core i7 processor 1 Core, Not Optimized	175 s	1.0x
Core i7 processor 1 Core, Optimized	116 s	1.5x
Literature Best [19]	~100 s	1.7x
Core i7 processor 4 Cores	56 s	3.1x
Core i7 processor Dual-Socket 12 Cores	35 s	5.0x
NVIDIA Tesla C2050	19 s	9.2x
Intel's Knights Ferry (32 Cores)	12 s	14.5x
Research CMP Simulation (128 Cores)	6 s	29.1x