Skip to content

P
pytensor
提交 2ad63271 authored 作者: Frederic Bastien's avatar Frederic Bastien
浏览文件
操作

fix some test case of fusion badbly tested on cpu and fix some cuda test disabled by default.

上级 57e2b381
隐藏空白字符变更
内嵌 并排
正在显示 包含 37 行增加 和 40 行删除
theano/tensor/tests/test_opt.py
浏览文件 @ 2ad63271
...@@ -614,34 +614,34 @@ class test_fusion(unittest.TestCase): ...@@ -614,34 +614,34 @@ class test_fusion(unittest.TestCase):
izv = theano._asarray(my_init(shp,num=70),dtype='int32') izv = theano._asarray(my_init(shp,num=70),dtype='int32')
fwx=fw+fx fwx=fw+fx
cases = [ cases = [
(fx+fy+fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+fzv,'float32'),#1 (fx+fy+fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+fzv,'float32'),#0
(fx*fy*fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv*fyv*fzv,'float32'), (fx*fy*fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv*fyv*fzv,'float32'),#1
(fx+fy*fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv*fzv,'float32'), (fx+fy*fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv*fzv,'float32'),
(fx*fy+fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv*fyv+fzv,'float32'), (fx*fy+fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv*fyv+fzv,'float32'),
(fw+fx+fy+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),#5 (fw+fx+fy+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),
((fw+fx)+(fy+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'), ((fw+fx)+(fy+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),#5
(((fw+fx)+fy)+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'), (((fw+fx)+fy)+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),
((fw+(fx+fy))+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'), ((fw+(fx+fy))+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),
((fw+(fx+fy)+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'), ((fw+(fx+fy)+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),
(fw+(fx+(fy+fz)),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),#10 (fw+(fx+(fy+fz)),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),
((fw+fx)+(fy+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'), ((fw+fx)+(fy+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv,'float32'),#10
(fw*fx*fy*fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv*fxv*fyv*fzv,'float32'), (fw*fx*fy*fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv*fxv*fyv*fzv,'float32'),
(fw+fx*fy*fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv*fyv*fzv,'float32'), (fw+fx*fy*fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv*fyv*fzv,'float32'),
(fx+fy*fz*fx,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv*fzv*fxv,'float32'), (fx+fy*fz*fx,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv*fzv*fxv,'float32'),
(fx*fy+fz+fy,(fx,fy,fz),(fxv,fyv,fzv),1,fxv*fyv+fzv+fyv,'float32'),#15 (fx*fy+fz+fy,(fx,fy,fz),(fxv,fyv,fzv),1,fxv*fyv+fzv+fyv,'float32'),
(fx*fy*fz*fw+fx+fy+fz+fw,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fxv*fyv*fzv*fwv+fxv+fyv+fzv+fwv,'float32'), (fx*fy*fz*fw+fx+fy+fz+fw,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fxv*fyv*fzv*fwv+fxv+fyv+fzv+fwv,'float32'),#15
#test with constant #test with constant
((fw+fx)+(fy+fz)+2,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'), ((fw+fx)+(fy+fz)+2,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'),
(((fw+fx)+2+fy)+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'), (((fw+fx)+2+fy)+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'),
((fw+(fx+2+fy))+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'), ((fw+(fx+2+fy))+fz,(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'),
((fw+(fx+fy)+2+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'),#20 ((fw+(fx+fy)+2+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'),
(fw+(fx+(fy+fz)+2),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'), (fw+(fx+(fy+fz)+2),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'),#20
(2+(fw+fx)+(fy+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'), (2+(fw+fx)+(fy+fz),(fw,fx,fy,fz),(fwv,fxv,fyv,fzv),1,fwv+fxv+fyv+fzv+2,'float32'),
#mix float32 and float64 #mix float32 and float64
(2+(dw+fx)+(fy+fz),(dw,fx,fy,fz),(dwv,fxv,fyv,fzv),1,dwv+fxv+fyv+fzv+2,'float64'), (2+(dw+fx)+(fy+fz),(dw,fx,fy,fz),(dwv,fxv,fyv,fzv),1,dwv+fxv+fyv+fzv+2,'float64'),
(2+(fw+dw)+(fy+fz),(fw,dw,fy,fz),(fwv,dwv,fyv,fzv),1,fwv+dwv+fyv+fzv+2,'float64'), (2+(fw+dw)+(fy+fz),(fw,dw,fy,fz),(fwv,dwv,fyv,fzv),1,fwv+dwv+fyv+fzv+2,'float64'),
(2+(fw+fx)+(dw+fz),(fw,fx,dw,fz),(fwv,fxv,dwv,fzv),1,fwv+fxv+dwv+fzv+2,'float64'),#25 (2+(fw+fx)+(dw+fz),(fw,fx,dw,fz),(fwv,fxv,dwv,fzv),1,fwv+fxv+dwv+fzv+2,'float64'),
(2+(fw+fx)+(fy+dw),(fw,fx,fy,dw),(fwv,fxv,fyv,dwv),1,fwv+fxv+fyv+dwv+2,'float64'), (2+(fw+fx)+(fy+dw),(fw,fx,fy,dw),(fwv,fxv,fyv,dwv),1,fwv+fxv+fyv+dwv+2,'float64'),#25
#test when their is other op then elemwise. #test when their is other op then elemwise.
#the good output for the next test. #the good output for the next test.
# (Pdb) p f.maker.env.toposort() # (Pdb) p f.maker.env.toposort()
...@@ -657,33 +657,33 @@ class test_fusion(unittest.TestCase): ...@@ -657,33 +657,33 @@ class test_fusion(unittest.TestCase):
#test other elemwise op #test other elemwise op
(fx+fy+cos(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.cos(fzv),'float32'), (fx+fy+cos(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.cos(fzv),'float32'),
(fx+fy+cosh(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.cosh(fzv),'float32'), (fx+fy+cosh(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.cosh(fzv),'float32'),
(fx+fy+abs(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.absolute(fzv),'float32'),#30 (fx+fy+abs(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.absolute(fzv),'float32'),
(ix+iy+abs(iz),(ix,iy,iz),(ixv,iyv,izv),1,ixv+iyv+numpy.absolute(izv),'int32'), (ix+iy+abs(iz),(ix,iy,iz),(ixv,iyv,izv),1,ixv+iyv+numpy.absolute(izv),'int32'),#30
(fx+fy+theano.tensor.log(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.log(fzv),'float32'), (fx+fy+theano.tensor.log(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.log(fzv),'float32'),
(fx+fy+theano.tensor.log2(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.log2(fzv),'float32'), (fx+fy+theano.tensor.log2(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.log2(fzv),'float32'),
(fx+fy+theano.tensor.log10(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.log10(fzv),'float32'), (fx+fy+theano.tensor.log10(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.log10(fzv),'float32'),
(fx+fy**fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv**fzv,'float32'),#pow #35 (fx+fy**fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv**fzv,'float32'),#pow
(fx+fy+theano.tensor.exp(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.exp(fzv),'float32'), (fx+fy+theano.tensor.exp(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv+fyv+numpy.exp(fzv),'float32'),#35
(fx-fy-fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv-fzv,'float32'), (fx-fy-fz,(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv-fzv,'float32'),
(fx-(fy/fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv/fzv),'float32'), (fx-(fy/fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv/fzv),'float32'),
(fx-theano.tensor.true_div(fy,2),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv/2),'float32'), (fx-theano.tensor.true_div(fy,2),(fx,fy),(fxv,fyv),1,fxv-(fyv/2),'float32'),
(fx-theano.tensor.true_div(fy,fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv/fzv),'float32'),#40 (fx-theano.tensor.true_div(fy,fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv/fzv),'float32'),
(fx-theano.tensor.int_div(ix*100,iy*1000),(fx,ix,iy),(fxv,ixv,iyv),4,fxv-((ixv*100)//(iyv*1000)),'float64'),#int32 - float32 = float64 #No c_code for int_div (fx-theano.tensor.int_div(ix*100,iy*1000),(fx,ix,iy),(fxv,ixv,iyv),4,fxv-((ixv*100)//(iyv*1000)),'float64'),#int32 - float32 = float64 #No c_code for int_div#40
(fx-(fy/2),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv/2),'float32'), (fx-(fy/2),(fx,fy),(fxv,fyv),1,fxv-(fyv/2),'float32'),
(fx-(fy%fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv%fzv),'float32'), (fx-(fy%fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv%fzv),'float32'),
(fx-(fy>fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv>fzv),'float32'), (fx-(fy>fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv>fzv),'float32'),
(fx-(fy>=fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv>=fzv),'float32'),#45 (fx-(fy>=fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv>=fzv),'float32'),
(fx-(fy<fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv<fzv),'float32'), (fx-(fy<fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv<fzv),'float32'),#45
(fx-(fy<=fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv<=fzv),'float32'),#TODO: bugged on the gpu (fx-(fy<=fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv<=fzv),'float32'),
(fx-(fy==fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv==fzv),'float32'),#TODO: bugged (fx-T.eq(fy,fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv==fzv),'float32'),
(fx-(fy!=fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv!=fzv),'float32'), (fx-T.neq(fy,fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-(fyv!=fzv),'float32'),
(fx-fy+tan(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.tan(fzv),'float32'),#50 (fx-fy+tan(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.tan(fzv),'float32'),
(fx-fy+tanh(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.tanh(fzv),'float32'), (fx-fy+tanh(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.tanh(fzv),'float32'),#50
(fx-fy+sin(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.sin(fzv),'float32'), (fx-fy+sin(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.sin(fzv),'float32'),
(fx-fy+sinh(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.sinh(fzv),'float32'), (fx-fy+sinh(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.sinh(fzv),'float32'),
(fx-fy+theano.tensor.sqr(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+(fzv*fzv),'float32'), (fx-fy+theano.tensor.sqr(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+(fzv*fzv),'float32'),
(fx-fy+theano.tensor.sqrt(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.sqrt(fzv),'float32'),#55 (fx-fy+theano.tensor.sqrt(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.sqrt(fzv),'float32'),
(fx-fy+theano.tensor.inv(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+(1/fzv),'float32'), (fx-fy+theano.tensor.inv(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+(1/fzv),'float32'),#55
(fx-fy+theano.tensor.neg(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+(-fzv),'float32'), (fx-fy+theano.tensor.neg(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+(-fzv),'float32'),
# (fx-fy+theano.tensor.iround(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.round(fzv),'float32'),#TODO: trouble with the output type. To my understanding, numpy and c round fct return the same type as the input. Why we don't do this? # (fx-fy+theano.tensor.iround(fz),(fx,fy,fz),(fxv,fyv,fzv),1,fxv-fyv+numpy.round(fzv),'float32'),#TODO: trouble with the output type. To my understanding, numpy and c round fct return the same type as the input. Why we don't do this?
...@@ -729,10 +729,9 @@ class test_fusion(unittest.TestCase): ...@@ -729,10 +729,9 @@ class test_fusion(unittest.TestCase):
fail1.append(id) fail1.append(id)
topo=f.maker.env.toposort() topo=f.maker.env.toposort()
if gpu: if gpu:
import theano_cuda_ndarray as tcn import theano.sandbox.cuda as cuda
topo_ = [x for x in topo if not isinstance(x.op,cuda.basic_ops.GpuFromHost) and not isinstance(x.op,cuda.basic_ops.HostFromGpu)]
topo_ = [x for x in topo if not isinstance(x.op,tcn.basic_ops.GpuFromHost)] gpu_ = [x for x in topo if isinstance(x.op,cuda.basic_ops.GpuFromHost)]
gpu_ = [x for x in topo if isinstance(x.op,tcn.basic_ops.GpuFromHost)]
if not len(gpu_)==len(sym_inputs): if not len(gpu_)==len(sym_inputs):
fail2.append((id,gpu_,sym_inputs)) fail2.append((id,gpu_,sym_inputs))
else: topo_=topo else: topo_=topo
...@@ -742,9 +741,6 @@ class test_fusion(unittest.TestCase): ...@@ -742,9 +741,6 @@ class test_fusion(unittest.TestCase):
if not out_dtype==out.dtype: if not out_dtype==out.dtype:
fail4.append((id,out_dtype,out.dtype)) fail4.append((id,out_dtype,out.dtype))
# cases[id]=None #to remove g, that link to out that link to the ndarray!
#g.owner.inputs[0] is out... make owner a weakref?
failed=len(fail1+fail2+fail3+fail4) failed=len(fail1+fail2+fail3+fail4)
print "Executed",len(cases),"cases", "failed", failed print "Executed",len(cases),"cases", "failed", failed
if failed>0: if failed>0:
...@@ -766,8 +762,9 @@ class test_fusion(unittest.TestCase): ...@@ -766,8 +762,9 @@ class test_fusion(unittest.TestCase):
mode=compile.mode.predefined_modes['FAST_COMPILE'] mode=compile.mode.predefined_modes['FAST_COMPILE']
mode=compile.mode.predefined_modes['FAST_RUN'] mode=compile.mode.predefined_modes['FAST_RUN']
mode=compile.mode.predefined_modes['DEBUG_MODE'] mode=compile.mode.predefined_modes['DEBUG_MODE']
import theano.sandbox.cuda as tcn mode = theano.compile.mode.get_mode(mode).including('gpu')
self.do(mode, tcn.shared_constructor, shp, gpu=True) import theano.sandbox.cuda as cuda
self.do(mode, cuda.float32_shared_constructor, shp, gpu=True)
def speed_fusion(self, shared_fn = shared, gpu = False, s=None): def speed_fusion(self, shared_fn = shared, gpu = False, s=None):
""" """
...@@ -803,8 +800,8 @@ class test_fusion(unittest.TestCase): ...@@ -803,8 +800,8 @@ class test_fusion(unittest.TestCase):
print "min", d.min(), "argmin", d.argmin(), "max", d.max(), "mean", d.mean(), "std", d.std() print "min", d.min(), "argmin", d.argmin(), "max", d.max(), "mean", d.mean(), "std", d.std()
def speed_fusion_gpu(self): def speed_fusion_gpu(self):
import theano_cuda_ndarray as tcn import theano.sandbox.cuda as cuda
self.speed_fusion(shared_fn=tcn.shared_constructor, gpu=True, s=slice(0,15)) self.speed_fusion(shared_fn=tcn.float32_shared_constructor, gpu=True, s=slice(0,15))
def speed_log_exp(self): def speed_log_exp(self):
s=slice(31,36) s=slice(31,36)
......
Markdown 格式
0% 或
您添加了 0 人 到此讨论。请谨慎行事。
请先完成此评论的编辑！
请 注册 或者 后发表评论