Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

[FIX,TOPI] Fix issue when running conv2d in autoscheduler #9900

Merged
merged 1 commit into from
Jan 12, 2022
Merged

[FIX,TOPI] Fix issue when running conv2d in autoscheduler #9900

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 8 additions & 4 deletions python/tvm/topi/generic/conv2d.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -143,8 +143,10 @@ def schedule_conv_NCHWc_cpu_common_int8(
# only in autotuning, input data of conv2d_NCHWc will be 4-D.
# skip this part during tuning to make records accurate.
# this part will be folded during Relay fold_constant pass.
s[data_vec].pragma(s[data_vec].op.axis[0], "debug_skip_region")
s[kernel_vec].pragma(s[kernel_vec].op.axis[0], "debug_skip_region")
if isinstance(data_vec.op, te.tensor.ComputeOp):
s[data_vec].pragma(s[data_vec].op.axis[0], "debug_skip_region")
if isinstance(kernel_vec.op, te.tensor.ComputeOp):
s[kernel_vec].pragma(s[kernel_vec].op.axis[0], "debug_skip_region")
elif isinstance(kernel_vec.op, te.tensor.ComputeOp) and kernel_vec.name == "kernel_vec":
# data and kernel are not pre-computed, schedule layout transform here.
# this should only be used by x86 conv2d_nchw, which is for
Expand Down Expand Up @@ -269,8 +271,10 @@ def schedule_conv_NCHWc_cpu_1x1_int8(
# only in autotuning, input data of conv2d_NCHWc will be 4-D.
# skip this part during tuning to make records accurate.
# this part will be folded during Relay fold_constant pass.
s[data_vec].pragma(s[data_vec].op.axis[0], "debug_skip_region")
s[kernel_vec].pragma(s[kernel_vec].op.axis[0], "debug_skip_region")
if isinstance(data_vec.op, te.tensor.ComputeOp):
s[data_vec].pragma(s[data_vec].op.axis[0], "debug_skip_region")
if isinstance(kernel_vec.op, te.tensor.ComputeOp):
s[kernel_vec].pragma(s[kernel_vec].op.axis[0], "debug_skip_region")
elif isinstance(kernel_vec.op, te.tensor.ComputeOp) and kernel_vec.name == "kernel_vec":
# data and kernel are not pre-computed, schedule layout transform here.
# this should only be used by x86 conv2d_nchw, which is for
Expand Down