Skip to content
Projects
Groups
Snippets
Help
This project
Loading...
Sign in / Register
Toggle navigation
T
tic
Overview
Overview
Details
Activity
Cycle Analytics
Repository
Repository
Files
Commits
Branches
Tags
Contributors
Graph
Compare
Charts
Issues
0
Issues
0
List
Board
Labels
Milestones
Merge Requests
0
Merge Requests
0
CI / CD
CI / CD
Pipelines
Jobs
Schedules
Charts
Wiki
Wiki
Snippets
Snippets
Members
Members
Collapse sidebar
Close sidebar
Activity
Graph
Charts
Create a new issue
Jobs
Commits
Issue Boards
Open sidebar
wenyuanbo
tic
Commits
8ca12d87
Commit
8ca12d87
authored
Aug 17, 2017
by
Haichen Shen
Committed by
Tianqi Chen
Aug 17, 2017
Browse files
Options
Browse Files
Download
Email Patches
Plain Diff
Add tutorial for convolution in CUDA (#343)
parent
d2a98a05
Show whitespace changes
Inline
Side-by-side
Showing
1 changed file
with
226 additions
and
0 deletions
+226
-0
tutorials/python/opt_conv_cuda.py
+226
-0
No files found.
tutorials/python/opt_conv_cuda.py
0 → 100644
View file @
8ca12d87
"""How to optimize convolution on GPU
==================================
**Author**: `Haichen Shen <https://homes.cs.washington.edu/~haichen/>`_
In this tutorial, we will demonstrate how to write a high performance
convolution implementation in TVM. We use square size input tensors and filters
as an example, and assume the input to convolution has a large batch. In this
example, we use a different layout to store the data in order to achieve better
data locality. The buffer layout is HWCN, which stands for height, width,
channel, batch.
"""
################################################################
# Preparation and Algorithm
# -------------------------
#
# We use the fixed size for input tensors with 256 channels and 14 x 14
# dimensions. The batch size is 256. Convolution filters contain 512 filters
# of size 3 x 3. We use stride size 1 and padding size 1 for the
# convolution. The following code defines the convolution algorithm in TVM.
#
import
numpy
as
np
import
tvm
# The sizes of inputs and filters
batch
=
256
in_channel
=
256
out_channel
=
512
in_size
=
14
kernel
=
3
pad
=
1
stride
=
1
# Algorithm
A
=
tvm
.
placeholder
((
in_size
,
in_size
,
in_channel
,
batch
),
name
=
'A'
)
W
=
tvm
.
placeholder
((
kernel
,
kernel
,
in_channel
,
out_channel
),
name
=
'W'
)
out_size
=
(
in_size
-
kernel
+
pad
)
//
stride
+
1
# Pad input
Apad
=
tvm
.
compute
(
(
in_size
+
pad
,
in_size
+
pad
,
in_channel
,
batch
),
lambda
yy
,
xx
,
cc
,
nn
:
tvm
.
select
(
tvm
.
all
(
yy
>=
pad
,
yy
-
pad
<
in_size
,
xx
>=
pad
,
xx
-
pad
<
in_size
),
A
[
yy
-
pad
,
xx
-
pad
,
cc
,
nn
],
tvm
.
const
(
0.
)),
name
=
'Apad'
)
# Create reduction variables
rc
=
tvm
.
reduce_axis
((
0
,
in_channel
),
name
=
'rc'
)
ry
=
tvm
.
reduce_axis
((
0
,
kernel
),
name
=
'ry'
)
rx
=
tvm
.
reduce_axis
((
0
,
kernel
),
name
=
'rx'
)
# Compute the convolution
B
=
tvm
.
compute
(
(
out_size
,
out_size
,
out_channel
,
batch
),
lambda
yy
,
xx
,
ff
,
nn
:
tvm
.
sum
(
Apad
[
yy
*
stride
+
ry
,
xx
*
stride
+
rx
,
rc
,
nn
]
*
W
[
ry
,
rx
,
rc
,
ff
],
axis
=
[
ry
,
rx
,
rc
]),
name
=
'B'
)
###############################################################################
# Memory Hierarchy
# ----------------
#
# We first specify the memory hierarchy for buffers. The figure below shows the
# GPU memory hierarchy. One important difference from CPU memory hierarchy is
# that GPU provides a cache buffer called shared memory, which is managed by
# programmers. Thus how to maximize the data reuse in the shared memory is
# critical to achieve high performance in GPU kernels.
#
# .. image:: https://github.com/dmlc/web-data/raw/master/tvm/tutorial/gpu_memory_hierarchy.png
# :align: center
# :height: 319px
# :width: 271px
#
# In this example, we load both Apad and W into buffer AA and WW, which are
# stored in the shared memory. These bufferes will be later shared by all
# threads within the same thread block to compute the convolution. Each thread
# then loads its own part from shared buffer into their local registers, AL and
# WL. BL is a local cache of output B, which is also stored in the thread local
# registers.
#
# Designate the memory hierarchy
s
=
tvm
.
create_schedule
(
B
.
op
)
s
[
Apad
]
.
compute_inline
()
# compute Apad inline
AA
=
s
.
cache_read
(
Apad
,
'shared'
,
[
B
])
WW
=
s
.
cache_read
(
W
,
"shared"
,
[
B
])
AL
=
s
.
cache_read
(
AA
,
"local"
,
[
B
])
WL
=
s
.
cache_read
(
WW
,
"local"
,
[
B
])
BL
=
s
.
cache_write
(
B
,
"local"
)
###############################################################################
# Blocking
# --------
#
# The following code splits the workload into thread blocks and individual
# threads. We follow the blocking scheme in the matrix multiply. As shown in the
# figure below, given a pixel coordinate (y, x), a thread block is responsible
# for computing a region of block_factor x block_factor (64 x 64) for output
# channels and batch. Due to the limit of shared memory space, we only load step
# x block_factor (8 x 64) data from Apad and B each time to buffers in the
# shared memory.
#
# .. image:: https://github.com/dmlc/web-data/raw/master/tvm/tutorial/conv_gpu_blocking.png
# :align: center
# :height: 308px
# :width: 317px
#
# tile consts
tile
=
8
num_thread
=
8
block_factor
=
tile
*
num_thread
step
=
8
vthread
=
2
# Get the GPU thread indices
block_x
=
tvm
.
thread_axis
(
"blockIdx.x"
)
block_y
=
tvm
.
thread_axis
(
"blockIdx.y"
)
block_z
=
tvm
.
thread_axis
(
"blockIdx.z"
)
thread_x
=
tvm
.
thread_axis
((
0
,
num_thread
),
"threadIdx.x"
)
thread_y
=
tvm
.
thread_axis
((
0
,
num_thread
),
"threadIdx.y"
)
thread_xz
=
tvm
.
thread_axis
((
0
,
vthread
),
"vthread"
,
name
=
"vx"
)
thread_yz
=
tvm
.
thread_axis
((
0
,
vthread
),
"vthread"
,
name
=
"vy"
)
# Split the workloads
hi
,
wi
,
fi
,
ni
=
s
[
B
]
.
op
.
axis
bz
=
s
[
B
]
.
fuse
(
hi
,
wi
)
by
,
fi
=
s
[
B
]
.
split
(
fi
,
factor
=
block_factor
)
bx
,
ni
=
s
[
B
]
.
split
(
ni
,
factor
=
block_factor
)
# Bind the iteration variables to GPU thread indices
s
[
B
]
.
bind
(
bz
,
block_z
)
s
[
B
]
.
bind
(
by
,
block_y
)
s
[
B
]
.
bind
(
bx
,
block_x
)
###############################################################################
# Virtual Thread Split
# --------------------
#
# We further split the workload from a thread block to individual threads. To
# avoid *memory bank conflict*, we use virtual thread to split the area into 4
# parts, and then tile into 8x8 grids. Therefore, shown in the figure below,
# each thread computes 4 strided grids, where size of each grid is 4 x 4.
#
# .. image:: https://github.com/dmlc/web-data/raw/master/tvm/tutorial/conv_gpu_vthread.png
# :align: center
# :height: 188px
# :width: 268px
#
tyz
,
fi
=
s
[
B
]
.
split
(
fi
,
nparts
=
vthread
)
# virtual thread split
txz
,
ni
=
s
[
B
]
.
split
(
ni
,
nparts
=
vthread
)
# virtual thread split
ty
,
fi
=
s
[
B
]
.
split
(
fi
,
nparts
=
num_thread
)
tx
,
ni
=
s
[
B
]
.
split
(
ni
,
nparts
=
num_thread
)
s
[
B
]
.
reorder
(
bz
,
by
,
bx
,
tyz
,
txz
,
ty
,
tx
,
fi
,
ni
)
s
[
B
]
.
bind
(
tyz
,
thread_yz
)
s
[
B
]
.
bind
(
txz
,
thread_xz
)
s
[
B
]
.
bind
(
ty
,
thread_y
)
s
[
B
]
.
bind
(
tx
,
thread_x
)
###############################################################################
# Cooperative Fetching
# --------------------
#
# As mentioned before, each time step we need to transfer step x block_factor
# data from GPU global memory to shared memory. In order to reduce the memory
# transfer per thread, the following code lets threads in the same thread block
# coopertively fetch dependent data from global memory.
#
# Schedule BL local write
s
[
BL
]
.
compute_at
(
s
[
B
],
tx
)
yi
,
xi
,
fi
,
ni
=
s
[
BL
]
.
op
.
axis
ry
,
rx
,
rc
=
s
[
BL
]
.
op
.
reduce_axis
rco
,
rci
=
s
[
BL
]
.
split
(
rc
,
factor
=
step
)
s
[
BL
]
.
reorder
(
rco
,
ry
,
rx
,
rci
,
fi
,
ni
)
# Attach computation to iteration variables
s
[
AA
]
.
compute_at
(
s
[
BL
],
rx
)
s
[
WW
]
.
compute_at
(
s
[
BL
],
rx
)
s
[
AL
]
.
compute_at
(
s
[
BL
],
rci
)
s
[
WL
]
.
compute_at
(
s
[
BL
],
rci
)
# Schedule for A's shared memory load
yi
,
xi
,
ci
,
ni
=
s
[
AA
]
.
op
.
axis
ty
,
ci
=
s
[
AA
]
.
split
(
ci
,
nparts
=
num_thread
)
tx
,
ni
=
s
[
AA
]
.
split
(
ni
,
nparts
=
num_thread
)
_
,
ni
=
s
[
AA
]
.
split
(
ni
,
factor
=
4
)
s
[
AA
]
.
reorder
(
ty
,
tx
,
yi
,
xi
,
ci
,
ni
)
s
[
AA
]
.
bind
(
ty
,
thread_y
)
s
[
AA
]
.
bind
(
tx
,
thread_x
)
s
[
AA
]
.
vectorize
(
ni
)
# vectorize memory load
# Schedule for W's shared memory load
yi
,
xi
,
ci
,
fi
=
s
[
WW
]
.
op
.
axis
ty
,
ci
=
s
[
WW
]
.
split
(
ci
,
nparts
=
num_thread
)
tx
,
fi
=
s
[
WW
]
.
split
(
fi
,
nparts
=
num_thread
)
_
,
fi
=
s
[
WW
]
.
split
(
fi
,
factor
=
4
)
s
[
WW
]
.
reorder
(
ty
,
tx
,
yi
,
xi
,
ci
,
fi
)
s
[
WW
]
.
bind
(
ty
,
thread_y
)
s
[
WW
]
.
bind
(
tx
,
thread_x
)
s
[
WW
]
.
vectorize
(
fi
)
# vectorize memory load
###############################################################################
# Generate CUDA Kernel
# --------------------
#
# Finally we use TVM to generate and compile the CUDA kernel, and evaluate the
# latency of convolution.
#
func
=
tvm
.
build
(
s
,
[
A
,
W
,
B
],
'cuda'
)
ctx
=
tvm
.
gpu
(
0
)
a_np
=
np
.
random
.
uniform
(
size
=
(
in_size
,
in_size
,
in_channel
,
batch
))
.
astype
(
A
.
dtype
)
w_np
=
np
.
random
.
uniform
(
size
=
(
kernel
,
kernel
,
in_channel
,
out_channel
))
.
astype
(
W
.
dtype
)
a
=
tvm
.
nd
.
array
(
a_np
,
ctx
)
w
=
tvm
.
nd
.
array
(
w_np
,
ctx
)
b
=
tvm
.
nd
.
array
(
np
.
zeros
((
out_size
,
out_size
,
out_channel
,
batch
),
dtype
=
B
.
dtype
),
ctx
)
func
(
a
,
w
,
b
)
evaluator
=
func
.
time_evaluator
(
func
.
entry_name
,
ctx
,
number
=
1
)
print
(
'Convolution:
%
f ms'
%
(
evaluator
(
a
,
w
,
b
)
.
mean
*
1e3
))
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment