Skip to content

P
pytensor
提交 ec3a90d4 authored 作者: James Bergstra's avatar James Bergstra
浏览文件
操作

Merge pull request #1303 from nouiz/gpusoftmax

Fix GpuSoftmax[WithBias] Ops to work for large rows
上级 97fff534 2335f829
显示空白字符变更
内嵌 并排
正在显示 包含 408 行增加 和 126 行删除
doc/dev_start_guide.txt
浏览文件 @ ec3a90d4
...@@ -505,5 +505,6 @@ Other tools that can help you ...@@ -505,5 +505,6 @@ Other tools that can help you
* `line_profiler <http://pypi.python.org/pypi/line_profiler/>`_: Line-by-line profiler. * `line_profiler <http://pypi.python.org/pypi/line_profiler/>`_: Line-by-line profiler.
* `memory_profiler <http://fseoane.net/blog/2012/line-by-line-report-of-memory-usage/>`_: memory profiler * `memory_profiler <http://fseoane.net/blog/2012/line-by-line-report-of-memory-usage/>`_: memory profiler
* `runsnake <http://www.vrplumber.com/programming/runsnakerun/>`_: Gui for cProfile(time profiler) and Meliae(memory profiler) * `runsnake <http://www.vrplumber.com/programming/runsnakerun/>`_: Gui for cProfile(time profiler) and Meliae(memory profiler)
* `Guppy <https://pypi.python.org/pypi/guppy/>`_: Supports object and heap memory sizing, profiling and debugging.
* `hub <https://github.com/defunkt/hub>`_: A tool that adds github commands to the git command line. * `hub <https://github.com/defunkt/hub>`_: A tool that adds github commands to the git command line.
* `git pull-requests <http://www.splitbrain.org/blog/2011-06/19-automate_github_pull_requests>`_: Another tool for git/github command line. * `git pull-requests <http://www.splitbrain.org/blog/2011-06/19-automate_github_pull_requests>`_: Another tool for git/github command line.
theano/misc/check_blas.py
浏览文件 @ ec3a90d4
...@@ -198,6 +198,8 @@ if __name__ == "__main__": ...@@ -198,6 +198,8 @@ if __name__ == "__main__":
cuda version 5.0 4.2 4.1 4.0 3.2 3.0 # note cuda version 5.0 4.2 4.1 4.0 3.2 3.0 # note
gpu gpu
K20m/ECC 0.07s
K20/NOECC 0.07s
M2070 0.25s 0.27s 0.32s M2070 0.25s 0.27s 0.32s
M2050(Amazon) 0.25s M2050(Amazon) 0.25s
C2075 0.25s C2075 0.25s
...@@ -215,7 +217,7 @@ if __name__ == "__main__": ...@@ -215,7 +217,7 @@ if __name__ == "__main__":
GTX 285 0.452s 0.452s 0.40s # cuda 3.0 seems faster? driver version? GTX 285 0.452s 0.452s 0.40s # cuda 3.0 seems faster? driver version?
GTX 550 Ti 0.57s GTX 550 Ti 0.57s
GT 520 2.68s 3.06s GT 520 2.68s 3.06s
520M 3.19s # with bumblebee on Ubuntu 12.04 520M 2.44s 3.19s # with bumblebee on Ubuntu 12.04
GT 220 3.80s GT 220 3.80s
GT 210 6.35s GT 210 6.35s
8500 GT 10.68s 8500 GT 10.68s
......
theano/sandbox/cuda/kernel_codegen.py
浏览文件 @ ec3a90d4
""" Helper routines for generating gpu kernels for nvcc. """ Helper routines for generating gpu kernels for nvcc.
""" """
def nvcc_kernel(name, params, body): def nvcc_kernel(name, params, body):
"""Return the c code of a kernel function. """Return the c code of a kernel function.
:param params: the parameters to the function as one or more strings :param params: the parameters to the function as one or more strings
:param body: the [nested] list of statements for the body of the function. These will be :param body: the [nested] list of statements for the body of the
separated by ';' characters. function. These will be separated by ';' characters.
""" """
paramstr = ', '.join(params) paramstr = ', '.join(params)
def flatbody(): def flatbody():
for b in body: for b in body:
if isinstance(b, (list, tuple)): if isinstance(b, (list, tuple)):
...@@ -21,12 +25,14 @@ def nvcc_kernel(name, params, body): ...@@ -21,12 +25,14 @@ def nvcc_kernel(name, params, body):
{ {
%(bodystr)s; %(bodystr)s;
} }
""" %locals() """ % locals()
def code_version(version): def code_version(version):
"""decorator to support version-based cache mechanism""" """decorator to support version-based cache mechanism"""
if not isinstance(version, tuple): if not isinstance(version, tuple):
raise TypeError('version must be tuple', version) raise TypeError('version must be tuple', version)
def deco(f): def deco(f):
f.code_version = version f.code_version = version
return f return f
...@@ -34,35 +40,38 @@ def code_version(version): ...@@ -34,35 +40,38 @@ def code_version(version):
UNVERSIONED = () UNVERSIONED = ()
@code_version((1,)) @code_version((1,))
def inline_reduce(N, buf, pos, count, manner_fn): def inline_reduce(N, buf, pos, count, manner_fn):
""" """Return C++ code for a function that reduces a contiguous buffer.
Return C++ code for a function that reduces a contiguous buffer.
:param N: length of the buffer :param N: length of the buffer
:param buf: buffer pointer :param buf: buffer pointer
:param pos: index of executing thread :param pos: index of executing thread
:param count: number of executing threads :param count: number of executing threads
:param manner_fn: a function that accepts strings of arguments a and b, and returns c code
for their reduction. (Example: return "%(a)s + %(b)s" for a sum reduction). :param manner_fn: a function that accepts strings of arguments a
and b, and returns c code for their reduction. (Example:
return "%(a)s + %(b)s" for a sum reduction).
:postcondition: :postcondition:
This function leaves the answer in position 0 of the buffer. The rest of the buffer is This function leaves the answer in position 0 of the buffer. The
trashed by this function. rest of the buffer is trashed by this function.
:note: buf should be in gpu shared memory, we access it many times. :note: buf should be in gpu shared memory, we access it many times.
""" """
loop_line = manner_fn("%s[%s]"%(buf,pos), "%s[i]" %(buf)) loop_line = manner_fn("%s[%s]" % (buf, pos), "%s[i]" % (buf))
r_16 = manner_fn("%s[%s]" %(buf, pos), "%s[%s+16]" %(buf, pos)) r_16 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+16]" % (buf, pos))
r_8 = manner_fn("%s[%s]" %(buf, pos), "%s[%s+8]" %(buf, pos)) r_8 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+8]" % (buf, pos))
r_4 = manner_fn("%s[%s]" %(buf, pos), "%s[%s+4]" %(buf, pos)) r_4 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+4]" % (buf, pos))
r_2 = manner_fn("%s[%s]" %(buf, pos), "%s[%s+2]" %(buf, pos)) r_2 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+2]" % (buf, pos))
r_1 = manner_fn("%s[%s]" %(buf, pos), "%s[%s+1]" %(buf, pos)) r_1 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+1]" % (buf, pos))
return """ return """
{ {
// This function trashes buf[1..N], leaving the reduction result in buf[0]. // This function trashes buf[1..warpSize],
// leaving the reduction result in buf[0].
if (%(pos)s < warpSize) if (%(pos)s < warpSize)
{ {
...@@ -88,24 +97,33 @@ def inline_reduce(N, buf, pos, count, manner_fn): ...@@ -88,24 +97,33 @@ def inline_reduce(N, buf, pos, count, manner_fn):
} }
""" % locals() """ % locals()
@code_version(inline_reduce.code_version) @code_version(inline_reduce.code_version)
def inline_reduce_max(N, buf, pos, count): def inline_reduce_max(N, buf, pos, count):
return inline_reduce(N, buf, pos, count, lambda a, b: "max(%s, %s)"%(a,b)) return inline_reduce(N, buf, pos, count,
lambda a, b: "max(%s, %s)" % (a, b))
@code_version(inline_reduce.code_version) @code_version(inline_reduce.code_version)
def inline_reduce_sum(N, buf, pos, count): def inline_reduce_sum(N, buf, pos, count):
return inline_reduce(N, buf, pos, count, lambda a, b: "%s + %s"%(a,b)) return inline_reduce(N, buf, pos, count,
lambda a, b: "%s + %s" % (a, b))
@code_version(inline_reduce.code_version) @code_version(inline_reduce.code_version)
def inline_reduce_min(N, buf, pos, count): def inline_reduce_min(N, buf, pos, count):
return inline_reduce(N, buf, pos, count, lambda a, b: "min(%s, %s)"%(a,b)) return inline_reduce(N, buf, pos, count,
lambda a, b: "min(%s, %s)" % (a, b))
@code_version(inline_reduce.code_version) @code_version(inline_reduce.code_version)
def inline_reduce_prod(N, buf, pos, count): def inline_reduce_prod(N, buf, pos, count):
return inline_reduce(N, buf, pos, count, lambda a, b: "%s * %s"%(a,b)) return inline_reduce(N, buf, pos, count,
lambda a, b: "%s * %s" % (a, b))
@code_version((2,) + inline_reduce_max.code_version + inline_reduce_sum.code_version) @code_version((2,) + inline_reduce_max.code_version +
inline_reduce_sum.code_version)
def inline_softmax(N, buf, buf2, threadPos, threadCount): def inline_softmax(N, buf, buf2, threadPos, threadCount):
""" """
...@@ -113,10 +131,12 @@ def inline_softmax(N, buf, buf2, threadPos, threadCount): ...@@ -113,10 +131,12 @@ def inline_softmax(N, buf, buf2, threadPos, threadCount):
:param threadPos: index of executing thread :param threadPos: index of executing thread
:param threadCount: number of executing threads :param threadCount: number of executing threads
:Precondition: buf and buf2 contain two identical copies of the input to softmax :Precondition: buf and buf2 contain two identical copies of the input
:Postcondition: buf contains the softmax, buf2 contains un-normalized softmax to softmax
:Postcondition: buf contains the softmax, buf2 contains un-normalized
softmax
:note: buf and buf2 should be in gpu shared memory, we access it many times. :note: buf and buf2 should be in gpu shared memory, we access it many times
:note2: We use __i as an int variable in a loop :note2: We use __i as an int variable in a loop
""" """
...@@ -124,20 +144,173 @@ def inline_softmax(N, buf, buf2, threadPos, threadCount): ...@@ -124,20 +144,173 @@ def inline_softmax(N, buf, buf2, threadPos, threadCount):
#get max of buf (trashing all but buf[0]) #get max of buf (trashing all but buf[0])
inline_reduce_max(N, buf, threadPos, threadCount), inline_reduce_max(N, buf, threadPos, threadCount),
'__syncthreads()', '__syncthreads()',
'float row_max = '+buf+'[0]', 'float row_max = ' + buf + '[0]',
'__syncthreads()', '__syncthreads()',
'for(int __i='+threadPos+'; __i<'+N+'; __i+='+threadCount+'){', 'for(int __i=' + threadPos + '; __i<' + N +
buf+'[__i] = exp('+buf2+'[__i] - row_max)', '; __i+=' + threadCount + '){',
buf2+'[__i] = '+buf+'[__i]', buf + '[__i] = exp(' + buf2 + '[__i] - row_max)',
buf2 + '[__i] = ' + buf + '[__i]',
'}', '}',
'__syncthreads()', '__syncthreads()',
inline_reduce_sum(N, buf, threadPos, threadCount), inline_reduce_sum(N, buf, threadPos, threadCount),
'__syncthreads()', '__syncthreads()',
'float row_sum = '+buf+'[0]', 'float row_sum = ' + buf + '[0]',
'__syncthreads()', '__syncthreads()',
# divide each exp() result by the sum to complete the job. # divide each exp() result by the sum to complete the job.
'for(int __i='+threadPos+'; __i<'+N+'; __i+='+threadCount+'){', 'for(int __i=' + threadPos + '; __i<' + N +
buf+'[__i] = '+buf2+'[__i] / row_sum', '; __i+=' + threadCount + '){',
buf + '[__i] = ' + buf2 + '[__i] / row_sum',
'}', '}',
'__syncthreads()', '__syncthreads()',
] ]
@code_version((1,))
def inline_reduce_fixed_shared(N, buf, x, stride_x, pos, count,
manner_fn, manner_init,
b='', stride_b=''):
"""Return C++ code for a function that reduces a contiguous buffer.
:param N: length of the buffer
:param buf: buffer pointer of size warpSize * sizeof(float)
:param pos: index of executing thread
:param count: number of executing threads
:param b: Optional, pointer to the bias
:param stride_b: Optional, the stride of b if b is provided
:param manner_fn: a function that accepts strings of arguments a
and b, and returns c code for their reduction. (Example:
return "%(a)s + %(b)s" for a sum reduction).
:param manner_init: a function that accepts strings of arguments a
and return c code for its initialization
:postcondition:
This function leaves the answer in position 0 of the buffer. The
rest of the buffer is trashed by this function.
:note: buf should be in gpu shared memory, we access it many times.
"""
if b:
init = manner_init("%(x)s[%(pos)s * %(stride_x)s] +"
" %(b)s[%(pos)s * %(stride_b)s]" % locals())
loop_line = manner_fn("red",
manner_init("%(x)s[i * %(stride_x)s] + "
"%(b)s[i * %(stride_b)s]" %
locals()))
else:
init = manner_init("%(x)s[%(pos)s * %(stride_x)s]" % locals())
loop_line = manner_fn("red", manner_init("%(x)s[i * %(stride_x)s]" %
locals()))
loop_line2 = manner_fn("%s[%s]" % (buf, pos),
"%s[i]" % buf)
r_16 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+16]" % (buf, pos))
r_8 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+8]" % (buf, pos))
r_4 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+4]" % (buf, pos))
r_2 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+2]" % (buf, pos))
r_1 = manner_fn("%s[%s]" % (buf, pos), "%s[%s+1]" % (buf, pos))
return """
{
// This function trashes buf[1..n_threads],
// leaving the reduction result in buf[0].
float red = %(init)s;
#pragma unroll 16
for (int i = %(pos)s + %(count)s; i<%(N)s; i += %(count)s){
red = %(loop_line)s;
}
buf[%(pos)s] = red;
__syncthreads();
if (%(pos)s < warpSize)
{
for (int i = %(pos)s + warpSize; i < %(count)s; i += warpSize)
{
%(buf)s[%(pos)s] = %(loop_line2)s;
}
if (%(pos)s < 16)
{
//reduce so that %(pos)s 0 has the reduction of everything
if(%(pos)s + 16 < %(N)s)
%(buf)s[%(pos)s] = %(r_16)s;
if(%(pos)s + 8 < %(N)s)
%(buf)s[%(pos)s] = %(r_8)s;
if(%(pos)s + 4 < %(N)s)
%(buf)s[%(pos)s] = %(r_4)s;
if(%(pos)s + 2 < %(N)s)
%(buf)s[%(pos)s] = %(r_2)s;
if(%(pos)s + 1 < %(N)s)
%(buf)s[%(pos)s] = %(r_1)s;
}
}
}
""" % locals()
@code_version(inline_reduce_fixed_shared.code_version)
def inline_reduce_fixed_shared_max(N, buf, x, stride_x, pos, count,
b='', stride_b=''):
return inline_reduce_fixed_shared(N, buf, x, stride_x, pos, count,
lambda a, b: "max(%s, %s)" % (a, b),
lambda a: a,
b, stride_b)
@code_version((1,) + inline_reduce_max.code_version +
inline_reduce_sum.code_version)
def inline_softmax_fixed_shared(N, buf, x, stride_x,
sm, sm_stride,
threadPos, threadCount,
b='', stride_b=''):
"""
:param N: length of the buffer, atleast waprSize(32).
:param buf: a shared memory buffer of size warpSize * sizeof(float)
:param x: a ptr to the gpu memory where the row is stored
:param stride_x: the stride between each element in x
:param sm: a ptr to the gpu memory to store the result
:param sm_stride: the stride between eash sm element
:param threadPos: index of executing thread
:param threadCount: number of executing threads
:param b: Optional, pointer to the bias
:param stride_b: Optional, the stride of b if b is provided
:Precondition: buf is empty
:Postcondition: buf[0] contains the softmax,
buf2 contains un-normalized softmax
:note: buf should be in gpu shared memory, we access it many times.
:note2: We use tx as an int variable in a loop
"""
ret = [
#get max of buf (trashing all but buf[0])
inline_reduce_fixed_shared_max(N, buf, x, stride_x,
threadPos, threadCount, b, stride_b),
'__syncthreads()',
'float row_max = ' + buf + '[0]',
'__syncthreads()',
inline_reduce_fixed_shared(N, buf, x, stride_x, threadPos, threadCount,
lambda a, b: "%s + %s" % (a, b),
lambda a: "exp(%s - row_max)" % a,
b, stride_b),
'__syncthreads()',
'float row_sum = ' + buf + '[0]',
'__syncthreads()',
"for (int tx = threadIdx.x; tx< N; tx += blockDim.x){",
]
# This set all value correctly
if b:
ret += [
"%(sm)s[tx * %(sm_stride)s] = "
" exp(%(x)s[tx * %(stride_x)s] +"
" %(b)s[tx * %(stride_b)s] - row_max)"
" / row_sum" % locals()]
else:
ret += [
"%(sm)s[tx * %(sm_stride)s] = "
"exp(%(x)s[tx * %(stride_x)s] - row_max) / row_sum" % locals()]
ret += [
"}",
'__syncthreads()',
]
return ret
theano/sandbox/cuda/nnet.py
浏览文件 @ ec3a90d4
from theano import Op, Type, Apply, Variable, Constant from theano import Op, Apply
from theano import tensor, scalar
import StringIO import StringIO
from theano.sandbox.cuda.type import CudaNdarrayType
from theano.sandbox.cuda import GpuOp from theano.sandbox.cuda import GpuOp
from theano.sandbox.cuda.kernel_codegen import (nvcc_kernel, inline_reduce_max, from theano.sandbox.cuda.kernel_codegen import (nvcc_kernel,
inline_reduce_sum, inline_softmax,
inline_softmax) inline_softmax_fixed_shared)
class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp): class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp):
...@@ -111,7 +109,8 @@ class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp): ...@@ -111,7 +109,8 @@ class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp):
PyErr_SetString(PyExc_ValueError, "b not 1d tensor"); PyErr_SetString(PyExc_ValueError, "b not 1d tensor");
%(fail)s; %(fail)s;
} }
if (CudaNdarray_HOST_DIMS(%(x)s)[0] != CudaNdarray_HOST_DIMS(%(y_idx)s)[0]) if (CudaNdarray_HOST_DIMS(%(x)s)[0] !=
CudaNdarray_HOST_DIMS(%(y_idx)s)[0])
{ {
PyErr_SetString(PyExc_ValueError, PyErr_SetString(PyExc_ValueError,
"dimension mismatch in x,y_idx arguments"); "dimension mismatch in x,y_idx arguments");
...@@ -124,56 +123,73 @@ class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp): ...@@ -124,56 +123,73 @@ class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp):
%(fail)s; %(fail)s;
} }
if ((NULL == %(nll)s) //initial condition if ((NULL == %(nll)s) //initial condition
|| (CudaNdarray_HOST_DIMS(%(nll)s)[0] != CudaNdarray_HOST_DIMS(%(y_idx)s)[0])) || (CudaNdarray_HOST_DIMS(%(nll)s)[0] !=
CudaNdarray_HOST_DIMS(%(y_idx)s)[0]))
{ {
Py_XDECREF(%(nll)s); Py_XDECREF(%(nll)s);
%(nll)s = (CudaNdarray*)CudaNdarray_NewDims(1, CudaNdarray_HOST_DIMS(%(y_idx)s)); %(nll)s = (CudaNdarray*)CudaNdarray_NewDims(1,
CudaNdarray_HOST_DIMS(%(y_idx)s));
if(!%(nll)s) if(!%(nll)s)
{ {
%(fail)s; %(fail)s;
} }
} }
if ((NULL == %(sm)s) if ((NULL == %(sm)s)
|| (CudaNdarray_HOST_DIMS(%(sm)s)[0] != CudaNdarray_HOST_DIMS(%(x)s)[0]) || (CudaNdarray_HOST_DIMS(%(sm)s)[0] !=
|| (CudaNdarray_HOST_DIMS(%(sm)s)[1] != CudaNdarray_HOST_DIMS(%(x)s)[1])) CudaNdarray_HOST_DIMS(%(x)s)[0])
|| (CudaNdarray_HOST_DIMS(%(sm)s)[1] !=
CudaNdarray_HOST_DIMS(%(x)s)[1]))
{ {
Py_XDECREF(%(sm)s); Py_XDECREF(%(sm)s);
%(sm)s = (CudaNdarray*) CudaNdarray_NewDims(2, CudaNdarray_HOST_DIMS(%(x)s)); %(sm)s = (CudaNdarray*) CudaNdarray_NewDims(2,
CudaNdarray_HOST_DIMS(%(x)s));
if(!%(sm)s) if(!%(sm)s)
{ {
PyErr_SetString(PyExc_MemoryError, PyErr_SetString(PyExc_MemoryError,
"failed to alloc sm output"); "failed to alloc sm output");
// no need to decref cnda_nll, the cleanup code should pick it up. // no need to decref cnda_nll, the cleanup code should do it up
%(fail)s; %(fail)s;
} }
} }
if ((NULL == %(am)s) if ((NULL == %(am)s)
|| (CudaNdarray_HOST_DIMS(%(am)s)[0] != CudaNdarray_HOST_DIMS(%(y_idx)s)[0])) || (CudaNdarray_HOST_DIMS(%(am)s)[0] !=
CudaNdarray_HOST_DIMS(%(y_idx)s)[0]))
{ {
Py_XDECREF(%(am)s); Py_XDECREF(%(am)s);
%(am)s = (CudaNdarray*) CudaNdarray_NewDims(1, CudaNdarray_HOST_DIMS(%(y_idx)s)); %(am)s = (CudaNdarray*) CudaNdarray_NewDims(1,
CudaNdarray_HOST_DIMS(%(y_idx)s));
if(!%(am)s) if(!%(am)s)
{ {
PyErr_SetString(PyExc_MemoryError, PyErr_SetString(PyExc_MemoryError,
"failed to alloc am output"); "failed to alloc am output");
// no need to decref nll amd sm, the cleanup code should pick it up. // no need to decref nll and sm,
// the cleanup code should do it up
%(fail)s; %(fail)s;
} }
} }
{ {
int n_blocks = CudaNdarray_HOST_DIMS(%(sm)s)[0]; int n_blocks = CudaNdarray_HOST_DIMS(%(sm)s)[0];
int n_threads = 1; //TODO: launch more threads per row and do parallel sum and max reductions. //TODO: launch more threads per row and do parallel sum and max reductions
int n_threads = 1;
int n_shared_bytes = 0; //n_threads * sizeof(float); int n_shared_bytes = 0; //n_threads * sizeof(float);
k_xent_sm_1hot_bias<<<n_blocks, n_threads, n_shared_bytes>>>( k_xent_sm_1hot_bias<<<n_blocks, n_threads, n_shared_bytes>>>(
CudaNdarray_HOST_DIMS(%(x)s)[0], CudaNdarray_HOST_DIMS(%(x)s)[0],
CudaNdarray_HOST_DIMS(%(x)s)[1], CudaNdarray_HOST_DIMS(%(x)s)[1],
CudaNdarray_DEV_DATA(%(x)s), CudaNdarray_HOST_STRIDES(%(x)s)[0], CudaNdarray_HOST_STRIDES(%(x)s)[1], CudaNdarray_DEV_DATA(%(x)s),
CudaNdarray_DEV_DATA(%(b)s), CudaNdarray_HOST_STRIDES(%(b)s)[0], CudaNdarray_HOST_STRIDES(%(x)s)[0],
CudaNdarray_DEV_DATA(%(y_idx)s), CudaNdarray_HOST_STRIDES(%(y_idx)s)[0], CudaNdarray_HOST_STRIDES(%(x)s)[1],
CudaNdarray_DEV_DATA(%(nll)s), CudaNdarray_HOST_STRIDES(%(nll)s)[0], CudaNdarray_DEV_DATA(%(b)s),
CudaNdarray_DEV_DATA(%(sm)s), CudaNdarray_HOST_STRIDES(%(sm)s)[0], CudaNdarray_HOST_STRIDES(%(sm)s)[1], CudaNdarray_HOST_STRIDES(%(b)s)[0],
CudaNdarray_DEV_DATA(%(am)s), CudaNdarray_HOST_STRIDES(%(am)s)[0]); CudaNdarray_DEV_DATA(%(y_idx)s),
CudaNdarray_HOST_STRIDES(%(y_idx)s)[0],
CudaNdarray_DEV_DATA(%(nll)s),
CudaNdarray_HOST_STRIDES(%(nll)s)[0],
CudaNdarray_DEV_DATA(%(sm)s),
CudaNdarray_HOST_STRIDES(%(sm)s)[0],
CudaNdarray_HOST_STRIDES(%(sm)s)[1],
CudaNdarray_DEV_DATA(%(am)s),
CudaNdarray_HOST_STRIDES(%(am)s)[0]);
CNDA_THREAD_SYNC; CNDA_THREAD_SYNC;
cudaError_t err = cudaGetLastError(); cudaError_t err = cudaGetLastError();
if (cudaSuccess != err) if (cudaSuccess != err)
...@@ -181,7 +197,7 @@ class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp): ...@@ -181,7 +197,7 @@ class GpuCrossentropySoftmaxArgmax1HotWithBias (GpuOp):
PyErr_Format(PyExc_RuntimeError, PyErr_Format(PyExc_RuntimeError,
"Cuda error: %(classname)s %(nodename)s: %%s.\\n", "Cuda error: %(classname)s %(nodename)s: %%s.\\n",
cudaGetErrorString(err)); cudaGetErrorString(err));
// no need to decref output vars the cleanup code should pick them up. // no need to decref output vars the cleanup code will do it
%(fail)s; %(fail)s;
} }
} }
...@@ -203,7 +219,7 @@ class GpuCrossentropySoftmax1HotWithBiasDx (GpuOp): ...@@ -203,7 +219,7 @@ class GpuCrossentropySoftmax1HotWithBiasDx (GpuOp):
nout = 1 nout = 1
"""Gradient wrt x of the CrossentropySoftmax1Hot Op""" """Gradient wrt x of the CrossentropySoftmax1Hot Op"""
def __init__(self, **kwargs): def __init__(self, **kwargs):
Op.__init__(self,**kwargs) Op.__init__(self, **kwargs)
def __eq__(self, other): def __eq__(self, other):
return type(self) == type(other) return type(self) == type(other)
...@@ -233,26 +249,33 @@ class GpuCrossentropySoftmax1HotWithBiasDx (GpuOp): ...@@ -233,26 +249,33 @@ class GpuCrossentropySoftmax1HotWithBiasDx (GpuOp):
PyErr_SetString(PyExc_ValueError, "rank error"); PyErr_SetString(PyExc_ValueError, "rank error");
%(fail)s; %(fail)s;
} }
if (CudaNdarray_HOST_DIMS(%(dnll)s)[0] != CudaNdarray_HOST_DIMS(%(sm)s)[0]) if (CudaNdarray_HOST_DIMS(%(dnll)s)[0] !=
CudaNdarray_HOST_DIMS(%(sm)s)[0])
{ {
PyErr_Format(PyExc_ValueError, "dnll.shape[0] == %%i, but sm.shape[0] == %%i", PyErr_Format(PyExc_ValueError,
CudaNdarray_HOST_DIMS(%(dnll)s)[0],CudaNdarray_HOST_DIMS(%(sm)s)[0]); "dnll.shape[0] == %%i, but sm.shape[0] == %%i",
CudaNdarray_HOST_DIMS(%(dnll)s)[0],
CudaNdarray_HOST_DIMS(%(sm)s)[0]);
%(fail)s; %(fail)s;
} }
if (CudaNdarray_HOST_DIMS(%(dnll)s)[0] != CudaNdarray_HOST_DIMS(%(y_idx)s)[0]) if (CudaNdarray_HOST_DIMS(%(dnll)s)[0] !=
CudaNdarray_HOST_DIMS(%(y_idx)s)[0])
{ {
PyErr_SetString(PyExc_ValueError, PyErr_SetString(PyExc_ValueError,
"dnll.shape[0] != y_idx.shape[0]"); "dnll.shape[0] != y_idx.shape[0]");
%(fail)s; %(fail)s;
} }
if ((NULL == %(dx)s) if ((NULL == %(dx)s)
|| (CudaNdarray_HOST_DIMS(%(dx)s)[0] != CudaNdarray_HOST_DIMS(%(sm)s)[0]) || (CudaNdarray_HOST_DIMS(%(dx)s)[0] !=
|| (CudaNdarray_HOST_DIMS(%(dx)s)[1] != CudaNdarray_HOST_DIMS(%(sm)s)[1])) CudaNdarray_HOST_DIMS(%(sm)s)[0])
|| (CudaNdarray_HOST_DIMS(%(dx)s)[1] !=
CudaNdarray_HOST_DIMS(%(sm)s)[1]))
{ {
Py_XDECREF(%(dx)s); Py_XDECREF(%(dx)s);
%(dx)s = (CudaNdarray*)CudaNdarray_New(); %(dx)s = (CudaNdarray*)CudaNdarray_New();
if ((NULL == %(dx)s) if ((NULL == %(dx)s)
|| CudaNdarray_alloc_contiguous(%(dx)s, 2, CudaNdarray_HOST_DIMS(%(sm)s))) || CudaNdarray_alloc_contiguous(%(dx)s, 2,
CudaNdarray_HOST_DIMS(%(sm)s)))
{ {
Py_XDECREF(%(dx)s); Py_XDECREF(%(dx)s);
%(dx)s = NULL; %(dx)s = NULL;
...@@ -314,13 +337,16 @@ class GpuCrossentropySoftmax1HotWithBiasDx (GpuOp): ...@@ -314,13 +337,16 @@ class GpuCrossentropySoftmax1HotWithBiasDx (GpuOp):
{ {
if (y_i == j) if (y_i == j)
{ {
dx[i * dx_s0 + j * dx_s1] = dnll_i * (sm[i * sm_s0 + j * sm_s1]-1.0); dx[i * dx_s0 + j * dx_s1] =
dnll_i * (sm[i * sm_s0 + j * sm_s1]-1.0);
} }
else else
{ {
dx[i * dx_s0 + j * dx_s1] = dnll_i * sm[i * sm_s0 + j * sm_s1]; dx[i * dx_s0 + j * dx_s1] =
dnll_i * sm[i * sm_s0 + j * sm_s1];
} }
//dx[i * dx_s0 + j * dx_s1] = dnll_i * sm[i * sm_s0 + j * sm_s1]; //dx[i * dx_s0 + j * dx_s1] =
// dnll_i * sm[i * sm_s0 + j * sm_s1];
//dx[i*dx_s0+j*dx_s1] = 0; //dx[i*dx_s0+j*dx_s1] = 0;
} }
} }
...@@ -350,8 +376,7 @@ class GpuSoftmax (GpuOp): ...@@ -350,8 +376,7 @@ class GpuSoftmax (GpuOp):
return shape return shape
def c_code_cache_version(self): def c_code_cache_version(self):
#return () return (9,) + inline_softmax.code_version
return (7,) + inline_softmax.code_version
def c_code(self, node, nodename, inp, out, sub): def c_code(self, node, nodename, inp, out, sub):
x, = inp x, = inp
...@@ -364,8 +389,10 @@ class GpuSoftmax (GpuOp): ...@@ -364,8 +389,10 @@ class GpuSoftmax (GpuOp):
%(fail)s; %(fail)s;
} }
if ((NULL == %(z)s) || if ((NULL == %(z)s) ||
(CudaNdarray_HOST_DIMS(%(z)s)[0] != CudaNdarray_HOST_DIMS(%(x)s)[0]) || (CudaNdarray_HOST_DIMS(%(z)s)[0] !=
(CudaNdarray_HOST_DIMS(%(z)s)[1] != CudaNdarray_HOST_DIMS(%(x)s)[1])) CudaNdarray_HOST_DIMS(%(x)s)[0]) ||
(CudaNdarray_HOST_DIMS(%(z)s)[1] !=
CudaNdarray_HOST_DIMS(%(x)s)[1]))
{ {
Py_XDECREF(%(z)s); Py_XDECREF(%(z)s);
%(z)s = (CudaNdarray*)CudaNdarray_New(); %(z)s = (CudaNdarray*)CudaNdarray_New();
...@@ -379,13 +406,19 @@ class GpuSoftmax (GpuOp): ...@@ -379,13 +406,19 @@ class GpuSoftmax (GpuOp):
} }
} }
{ {
int n_blocks = std::min(CudaNdarray_HOST_DIMS(%(x)s)[0], 32 * 1024); int n_blocks = std::min(CudaNdarray_HOST_DIMS(%(x)s)[0],
32 * 1024);
//TODO, detect the maximum number of thread per block. //TODO, detect the maximum number of thread per block.
int n_threads = std::min(CudaNdarray_HOST_DIMS(%(x)s)[1], 512); int n_threads = std::min(CudaNdarray_HOST_DIMS(%(x)s)[1], 512);
int n_shared_bytes = CudaNdarray_HOST_DIMS(%(x)s)[1] * 2 * sizeof(float); int n_shared_bytes = CudaNdarray_HOST_DIMS(%(x)s)[1] *
2 * sizeof(float);
if (CudaNdarray_HOST_DIMS(%(x)s)[0] > 0) if (CudaNdarray_HOST_DIMS(%(x)s)[0] > 0)
{ {
//Those numbers are based on not too recent GPU
//to make them compatible with more GPU.
//TODO: read the information from the card.
if(n_shared_bytes < (32 * 1024 - 500)){
kSoftmax_%(nodename)s kSoftmax_%(nodename)s
<<< <<<
n_blocks, n_blocks,
...@@ -403,6 +436,25 @@ class GpuSoftmax (GpuOp): ...@@ -403,6 +436,25 @@ class GpuSoftmax (GpuOp):
CudaNdarray_HOST_STRIDES(%(z)s)[0], CudaNdarray_HOST_STRIDES(%(z)s)[0],
CudaNdarray_HOST_STRIDES(%(z)s)[1] CudaNdarray_HOST_STRIDES(%(z)s)[1]
); );
}else{
kSoftmax_fixed_shared%(nodename)s
<<<
n_blocks,
n_threads,
n_threads * sizeof(float)
>>>(
CudaNdarray_HOST_DIMS(%(x)s)[0],
CudaNdarray_HOST_DIMS(%(x)s)[1],
CudaNdarray_DEV_DATA(%(x)s),
CudaNdarray_HOST_STRIDES(%(x)s)[0],
CudaNdarray_HOST_STRIDES(%(x)s)[1],
CudaNdarray_DEV_DATA(%(z)s),
CudaNdarray_HOST_STRIDES(%(z)s)[0],
CudaNdarray_HOST_STRIDES(%(z)s)[1]
);
}
CNDA_THREAD_SYNC; CNDA_THREAD_SYNC;
cudaError_t err = cudaGetLastError(); cudaError_t err = cudaGetLastError();
if( cudaSuccess != err) if( cudaSuccess != err)
...@@ -410,7 +462,8 @@ class GpuSoftmax (GpuOp): ...@@ -410,7 +462,8 @@ class GpuSoftmax (GpuOp):
PyErr_Format(PyExc_RuntimeError, PyErr_Format(PyExc_RuntimeError,
"Cuda error: %%s: %%s.\\n Used %%d blocks," "Cuda error: %%s: %%s.\\n Used %%d blocks,"
" %%d threads %%d bytes of shared memory", " %%d threads %%d bytes of shared memory",
"kSoftmax_%(nodename)s", cudaGetErrorString(err), "kSoftmax[_fixed_shared]%(nodename)s",
cudaGetErrorString(err),
n_blocks, n_threads, n_shared_bytes); n_blocks, n_threads, n_shared_bytes);
%(fail)s; %(fail)s;
} }
...@@ -420,14 +473,15 @@ class GpuSoftmax (GpuOp): ...@@ -420,14 +473,15 @@ class GpuSoftmax (GpuOp):
""" % locals() """ % locals()
def c_support_code_apply(self, node, nodename): def c_support_code_apply(self, node, nodename):
return nvcc_kernel("kSoftmax_%s" % nodename, ret1 = nvcc_kernel("kSoftmax_%s" % nodename,
params=['int M', 'int N', params=['int M', 'int N',
'const float * x', 'const int sx0', 'const int sx1', 'const float * x', 'const int sx0', 'const int sx1',
'float * sm', 'const int sm_s0', 'const int sm_s1'], 'float * sm', 'const int sm_s0', 'const int sm_s1'],
body=[ body=[
"extern __shared__ float buf[]", "extern __shared__ float buf[]",
"float * buf2 = buf + N", "float * buf2 = buf + N",
"for (int blockIDX = blockIdx.x; blockIDX < M; blockIDX += gridDim.x){", "for (int blockIDX = blockIdx.x; blockIDX < M;"
" blockIDX += gridDim.x){",
"for (int tx = threadIdx.x; tx< N; tx += blockDim.x){", "for (int tx = threadIdx.x; tx< N; tx += blockDim.x){",
"buf[tx] = x[blockIDX * sx0 + tx * sx1]", "buf[tx] = x[blockIDX * sx0 + tx * sx1]",
"buf2[tx] = buf[tx]", "buf2[tx] = buf[tx]",
...@@ -442,6 +496,23 @@ class GpuSoftmax (GpuOp): ...@@ -442,6 +496,23 @@ class GpuSoftmax (GpuOp):
"__syncthreads()", "__syncthreads()",
"}", "}",
]) ])
ret2 = nvcc_kernel("kSoftmax_fixed_shared%s" % nodename,
params=['int M', 'int N',
'const float * x', 'const int sx0', 'const int sx1',
'float * sm', 'const int sm_s0', 'const int sm_s1'],
body=[
"extern __shared__ float buf[]",
"for (int blockIDX = blockIdx.x; blockIDX < M;"
" blockIDX += gridDim.x){",
"const float *x_ptr = &x[blockIDX * sx0]",
"float *sm_ptr = &sm[blockIDX * sm_s0]",
inline_softmax_fixed_shared('N', 'buf', 'x_ptr', 'sx1',
'sm_ptr', 'sm_s1',
'threadIdx.x', 'blockDim.x'),
"__syncthreads()",
"}",
])
return ret1 + "\n" + ret2
gpu_softmax = GpuSoftmax() gpu_softmax = GpuSoftmax()
...@@ -470,7 +541,7 @@ class GpuSoftmaxWithBias (GpuOp): ...@@ -470,7 +541,7 @@ class GpuSoftmaxWithBias (GpuOp):
def c_code_cache_version(self): def c_code_cache_version(self):
#return () #return ()
return (7,) + inline_softmax.code_version return (8,) + inline_softmax.code_version
def c_code(self, node, nodename, inp, out, sub): def c_code(self, node, nodename, inp, out, sub):
x, b = inp x, b = inp
...@@ -487,20 +558,27 @@ class GpuSoftmaxWithBias (GpuOp): ...@@ -487,20 +558,27 @@ class GpuSoftmaxWithBias (GpuOp):
PyErr_SetString(PyExc_ValueError, "rank error for the bias"); PyErr_SetString(PyExc_ValueError, "rank error for the bias");
%(fail)s; %(fail)s;
} }
if ((CudaNdarray_HOST_DIMS(%(x)s)[1] != CudaNdarray_HOST_DIMS(%(b)s)[0])) if ((CudaNdarray_HOST_DIMS(%(x)s)[1] !=
CudaNdarray_HOST_DIMS(%(b)s)[0]))
{ {
PyErr_Format(PyExc_ValueError, "number of columns in x (%%ld) does not match length of b (%%ld)", PyErr_Format(PyExc_ValueError,
(long int)CudaNdarray_HOST_DIMS(%(x)s)[1], (long int)CudaNdarray_HOST_DIMS(%(b)s)[0]); "number of columns in x (%%ld)"
" does not match length of b (%%ld)",
(long int)CudaNdarray_HOST_DIMS(%(x)s)[1],
(long int)CudaNdarray_HOST_DIMS(%(b)s)[0]);
%(fail)s; %(fail)s;
} }
if ((NULL == %(z)s) if ((NULL == %(z)s)
|| (CudaNdarray_HOST_DIMS(%(z)s)[0] != CudaNdarray_HOST_DIMS(%(x)s)[0]) || (CudaNdarray_HOST_DIMS(%(z)s)[0] !=
|| (CudaNdarray_HOST_DIMS(%(z)s)[1] != CudaNdarray_HOST_DIMS(%(x)s)[1])) CudaNdarray_HOST_DIMS(%(x)s)[0])
|| (CudaNdarray_HOST_DIMS(%(z)s)[1] !=
CudaNdarray_HOST_DIMS(%(x)s)[1]))
{ {
Py_XDECREF(%(z)s); Py_XDECREF(%(z)s);
%(z)s = (CudaNdarray*)CudaNdarray_New(); %(z)s = (CudaNdarray*)CudaNdarray_New();
if ((NULL == %(z)s) if ((NULL == %(z)s)
|| CudaNdarray_alloc_contiguous(%(z)s, 2, CudaNdarray_HOST_DIMS(%(x)s))) || CudaNdarray_alloc_contiguous(%(z)s, 2,
CudaNdarray_HOST_DIMS(%(x)s)))
{ {
Py_XDECREF(%(z)s); Py_XDECREF(%(z)s);
%(z)s = NULL; %(z)s = NULL;
...@@ -511,13 +589,13 @@ class GpuSoftmaxWithBias (GpuOp): ...@@ -511,13 +589,13 @@ class GpuSoftmaxWithBias (GpuOp):
int n_blocks = std::min(CudaNdarray_HOST_DIMS(%(x)s)[0],32*1024); int n_blocks = std::min(CudaNdarray_HOST_DIMS(%(x)s)[0],32*1024);
//TODO, detect the maximum number of thread per block. //TODO, detect the maximum number of thread per block.
int n_threads = std::min(CudaNdarray_HOST_DIMS(%(x)s)[1], 512); int n_threads = std::min(CudaNdarray_HOST_DIMS(%(x)s)[1], 512);
int n_shared_bytes = CudaNdarray_HOST_DIMS(%(x)s)[1] * 2 * sizeof(float); int n_shared_bytes = CudaNdarray_HOST_DIMS(%(x)s)[1] *
2 * sizeof(float);
if (CudaNdarray_HOST_DIMS(%(x)s)[0] > 0) if (CudaNdarray_HOST_DIMS(%(x)s)[0] > 0)
{ {
if(n_shared_bytes < (32 * 1024 - 500)){
kSoftmaxWithBias_%(nodename)s kSoftmaxWithBias_%(nodename)s
<<< <<<
// todo: cap these at the card limits,
// implement loops in kernel
n_blocks, n_blocks,
n_threads, n_threads,
n_shared_bytes n_shared_bytes
...@@ -536,6 +614,28 @@ class GpuSoftmaxWithBias (GpuOp): ...@@ -536,6 +614,28 @@ class GpuSoftmaxWithBias (GpuOp):
CudaNdarray_HOST_STRIDES(%(z)s)[0], CudaNdarray_HOST_STRIDES(%(z)s)[0],
CudaNdarray_HOST_STRIDES(%(z)s)[1] CudaNdarray_HOST_STRIDES(%(z)s)[1]
); );
}else{
kSoftmaxWithBias_fixed_shared%(nodename)s
<<<
n_blocks,
n_threads,
n_threads * sizeof(float)
>>>(
CudaNdarray_HOST_DIMS(%(x)s)[0],
CudaNdarray_HOST_DIMS(%(x)s)[1],
CudaNdarray_DEV_DATA(%(x)s),
CudaNdarray_HOST_STRIDES(%(x)s)[0],
CudaNdarray_HOST_STRIDES(%(x)s)[1],
CudaNdarray_DEV_DATA(%(b)s),
CudaNdarray_HOST_STRIDES(%(b)s)[0],
CudaNdarray_DEV_DATA(%(z)s),
CudaNdarray_HOST_STRIDES(%(z)s)[0],
CudaNdarray_HOST_STRIDES(%(z)s)[1]
);
}
CNDA_THREAD_SYNC; CNDA_THREAD_SYNC;
cudaError_t err = cudaGetLastError(); cudaError_t err = cudaGetLastError();
if( cudaSuccess != err) if( cudaSuccess != err)
...@@ -552,27 +652,52 @@ class GpuSoftmaxWithBias (GpuOp): ...@@ -552,27 +652,52 @@ class GpuSoftmaxWithBias (GpuOp):
""" % locals() """ % locals()
def c_support_code_apply(self, node, nodename): def c_support_code_apply(self, node, nodename):
return nvcc_kernel("kSoftmaxWithBias_%s"%nodename, ret1 = nvcc_kernel("kSoftmaxWithBias_%s" % nodename,
params=['int M', 'int N', params=['int M', 'int N',
'const float * x', 'const int sx0', 'const int sx1', 'const float * x', 'const int sx0', 'const int sx1',
'const float * b', 'const int sb0', 'const float * b', 'const int sb0',
'float * sm', 'const int ssm0', 'const int ssm1'], 'float * sm', 'const int sm_s0', 'const int sm_s1'],
body=[ body=[
"extern __shared__ float buf[]", "extern __shared__ float buf[]",
"float * buf2 = buf + N", "float * buf2 = buf + N",
"for (int blockIDX = blockIdx.x; blockIDX < M; blockIDX += gridDim.x){", "for (int blockIDX = blockIdx.x; blockIDX < M;"
" blockIDX += gridDim.x){",
"for (int tx = threadIdx.x; tx< N; tx += blockDim.x){", "for (int tx = threadIdx.x; tx< N; tx += blockDim.x){",
"buf[tx] = x[blockIDX * sx0 + tx * sx1]", "buf[tx] = x[blockIDX * sx0 + tx * sx1]",
"buf[tx] += b[tx * sb0]", "buf[tx] += b[tx * sb0]",
"buf2[tx] = buf[tx]", "buf2[tx] = buf[tx]",
"}", "}",
"__syncthreads()", "__syncthreads()",
inline_softmax('N', 'buf', 'buf2', 'threadIdx.x', 'blockDim.x'), inline_softmax('N', 'buf', 'buf2',
'threadIdx.x', 'blockDim.x'),
"for (int tx = threadIdx.x; tx< N; tx += blockDim.x){", "for (int tx = threadIdx.x; tx< N; tx += blockDim.x){",
"sm[blockIDX * ssm0 + tx * ssm1] = buf[tx]", "sm[blockIDX * sm_s0 + tx * sm_s1] = buf[tx]",
"}",
"__syncthreads()",
"}", "}",
])
ret2 = nvcc_kernel("kSoftmaxWithBias_fixed_shared%s" % nodename,
params=['int M', 'int N',
'const float * x',
'const int sx0', 'const int sx1',
'const float * b', 'const int sb0',
'float * sm',
'const int sm_s0', 'const int sm_s1'],
body=[
"extern __shared__ float buf[]",
"for (int blockIDX = blockIdx.x; blockIDX < M;"
" blockIDX += gridDim.x){",
"const float *x_ptr = &x[blockIDX * sx0]",
"float *sm_ptr = &sm[blockIDX * sm_s0]",
inline_softmax_fixed_shared('N', 'buf',
'x_ptr', 'sx1',
'sm_ptr', 'sm_s1',
'threadIdx.x',
'blockDim.x',
'b', 'sb0'),
"__syncthreads()", "__syncthreads()",
"}", "}",
]) ])
return ret1 + "\n" + ret2
gpu_softmax_with_bias = GpuSoftmaxWithBias() gpu_softmax_with_bias = GpuSoftmaxWithBias()
theano/sandbox/cuda/tests/test_nnet.py
浏览文件 @ ec3a90d4
...@@ -172,8 +172,8 @@ def test_softmax_with_bias(): ...@@ -172,8 +172,8 @@ def test_softmax_with_bias():
x = T.fmatrix('x') x = T.fmatrix('x')
# We can't use zeros_like(x[0,::]) as this don't allow to test with # We can't use zeros_like(x[0,::]) as this don't allow to test with
# 0 shape. # 0 shape.
z = T.nnet.softmax_with_bias(x, T.alloc(numpy.asarray(0, dtype='float32'), z = T.nnet.softmax_with_bias(x, T.arange(x.shape[1] * 2,
x.shape[1])) dtype='float32')[::2])
f = theano.function([x], z, mode=mode_without_gpu) f = theano.function([x], z, mode=mode_without_gpu)
f_gpu = theano.function([x], z, mode=mode_with_gpu) f_gpu = theano.function([x], z, mode=mode_with_gpu)
...@@ -181,24 +181,12 @@ def test_softmax_with_bias(): ...@@ -181,24 +181,12 @@ def test_softmax_with_bias():
assert isinstance(f_gpu.maker.fgraph.toposort()[-2].op, assert isinstance(f_gpu.maker.fgraph.toposort()[-2].op,
cuda.nnet.GpuSoftmaxWithBias) cuda.nnet.GpuSoftmaxWithBias)
def cmp(n, m, catch=False): def cmp(n, m):
"""Some old card won't accet the configuration arguments of
this implementation. For those cases set catch=True to skip
those errors.
"""
try:
#print "test_softmax",n,m #print "test_softmax",n,m
data = numpy.arange(n * m, dtype='float32').reshape(n, m) data = numpy.arange(n * m, dtype='float32').reshape(n, m)
out = f(data) out = f(data)
gout = f_gpu(data) gout = f_gpu(data)
assert numpy.allclose(out, gout), numpy.absolute(out - gout) assert numpy.allclose(out, gout), numpy.absolute(out - gout)
except RuntimeError, e:
if not catch:
raise
# Different CUDA driver have different error message
assert (e.args[0].startswith(
'Cuda error: kSoftmaxWithBias_node_0: invalid configuration argument.\n') or
e.args[0].startswith('Cuda error: kSoftmaxWithBias_node_0: invalid argument.\n'))
cmp(2, 5) cmp(2, 5)
#we need to test n>32*1024 to check that we make the block loop. #we need to test n>32*1024 to check that we make the block loop.
...@@ -211,7 +199,11 @@ def test_softmax_with_bias(): ...@@ -211,7 +199,11 @@ def test_softmax_with_bias():
cmp(4, 2000) cmp(4, 2000)
cmp(4, 2024) cmp(4, 2024)
#GTX285 don't have enough shared mem for this case. #GTX285 don't have enough shared mem for this case.
cmp(4, 4074, True) cmp(4, 4074)
# The GTX580, 680 and kepler don't have enough shared memory.
cmp(2, 10000)
cmp(128, 16 * 1024)
cmp(128, 64 * 1024)
def test_softmax(): def test_softmax():
...@@ -219,9 +211,7 @@ def test_softmax(): ...@@ -219,9 +211,7 @@ def test_softmax():
This is basic test for GpuSoftmax This is basic test for GpuSoftmax
We check that we loop when their is too much block We check that we loop when their is too much block
We use slower code when there isn't enough shared memory
TODO: check that we loop when their is too much thread.(THIS IS
NOT IMPLEMENTED)
""" """
x = T.fmatrix('x') x = T.fmatrix('x')
...@@ -232,25 +222,12 @@ def test_softmax(): ...@@ -232,25 +222,12 @@ def test_softmax():
assert isinstance(f_gpu.maker.fgraph.toposort()[-2].op, assert isinstance(f_gpu.maker.fgraph.toposort()[-2].op,
cuda.nnet.GpuSoftmax) cuda.nnet.GpuSoftmax)
def cmp(n, m, catch=False): def cmp(n, m):
"""Some old card won't accept the configuration arguments of
this implementation. For those cases set catch=True to skip
those errors.
"""
try:
#print "test_softmax",n,m #print "test_softmax",n,m
data = numpy.arange(n * m, dtype='float32').reshape(n, m) data = numpy.arange(n * m, dtype='float32').reshape(n, m)
out = f(data) out = f(data)
gout = f_gpu(data) gout = f_gpu(data)
assert numpy.allclose(out, gout), numpy.absolute(out - gout) assert numpy.allclose(out, gout), numpy.absolute(out - gout)
except RuntimeError, e:
if not catch:
raise
# Different CUDA driver have different error message
assert (e.args[0].startswith(
'Cuda error: kSoftmax_node_0: invalid configuration argument.\n') or
e.args[0].startswith('Cuda error: kSoftmax_node_0: invalid argument.\n'))
#we need to test n>32*1024 to check that we make the block loop. #we need to test n>32*1024 to check that we make the block loop.
cmp(2, 5) cmp(2, 5)
...@@ -262,5 +239,9 @@ def test_softmax(): ...@@ -262,5 +239,9 @@ def test_softmax():
cmp(4, 1024) cmp(4, 1024)
cmp(4, 2000) cmp(4, 2000)
cmp(4, 2024) cmp(4, 2024)
#GTX285 don't have enough shared mem for this case. # The GTX285 don't have enough shared memory.
cmp(4, 4074, True) cmp(4, 4074)
# The GTX580, 680 and kepler don't have enough shared memory.
cmp(2, 10000)
cmp(128, 16 * 1024)
cmp(128, 64 * 1024)
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