[TOPI] Depth wise Conv for NHWC by weitliu · Pull Request #325 · apache/tvm

weitliu · 2017-08-15T00:15:04Z

It still cannot pass the test case for the fusion with bn and relu. The main problem is in the scheduler does not work with the scale_shift_nhwc function, which is added in the nn.mapping.

Any help is valuable to me. Thank you!

…wise

tqchen · 2017-08-15T00:56:01Z

@Huyuwei do you mind help to do a review?

tqchen · 2017-08-15T01:44:59Z

+
+        s[temp].compute_inline()
+        if DepthwiseConv2d.op in s.outputs:
+            Output = DepthwiseConv2d


You will need to cache_write here so this compute block is in local register

tqchen · 2017-08-15T01:45:16Z

+            Output = DepthwiseConv2d
+        else:
+            Output = op.output(0)
+            #s[DepthwiseConv2d].set_scope("local")


set the scope of depthwise to be local

Huyuwei · 2017-08-15T03:58:16Z

according to @tqchen's comment in #315:

the schedule interface should take Array of Tensors(or a single Tensor) instead of op. Also in terms of naming, map suffix is removed from functions and file names, and we assume all complex functions like conv is able to schedule with follwup ewise ops

Besides, it's better to add topi.testing.depthwise_conv2d

Could you refer to this PR #328 and do the same thing for depthwise_conv2d_nhwc @wetliu

weitliu · 2017-08-15T17:55:11Z

I pulled the most recent TVM and try to build by make pylint. I got this error:
************* Module topi.nn.softmax
E: 22,15: Unexpected keyword argument 'axis' in function call (unexpected-keyword-arg)
E: 23, 8: No value for argument 'axis' in function call (no-value-for-parameter)

Any idea what I can do to fix it so that I can build by make pylint? Thank you.

tqchen · 2017-08-15T18:16:28Z

This is the problem of upstream, it shall be fixed by #332

weitliu · 2017-08-15T18:27:14Z

Can you please check what's happening as it cannot be built? @tqchen Is it the last version before you guys finish modifying the conv and depth conv? I don't want to modify the code again and again only because the upstream conv and deconv are still under developing. Thank you!

tqchen · 2017-08-15T18:28:36Z

You will need to do a rebase on the upstream master, i think the conv will be freezed for a while from now on

weitliu · 2017-08-15T18:29:15Z

OK. Thank you!

weitliu · 2017-08-15T20:11:44Z

[solved]
I cannot bind the scheduler after merging the newest pr. It seems to me the tag is messed up if I add the name "depthwise_conv2d_nchw," etc. Where do I need to fix the tags?

Thank you!

tqchen · 2017-08-15T21:10:55Z

+  auto kw = tvm::reduce_axis(tvm::Range{0, W->shape[1]}, "kw");
+  auto T = (pad_h == 0 && pad_w == 0)
+               ? I
+               : pad(I, {tvm::Expr(0), tvm::Expr(0), pad_h, pad_w});


This should be {0, pad_h, pad_w, 0}

tqchen · 2017-08-15T21:11:10Z

+                                         int stride_h = 1,
+                                         int stride_w = 1,
+                                         std::string name = "tensor",
+                                         std::string tag = kDepthwiseConv2d_nhwc) {


should be kDepthwiseConv2dNHWC

tqchen · 2017-08-15T21:16:19Z

please fix the updated comment and do another pass over the code

tqchen · 2017-08-16T00:10:47Z

Thanks, this is merged

…pache#325) (apache#326) * Added elementwise-add test * Fix typo * Fixed elem-wise ops for lists with len>2

…325) (#326) * Added elementwise-add test * Fix typo * Fixed elem-wise ops for lists with len>2

…pache#325) (apache#326) * Added elementwise-add test * Fix typo * Fixed elem-wise ops for lists with len>2

This PR add hash/equal/json support for shape tuple.

rebased

wetliu added 5 commits August 13, 2017 19:54

rename the nchw and pass the unit test; going to do it for nhwc depth…

2b586d2

…wise

bug with fusion

b65d03f

resolved conflicts

0c01fc0

nchw works fine; nhwc float32 problem remains

68f1292

still cannot bind them together

a9a2975

tqchen reviewed Aug 15, 2017

View reviewed changes

wetliu added 2 commits August 14, 2017 20:20

fusion works

f6108c4

resolved conflicts

ac264c2

wetliu and others added 3 commits August 14, 2017 22:54

modify after pr 328

d935fce

Merge branch 'master' of https://github.com/dmlc/tvm

1f373f9

Merge branch 'master' of https://github.com/dmlc/tvm

ce5da31

syntax fix

83703c1

weitang added 2 commits August 15, 2017 20:23

after merge, not working; cannot bind

4edcb4a

all bugs fixed; test cases pass

c1ae53d

tqchen requested changes Aug 15, 2017

View reviewed changes

minor fix on nn.h

183e145

tqchen approved these changes Aug 16, 2017

View reviewed changes

tqchen merged commit 989e99e into apache:master Aug 16, 2017

tqchen changed the title ~~[WIP] [TOPI] Depth wise Conv for NHWC~~ [TOPI] Depth wise Conv for NHWC Aug 16, 2017

tqchen pushed a commit to tqchen/tvm that referenced this pull request May 26, 2018

Fix keras frontend elementwise-ops for lists with len>2 (fixes issue a…

a146000

…pache#325) (apache#326) * Added elementwise-add test * Fix typo * Fixed elem-wise ops for lists with len>2

tqchen pushed a commit that referenced this pull request May 29, 2018

Fix keras frontend elementwise-ops for lists with len>2 (fixes issue #…

ab0b65c

…325) (#326) * Added elementwise-add test * Fix typo * Fixed elem-wise ops for lists with len>2

tqchen pushed a commit to tqchen/tvm that referenced this pull request Jul 6, 2018

Fix keras frontend elementwise-ops for lists with len>2 (fixes issue a…

9998f33

…pache#325) (apache#326) * Added elementwise-add test * Fix typo * Fixed elem-wise ops for lists with len>2

sergei-mironov pushed a commit to sergei-mironov/tvm that referenced this pull request Aug 8, 2018

Fix keras frontend elementwise-ops for lists with len>2 (fixes issue a…

cbd573b

…pache#325) (apache#326) * Added elementwise-add test * Fix typo * Fixed elem-wise ops for lists with len>2

vinx13 pushed a commit to vinx13/tvm that referenced this pull request Mar 9, 2022

[BugFix] warp_testcase & tvmscript printer (apache#325)

34fb83a

gigiblender pushed a commit to gigiblender/tvm that referenced this pull request Jan 19, 2023

[CONTAINER] Hash/Equal/JSON support for ShapeTuple (apache#325)

80710a8

This PR add hash/equal/json support for shape tuple.

tlopex pushed a commit to tlopex/tvm that referenced this pull request Mar 9, 2026

[fix] fix batch-attn prelogue & add tp e2e profiler (apache#325)

a992b6a

rebased

Uh oh!

Conversation

weitliu commented Aug 15, 2017

Uh oh!

tqchen commented Aug 15, 2017

Uh oh!

tqchen Aug 15, 2017

Choose a reason for hiding this comment

Uh oh!

tqchen Aug 15, 2017

Choose a reason for hiding this comment

Uh oh!

Huyuwei commented Aug 15, 2017

Uh oh!

weitliu commented Aug 15, 2017

Uh oh!

tqchen commented Aug 15, 2017

Uh oh!

weitliu commented Aug 15, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tqchen commented Aug 15, 2017

Uh oh!

weitliu commented Aug 15, 2017

Uh oh!

weitliu commented Aug 15, 2017 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

tqchen Aug 15, 2017

Choose a reason for hiding this comment

Uh oh!

tqchen Aug 15, 2017

Choose a reason for hiding this comment

Uh oh!

tqchen commented Aug 15, 2017

Uh oh!

tqchen commented Aug 16, 2017

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

weitliu commented Aug 15, 2017 •

edited

Loading

weitliu commented Aug 15, 2017 •

edited

Loading