1 回答
TA贡献1804条经验 获得超2个赞
可以说该发布的代码中至少有两个错误:
这不可能是正确的范围检查:
if x >= C.shape[0] and y >= C.shape[1]:
为了让我们决定网格中的特定线程不执行任何加载活动,我们要求要么超出
x
范围,要么y
超出范围。应该and
是一个or
.如果块中的所有线程都无法参与该语句,则在条件代码中使用是非法的。
cuda.syncthreads()
上面第 1 项中的前一个return
语句(即使从and
to更正or
)几乎保证了对于问题大小不能被线程块大小整除的这种非法行为。
因此,要解决这些问题,我们不能仅使用简单的return
语句来处理越界线程。相反,在加载点,如果计算出的全局加载索引(forA
或B
)在边界内(根据定义,共享索引在边界内),我们必须只允许线程从全局加载到共享内存。此外,在写入结果时,我们必须只写入在 的范围内的计算结果C
。
以下代码修复了这些项目。对于您给定的测试用例,它似乎可以正常工作:
$ cat t49.py
from numba import cuda, float32
import numpy as np
import math
@cuda.jit
def fast_matmul(A, B, C):
# Define an array in the shared memory
# The size and type of the arrays must be known at compile time
sA = cuda.shared.array(shape=(TPB, TPB), dtype=float32)
sB = cuda.shared.array(shape=(TPB, TPB), dtype=float32)
x, y = cuda.grid(2)
tx = cuda.threadIdx.x
ty = cuda.threadIdx.y
bpg = cuda.gridDim.x # blocks per grid
# Each thread computes one element in the result matrix.
# The dot product is chunked into dot products of TPB-long vectors.
tmp = float32(0.)
for i in range(bpg):
# Preload data into shared memory
sA[tx, ty] = 0
sB[tx, ty] = 0
if x < A.shape[0] and (ty+i*TPB) < A.shape[1]:
sA[tx, ty] = A[x, ty + i * TPB]
if y < B.shape[1] and (tx+i*TPB) < B.shape[0]:
sB[tx, ty] = B[tx + i * TPB, y]
# Wait until all threads finish preloading
cuda.syncthreads()
# Computes partial product on the shared memory
for j in range(TPB):
tmp += sA[tx, j] * sB[j, ty]
# Wait until all threads finish computing
cuda.syncthreads()
if x < C.shape[0] and y < C.shape[1]:
C[x, y] = tmp
#%%
x_h = np.arange(16).reshape([4,4])
y_h = np.ones([4,4])
z_h = np.zeros([4,4])
x_d = cuda.to_device(x_h)
y_d = cuda.to_device(y_h)
z_d = cuda.to_device(z_h)
TPB = 3
threadsperblock = (TPB, TPB)
blockspergrid_x = math.ceil(z_h.shape[0] / threadsperblock[0])
blockspergrid_y = math.ceil(z_h.shape[1] / threadsperblock[1])
blockspergrid = (blockspergrid_x, blockspergrid_y)
fast_matmul[blockspergrid, threadsperblock](x_d, y_d, z_d)
z_h = z_d.copy_to_host()
print(z_h)
print(x_h@y_h)
$ cuda-memcheck python t49.py
========= CUDA-MEMCHECK
[[ 6. 6. 6. 6.]
[22. 22. 22. 22.]
[38. 38. 38. 38.]
[54. 54. 54. 54.]]
[[ 6. 6. 6. 6.]
[22. 22. 22. 22.]
[38. 38. 38. 38.]
[54. 54. 54. 54.]]
========= ERROR SUMMARY: 0 errors
$
(请注意,在边界测试中使用and此处是正确的。与测试一组索引是否越界相比,测试一组索引是否为入界在布尔意义上是不同的。在入界测试中,我们要求两者都在界内。在出界测试中,任何一个索引越界都是不合格的)。
我并不是说上面的代码没有缺陷或适合任何特定目的。它旨在演示我发现的问题的可能修复方法。正如您所发现的,让共享内存平铺矩阵乘法在每个可以想象的配置中工作并非易事,并且除了此处显示的内容之外,我还没有对它进行测试。(例如,如果您决定使 TPB 大于 32,您会遇到其他问题。此外,原始发布的代码仅用于方阵乘法,这在一般非方阵情况下不起作用。)
如上所述,发布的代码和上面带有“修复”的代码将无法正确处理一般的非方形情况。我相信一些简单的修改将使我们能够处理非方形的情况。简而言之,我们必须将网格设置得足够大,以处理两个输入矩阵的维度,同时仍然只写入输出矩阵的界内值的结果。这是一个经过简单测试的示例:
$ cat t49.py
from numba import cuda, float32
import numpy as np
import math
@cuda.jit
def fast_matmul(A, B, C):
# Define an array in the shared memory
# The size and type of the arrays must be known at compile time
sA = cuda.shared.array(shape=(TPB, TPB), dtype=float32)
sB = cuda.shared.array(shape=(TPB, TPB), dtype=float32)
x, y = cuda.grid(2)
tx = cuda.threadIdx.x
ty = cuda.threadIdx.y
bpg = cuda.gridDim.x # blocks per grid
# Each thread computes one element in the result matrix.
# The dot product is chunked into dot products of TPB-long vectors.
tmp = float32(0.)
for i in range(bpg):
# Preload data into shared memory
sA[ty, tx] = 0
sB[ty, tx] = 0
if y < A.shape[0] and (tx+i*TPB) < A.shape[1]:
sA[ty, tx] = A[y, tx + i * TPB]
if x < B.shape[1] and (ty+i*TPB) < B.shape[0]:
sB[ty, tx] = B[ty + i * TPB, x]
# Wait until all threads finish preloading
cuda.syncthreads()
# Computes partial product on the shared memory
for j in range(TPB):
tmp += sA[ty, j] * sB[j, tx]
# Wait until all threads finish computing
cuda.syncthreads()
if y < C.shape[0] and x < C.shape[1]:
C[y, x] = tmp
#%%
x_h = np.arange(115).reshape([5,23])
y_h = np.ones([23,7])
z_h = np.zeros([5,7])
x_d = cuda.to_device(x_h)
y_d = cuda.to_device(y_h)
z_d = cuda.to_device(z_h)
#TPB must be an integer between 1 and 32
TPB = 32
threadsperblock = (TPB, TPB)
grid_y_max = max(x_h.shape[0],y_h.shape[0])
grid_x_max = max(x_h.shape[1],y_h.shape[1])
blockspergrid_x = math.ceil(grid_x_max / threadsperblock[0])
blockspergrid_y = math.ceil(grid_y_max / threadsperblock[1])
blockspergrid = (blockspergrid_x, blockspergrid_y)
fast_matmul[blockspergrid, threadsperblock](x_d, y_d, z_d)
z_h = z_d.copy_to_host()
print(z_h)
print(x_h@y_h)
$ cuda-memcheck python t49.py
========= CUDA-MEMCHECK
[[ 253. 253. 253. 253. 253. 253. 253.]
[ 782. 782. 782. 782. 782. 782. 782.]
[1311. 1311. 1311. 1311. 1311. 1311. 1311.]
[1840. 1840. 1840. 1840. 1840. 1840. 1840.]
[2369. 2369. 2369. 2369. 2369. 2369. 2369.]]
[[ 253. 253. 253. 253. 253. 253. 253.]
[ 782. 782. 782. 782. 782. 782. 782.]
[1311. 1311. 1311. 1311. 1311. 1311. 1311.]
[1840. 1840. 1840. 1840. 1840. 1840. 1840.]
[2369. 2369. 2369. 2369. 2369. 2369. 2369.]]
========= ERROR SUMMARY: 0 errors
$
x我还重新排序了and的含义(以及andy的用法)以修复上述代码中的性能问题。原始发布的文档代码中也存在相同的性能问题。txty
再次强调,不声称无缺陷。此外,我确信可以得出“更优化”的代码。然而,优化矩阵乘法是一项应该很快就会导致使用库实现的练习。此处使用cupy
GPU 方法应该是一种相当简单的方法,可以在 GPU 上利用高质量的矩阵乘法例程。
添加回答
举报