id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_agwx3vvgsne2zexujz6g3vtzje	Sándor Szénási	Solving the inverse heat conduction problem using NVLink capable Power architecture	2017	20	.pdf	application/pdf	7676	884	70	Keywords GPU, CUDA, Inverse heat conduction problem, Heat transfer, Parallelisation, Dataparallel algorithm, Simulation, NVLink, Graphics accelerator, Optimisation How to cite this article Szénási (2017), Solving the inverse heat conduction problem using NVLink capable Power architecture. POWER8 CPUs and NVIDIA's Pascal GPUs. Data transfer between the host and device speed-up for the GPU implementation (based on other parameters, like population size, Table 1 Runtime values for different population sizes with the GeForce Titan Black cards. Table 3 Runtime of the DHCP solver for different population sizes and CPU core counts (thread the CPU would be faster in the case of small population sizes (where the GPUs cannot take Figure 7 Memory transfer time (µs) for different population sizes with GeForce Titan Black cards. Figure 8 Memory transfer time (µs) for different population sizes with P100 cards. • In the case of the IHCP, the runtime of both the CPU and the GPU implementations	./cache/work_agwx3vvgsne2zexujz6g3vtzje.pdf	./txt/work_agwx3vvgsne2zexujz6g3vtzje.txt