gpu slower than cpu for neural networks in matlab -


i have following code:

tic; h = rand(100, 1000);     f = rand(1, 1000);     net = newff(h, f, [30, 10], { 'tansig' 'tansig'}, 'traingdx', 'learngdm', 'mse');      net.trainparam.epochs = 400;     net.performparam.regularization = 0.05;      net.divideparam.trainratio = 1;     net.divideparam.valratio = 0;     net.divideparam.testratio = 0;      net.trainparam.showwindow = 0;     net.trainparam.showcommandline = 0;      % net = train(net, h, f, 'usegpu', 'yes', 'showresources', 'yes'); % line 1     net = train(net, h, f, 'showresources', 'yes'); % line 2  toc;

with line 2 uncommented 

computing resources: gpu device #1, geforce 800m  elapsed time 5.084222 seconds.

and line 1 uncommented 

computing resources: mex2  elapsed time 1.870803 seconds.

why gpu slower cpu? gpu properties:

cudadevice properties:                    name: 'geforce 800m'                  index: 1      computecapability: '2.1'         supportsdouble: 1          driverversion: 6         toolkitversion: 5     maxthreadsperblock: 1024       maxshmemperblock: 49152     maxthreadblocksize: [1024 1024 64]            maxgridsize: [65535 65535 65535]              simdwidth: 32            totalmemory: 2.1475e+09             freememory: 1.9886e+09    multiprocessorcount: 1           clockratekhz: 1475000            computemode: 'default'   gpuoverlapstransfers: 1 kernelexecutiontimeout: 1       canmaphostmemory: 1        devicesupported: 1         deviceselected: 1`


-

Comments

Post a Comment

Popular posts from this blog

serialization - Convert Any type in scala to Array[Byte] and back -

i have following question: i have variable value in program declared value. i want convert value byte array.. how can serialize byte array , back? found examples related other types such double or int, not any. this should need. it's pretty similar how 1 in java. import java.io.{bytearrayinputstream, bytearrayoutputstream, objectinputstream, objectoutputstream} object serialization extends app { def serialise(value: any): array[byte] = { val stream: bytearrayoutputstream = new bytearrayoutputstream() val oos = new objectoutputstream(stream) oos.writeobject(value) oos.close stream.tobytearray } def deserialise(bytes: array[byte]): = { val ois = new objectinputstream(new bytearrayinputstream(bytes)) val value = ois.readobject ois.close value } println(deserialise(serialise("my test"))) println(deserialise(serialise(list(1)))) println(deserialise(serialise(map(1 -> 2)))) println(deserialise(seri
Read more

matplotlib support failed in PyCharm on OSX -

any idea how fix ? how can debug might problem ? not seem : https://youtrack.jetbrains.com/issue/py-15520 i in console: python 2.6.9 (unknown, jul 14 2015, 19:46:31) [gcc 4.2.1 compatible apple llvm 6.0 (clang-600.0.39)] on darwin >>> import matplotlib.pyplot plt matplotlib support failed the backend : >>> import matplotlib >>> print(matplotlib.rcparams['backend']) macosx
Read more

python - Matplotlib: TypeError: 'AxesSubplot' object is not callable -

i new programming please pardon me if rookie mistake. i writing python script have defined functions create subplots of distributions. i have slider widget end user change value of variables(mu , sigma) , see effect on distribution. here's code import numpy np print np.__version__ import scipy print scipy.__version__ scipy.stats import norm, lognorm, uniform import matplotlib print matplotlib.__version__ import matplotlib.pyplot plt matplotlib.widgets import slider, button, radiobuttons, checkbuttons #####calculating mean , standard deviation##### global mu_a1 mu_a1 = 1 global mu_b1 mu_b1 = 10 global mu_c1 mu_c1 = -13 global sigma_a1 sigma_a1 = 0.14 global sigma_b1 sigma_b1 = 1.16 global sigma_c1 sigma_c1 = 2.87 mu_x01 = -11 sigma_x01 = 1.9 #####_____##### #####generating random data##### a1 = 0.75*mu_a1 + (1.25 - 0.75)*sigma_a1*np.random.sample(10000) b1 = 8*mu_b1 + (12 - 8)*sigma_b1*np.random.sample(10000) c1 = -12*mu_c1 + 2*sigma_c1*np.random.sample(10000)
Read more