Below is the screen output of the Python script run_mgs_timings.py
comparing the running times of the modified Gram-Schmidt method 
for the NVIDIA Tesla K20C GPU with a 2.60GHz CPU,
done with complex double precision arithmetic.

$ python run_mgs_timings.py
running the modified Gram-Schmidt method
type precision level (d, dd, or qd) : d
give start dimension (>= 2) : 16
give stop dimension (<= 256) : 256
give the step : 16
give frequency per dimension (>= 1) : 10000
give the name of the output file : run_mgs_d_times.txt
running for dimension 16 and frequency 10000 on CPU
['real 2.25', 'user 2.23', 'sys 0.02']
running for dimension 16 and frequency 10000 on GPU
['real 1.23', 'user 0.45', 'sys 0.63']
running for dimension 32 and frequency 10000 on CPU
['real 16.31', 'user 16.20', 'sys 0.08']
running for dimension 32 and frequency 10000 on GPU
['real 6.51', 'user 3.88', 'sys 2.50']
running for dimension 48 and frequency 10000 on CPU
['real 53.12', 'user 52.88', 'sys 0.15']
running for dimension 48 and frequency 10000 on GPU
['real 10.24', 'user 6.40', 'sys 3.67']
running for dimension 64 and frequency 10000 on CPU
['real 123.70', 'user 123.19', 'sys 0.31']
running for dimension 64 and frequency 10000 on GPU
['real 2.55', 'user 1.66', 'sys 0.71']
running for dimension 80 and frequency 10000 on CPU
['real 239.62', 'user 238.34', 'sys 0.88']
running for dimension 80 and frequency 10000 on GPU
['real 22.41', 'user 13.94', 'sys 8.20']
running for dimension 96 and frequency 10000 on CPU
['real 409.30', 'user 407.17', 'sys 1.47']
running for dimension 96 and frequency 10000 on GPU
['real 28.64', 'user 17.94', 'sys 10.50']
running for dimension 112 and frequency 10000 on CPU
['real 649.00', 'user 645.68', 'sys 2.29']
running for dimension 112 and frequency 10000 on GPU
['real 36.59', 'user 22.38', 'sys 13.87']
running for dimension 128 and frequency 10000 on CPU
['real 962.17', 'user 957.79', 'sys 2.71']
running for dimension 128 and frequency 10000 on GPU
['real 45.29', 'user 28.02', 'sys 16.99']
running for dimension 144 and frequency 10000 on CPU
['real 1363.57', 'user 1358.10', 'sys 3.30']
running for dimension 144 and frequency 10000 on GPU
['real 59.48', 'user 36.73', 'sys 22.45']
running for dimension 160 and frequency 10000 on CPU
['real 1867.36', 'user 1858.74', 'sys 5.56']
running for dimension 160 and frequency 10000 on GPU
['real 70.54', 'user 42.69', 'sys 27.56']
running for dimension 176 and frequency 10000 on CPU
['real 2480.18', 'user 2469.74', 'sys 6.24']
running for dimension 176 and frequency 10000 on GPU
['real 84.26', 'user 51.40', 'sys 32.52']
running for dimension 192 and frequency 10000 on CPU
['real 3221.12', 'user 3206.63', 'sys 8.81']
running for dimension 192 and frequency 10000 on GPU
['real 97.62', 'user 58.97', 'sys 38.17']
running for dimension 208 and frequency 10000 on CPU
['real 4080.77', 'user 4064.61', 'sys 9.46']
running for dimension 208 and frequency 10000 on GPU
['real 112.19', 'user 67.24', 'sys 44.56']
running for dimension 224 and frequency 10000 on CPU
['real 5099.98', 'user 5077.51', 'sys 13.65']
running for dimension 224 and frequency 10000 on GPU
['real 128.11', 'user 76.88', 'sys 50.85']
running for dimension 240 and frequency 10000 on CPU
['real 6253.65', 'user 6230.79', 'sys 12.75']
running for dimension 240 and frequency 10000 on GPU
['real 146.31', 'user 88.93', 'sys 56.93']
running for dimension 256 and frequency 10000 on CPU
['real 7575.85', 'user 7549.78', 'sys 14.52']
running for dimension 256 and frequency 10000 on GPU
['real 164.33', 'user 99.73', 'sys 64.22']
[jan@kepler Timings]$ 
