Partitioned Memory Parallel Programming Library (PMLIB)
Based on BSP-RAMP: Partitioned Memory Parallel Programming Framework
			Prof. Subodh Kumar	Prof. Sorav Bansal
Tarun Beri
Indian Institute of Technology, Delhi
		In-Memory Data Compression
Experimental Configuration									2
Four 64-bit Ubuntu Linux 8.04.2 Eight Core Machines [Intel Xeon CPU E5450 3.00 GHz with 16 GB physical memory];									1
Two machines have one Tesla C1060 GPU Card each									1
Watch Dog/Kernel Execution Timeout Disabled; mpiexec with 4 processes on different machines;
CUDA Version 3.1; OpenMPI Version 1.78; OpenMP Version 3.0; gcc Version 4.2.4
Equal task partitioning among all processing elements
Data Size (in MB)	Compression Time (in seconds)
	Serial Task	1 host CPU Task	1 host GPU Task	1 host Task CPU+GPU	Cluster CPU Task	Cluster GPU Task	Cluster Task CPU+GPU
1024	141.78	18.57			24.68
2048	284.14	37.14			52.26
Comparison of Serial and PMLIB Tasks
Comparison of PMLIB Tasks
Comparison of Serial Task, Fastest PMLIB Task and Slowest PMLIB Task
Data Size (in MB)	Compression Time (in seconds)
	Serial Task	1 host CPU Task	1 host GPU Task	1 host Task CPU+GPU	Cluster CPU Task	Cluster GPU Task	Cluster Task CPU+GPU	Fastest PMLIB Task	Slowest PMLIB Task
1024	141.78	18.57			24.68			18.57	24.68
2048	284.14	37.14			52.26			37.14	52.26
								Fastest LPC Task	Slowest LPC Task
Data Size (in MB)	Compression Time (in seconds)			% Speedup of PMLIB Task over Serial Task
	Serial Task	Fastest PMLIB Task	Slowest PMLIB Task	Fastest PMLIB Task	Slowest PMLIB Task
1024	141.78	18.57	24.68	763.49	574.47
2048	284.14	37.14	52.26	765.05	543.70
							Max PMLIB Speedup over Serial Task
							7.65x
		PMLIB Tasks slower than Serial Task		PMLIB Tasks faster than Serial Task