junrushao1994 commented on PR #11234:
URL: https://github.com/apache/tvm/pull/11234#issuecomment-1120119963
That makes a lot of sense @wrong
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the
wrongtest commented on PR #11234:
URL: https://github.com/apache/tvm/pull/11234#issuecomment-1120111692
> Reading the cryptic error message, it seems that the `IntervalSet`
sometimes is used to represent `handle` vs `int`, so it cans `CanProveEqual` -
a bit weird to me, but definitely
This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a change to branch last-successful
in repository https://gitbox.apache.org/repos/asf/tvm.git
from 98aa41e329 [Relay] Flexible shape dispatch transformation (#11199)
add 4622dfefd6 [CI] fix docker
junrushao1994 commented on PR #11234:
URL: https://github.com/apache/tvm/pull/11234#issuecomment-1120091444
Reading the cryptic error message, it seems that the `IntervalSet` sometimes
is used to represent `handle` vs `int`, so it cans `CanProveEqual` - a bit
weird to me, but definitely
denise-k opened a new pull request, #69:
URL: https://github.com/apache/tvm-rfcs/pull/69
cc @tqchen @areusch
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To
adstraw opened a new pull request, #11239:
URL: https://github.com/apache/tvm/pull/11239
Captures Hexagon gtest output in a string and passes back over the FFI.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL
tkonolige commented on issue #10842:
URL: https://github.com/apache/tvm/issues/10842#issuecomment-1120079854
@jackyflame Are you still having this issue? If so, could you provide some
more information to help us debug? For example: what version of bash and sh are
you using, what kind of
tkonolige opened a new pull request, #11238:
URL: https://github.com/apache/tvm/pull/11238
Refactor roofline_analysis to use a pass instrument to save TIR code from
compilation for feature extraction. This should support different compilation
pipelines and avoids recompiling the module
This is an automated email from the ASF dual-hosted git repository.
wuwei pushed a change to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
from 31be30062b [TIR] Add schedule primitive SetAxisSeparator (#11225)
add 5393f6dc2d Update debugger.rst (#11231)
No new
vinx13 merged PR #11231:
URL: https://github.com/apache/tvm/pull/11231
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
csullivan commented on PR #11225:
URL: https://github.com/apache/tvm/pull/11225#issuecomment-1120051709
Many thanks @vinx13 @quic-sanirudh @junrushao1994 @Lunderberg, this is
merged!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
This is an automated email from the ASF dual-hosted git repository.
csullivan pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new 31be30062b [TIR] Add schedule primitive
csullivan merged PR #11225:
URL: https://github.com/apache/tvm/pull/11225
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
masahi commented on PR #11190:
URL: https://github.com/apache/tvm/pull/11190#issuecomment-1120048336
Yes, please send another job
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
areusch commented on issue #11220:
URL: https://github.com/apache/tvm/issues/11220#issuecomment-1120047383
sorry for this break! can we close now that the revert is merged?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and
mehrdadh opened a new issue, #11237:
URL: https://github.com/apache/tvm/issues/11237
Running hexagon launcher on HDK boards shows this error after enabling gtest:
```
[08500/02] 34:12.022 7567:e85: CDSP: remote_handle_open: opened handle 0x0
for apps_remotectl 0123
areusch commented on PR #60:
URL: https://github.com/apache/tvm-rfcs/pull/60#issuecomment-1120042223
@MichaelJKlaiber @cgerum any more thoughts on @manupa-arm 's two questions
above? would be great to get this RFC in!
--
This is an automated message from the Apache Git Service.
To
areusch commented on PR #11184:
URL: https://github.com/apache/tvm/pull/11184#issuecomment-1120040692
thanks @quic-sanirudh and sorry for the delay!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go
areusch merged PR #11184:
URL: https://github.com/apache/tvm/pull/11184
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
This is an automated email from the ASF dual-hosted git repository.
areusch pushed a change to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
from 98aa41e329 [Relay] Flexible shape dispatch transformation (#11199)
add 4622dfefd6 [CI] fix docker group exists with
mehrdadh opened a new pull request, #11236:
URL: https://github.com/apache/tvm/pull/11236
Disable since it breaks the build on HDK boards.
cc @adstraw @csullivan
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use
This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a change to branch last-successful
in repository https://gitbox.apache.org/repos/asf/tvm.git
from fb32997369 [CI] Update all Docker Images to 20220505-060045-500703308
(#11219)
add 98aa41e329 [Relay]
Lunderberg commented on PR #11235:
URL: https://github.com/apache/tvm/pull/11235#issuecomment-1120012794
Also, name suggestions for "non-surjective inverse" would be appreciated.
It's something of a mouthful, and describes what it is doing without describing
why it would be desired.
--
Lunderberg opened a new pull request, #11235:
URL: https://github.com/apache/tvm/pull/11235
The goal of this PR is to allow `IndexMap` to express transformations that
introduce padding in the transformed shape. For an arbitrary input shape, the
new method `IndexMap.non_surjective_inverse`
vinx13 commented on PR #11234:
URL: https://github.com/apache/tvm/pull/11234#issuecomment-1119996674
CUDA will call `cache_write` to introduce a cache block first, so the block
used for `reverse_compute_at` is different, and the bindings we are dealing
with might also be different
--
vinx13 opened a new pull request, #11234:
URL: https://github.com/apache/tvm/pull/11234
This PR fixed `compute_at` / `reverse_compute_at` when block access regions
contain constants (e.g. `A[0, vi, vj]`).
Background:
https://github.com/apache/tvm/pull/11187 added additional
Wheest commented on PR #11231:
URL: https://github.com/apache/tvm/pull/11231#issuecomment-1119985815
> @Wheest can you create a new issue with exactly what you said above. That
way we can keep track of the problem.
Good point, thanks @tkonolige. The issue is #11233.
--
This
Wheest opened a new issue, #11233:
URL: https://github.com/apache/tvm/issues/11233
### Expected behavior
Compiling a model with `debug_executor` allows one to run the model and get
the tracing output.
### Actual behavior
Process fails during the module creation:
NicolaLancellotti commented on PR #11226:
URL: https://github.com/apache/tvm/pull/11226#issuecomment-1119979328
Thank you very much!
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
tkonolige commented on PR #11231:
URL: https://github.com/apache/tvm/pull/11231#issuecomment-1119973529
@Wheest can you create a new issue with exactly what you said above. That
way we can keep track of the problem.
--
This is an automated message from the Apache Git Service.
To respond
Wheest commented on PR #11231:
URL: https://github.com/apache/tvm/pull/11231#issuecomment-1119970240
That's correct @tkonolige, [this Python
script](https://gist.github.com/Wheest/9ad2d6a47bbd2cfaa4be530c68ba2f6c) shows
the issue.
The script has three modes: `["tutorial", "alt",
This is an automated email from the ASF dual-hosted git repository.
jwfromm pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new 98aa41e329 [Relay] Flexible shape dispatch
jwfromm merged PR #11199:
URL: https://github.com/apache/tvm/pull/11199
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
mkatanbaf opened a new pull request, #11232:
URL: https://github.com/apache/tvm/pull/11232
This adds logging to rpc communications, by adding a wrapper around
"RPCChannel". The wrapper, "RPCChannelLogging", has a "NanoRPCListener" that
only listens to ongoing communication and log them in
Wheest opened a new pull request, #11231:
URL: https://github.com/apache/tvm/pull/11231
Updating debugger docs to reflect a way of using the debugger that works,
see [TVM forum
post](https://discuss.tvm.apache.org/t/runnig-a-model-with-tvm-debugger/9869/8?u=wheest).
@tkonolige, this
jsheng-jian commented on PR #11190:
URL: https://github.com/apache/tvm/pull/11190#issuecomment-1119903278
It looks the CI failure is resolved in main branch, do I need to rebase my
changes?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log
quic-sanirudh commented on PR #11225:
URL: https://github.com/apache/tvm/pull/11225#issuecomment-1119853001
> we can have a user convenience API backed by these two schedule primitives
Ah okay, thanks for the reply @vinx13
--
This is an automated message from the Apache Git
vinx13 commented on code in PR #11225:
URL: https://github.com/apache/tvm/pull/11225#discussion_r867038052
##
src/tir/schedule/transform.h:
##
@@ -66,6 +72,122 @@ Array ReplaceBuffer(Array
regions, const Buffer& sou
Array ReplaceBuffer(Array match_buffers,
const Buffer&
grant-arm opened a new pull request, #11230:
URL: https://github.com/apache/tvm/pull/11230
This PR implements [Inclusive Language RFC
(#68)](https://github.com/apache/tvm-rfcs/pull/68)
The RFC proposes to remove some non-inclusive language from the existing TVM
codebase and
This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a change to branch last-successful
in repository https://gitbox.apache.org/repos/asf/tvm.git
from ff7efe767a Revert "Implemented rpc logging (#10967)" (#11227)
add f9d9221739 [COMMUNITY] Nicola
grant-arm opened a new pull request, #68:
URL: https://github.com/apache/tvm-rfcs/pull/68
This RFC proposes to:
* Remove some non-inclusive language from the existing TVM codebase and
documentation.
* Introduce a linting script to CI that will prevent these non-inclusive
terms from
vinx13 commented on PR #11225:
URL: https://github.com/apache/tvm/pull/11225#issuecomment-1119827206
@quic-sanirudh Thanks for the comments. The motivation to have a separate
schedule primitive is to decouple the logical-physical mapping from generic
layout transformation which doesn't
manupa-arm commented on code in PR #10251:
URL: https://github.com/apache/tvm/pull/10251#discussion_r867009590
##
tests/python/contrib/test_ethosu/test_replace_conv2d.py:
##
@@ -132,6 +132,28 @@ def _create_serial_conv2d_params(
]
+def get_conv2d_args(call,
leandron commented on PR #11229:
URL: https://github.com/apache/tvm/pull/11229#issuecomment-1119812970
cc @ashutosh-arm @grant-arm
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific
csullivan commented on code in PR #11225:
URL: https://github.com/apache/tvm/pull/11225#discussion_r866996238
##
src/tir/schedule/transform.h:
##
@@ -66,6 +72,122 @@ Array ReplaceBuffer(Array
regions, const Buffer& sou
Array ReplaceBuffer(Array match_buffers,
const Buffer&
csullivan commented on code in PR #11225:
URL: https://github.com/apache/tvm/pull/11225#discussion_r866976226
##
src/tir/schedule/transform.h:
##
@@ -66,6 +72,122 @@ Array ReplaceBuffer(Array
regions, const Buffer& sou
Array ReplaceBuffer(Array match_buffers,
const Buffer&
mehrdadh commented on PR #11223:
URL: https://github.com/apache/tvm/pull/11223#issuecomment-1119794266
closing due to https://github.com/apache/tvm/pull/11227
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL
mehrdadh closed pull request #11223: [Hot fix][RPC] Fix RPC build on windows
URL: https://github.com/apache/tvm/pull/11223
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To
areusch commented on PR #11227:
URL: https://github.com/apache/tvm/pull/11227#issuecomment-1119792891
sorry for the breakage everyone! we'll fix this and push a new PR.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use
This is an automated email from the ASF dual-hosted git repository.
leandron pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new fb32997369 [CI] Update all Docker Images to
leandron closed issue #11200: [CI Image] Update ci_lint with blocklint
URL: https://github.com/apache/tvm/issues/11200
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To
leandron merged PR #11219:
URL: https://github.com/apache/tvm/pull/11219
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
leandron closed issue #11202: [CI Image] Upgrade ci_wasm, ci_gpu, ci_cpu,
ci_i386, ci_hexagon, ci_qemu and ci_arm to get newer GoogleTest
URL: https://github.com/apache/tvm/issues/11202
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
This is an automated email from the ASF dual-hosted git repository.
leandron pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new f9d9221739 [COMMUNITY] Nicola Lancellotti ->
leandron merged PR #11226:
URL: https://github.com/apache/tvm/pull/11226
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
Mousius opened a new pull request, #11229:
URL: https://github.com/apache/tvm/pull/11229
This ensures that CMSIS-NN is only used when the batch size and layout are
correct for the library calls.
--
This is an automated message from the Apache Git Service.
To respond to the message,
quic-sanirudh commented on PR #11225:
URL: https://github.com/apache/tvm/pull/11225#issuecomment-1119688463
@Lunderberg @vinx13 This is great, thanks a lot, I would love to start using
this.
I do have a small doubt however. Is there a reason why we're introducing a
new schedule
tmoreau89 commented on PR #11225:
URL: https://github.com/apache/tvm/pull/11225#issuecomment-1119634333
cc @quic-sanirudh
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
ibsidorenko commented on PR #11228:
URL: https://github.com/apache/tvm/pull/11228#issuecomment-1119630885
cc: @manupa-arm @masahi @apeskov
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the
This is an automated email from the ASF dual-hosted git repository.
github-bot pushed a change to branch last-successful
in repository https://gitbox.apache.org/repos/asf/tvm.git
from bafa3e9c22 [Hexagon] Add AoT capability to Hexagon launcher (#11214)
add aa3bcd9d33 Implemented rpc
tmoreau89 commented on code in PR #11226:
URL: https://github.com/apache/tvm/pull/11226#discussion_r866817969
##
CONTRIBUTORS.md:
##
@@ -117,6 +117,7 @@ We do encourage everyone to work anything they are
interested in.
- [Marisa Kirisame](https://github.com/MarisaKirisame):
kparzysz-quic commented on PR #11181:
URL: https://github.com/apache/tvm/pull/11181#issuecomment-1119599284
> The clarification RFC says `T.buffer_decl creates a buffer alias if the
underlying data variable (.data field) overlaps with another buffer. ` And this
should be the unique way to
NicolaLancellotti commented on code in PR #11226:
URL: https://github.com/apache/tvm/pull/11226#discussion_r866684707
##
CONTRIBUTORS.md:
##
@@ -117,6 +117,7 @@ We do encourage everyone to work anything they are
interested in.
- [Marisa
ibsidorenko commented on code in PR #11228:
URL: https://github.com/apache/tvm/pull/11228#discussion_r866767745
##
python/tvm/relay/transform/transform.py:
##
@@ -261,32 +261,39 @@ def LazyGradientInit():
return _ffi_api.LazyGradientInit()
-def FoldConstantExpr(expr,
ibsidorenko commented on code in PR #11228:
URL: https://github.com/apache/tvm/pull/11228#discussion_r866767745
##
python/tvm/relay/transform/transform.py:
##
@@ -261,32 +261,39 @@ def LazyGradientInit():
return _ffi_api.LazyGradientInit()
-def FoldConstantExpr(expr,
leandron commented on code in PR #11228:
URL: https://github.com/apache/tvm/pull/11228#discussion_r866763690
##
python/tvm/relay/transform/transform.py:
##
@@ -261,32 +261,39 @@ def LazyGradientInit():
return _ffi_api.LazyGradientInit()
-def FoldConstantExpr(expr,
This is an automated email from the ASF dual-hosted git repository.
mousius pushed a change to branch update-images-with-shiny-new-tag
in repository https://gitbox.apache.org/repos/asf/tvm.git
omit d6b43e1d16 [CI] Update all Docker Images to 20220505-060045-500703308
add eae836cdf6 Fix
manupa-arm commented on PR #11227:
URL: https://github.com/apache/tvm/pull/11227#issuecomment-1119507247
This is merged in favor of unblocking the CI...
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to
This is an automated email from the ASF dual-hosted git repository.
manupa pushed a commit to branch main
in repository https://gitbox.apache.org/repos/asf/tvm.git
The following commit(s) were added to refs/heads/main by this push:
new ff7efe767a Revert "Implemented rpc logging (#10967)"
manupa-arm merged PR #11227:
URL: https://github.com/apache/tvm/pull/11227
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail:
echuraev commented on code in PR #11180:
URL: https://github.com/apache/tvm/pull/11180#discussion_r866715240
##
src/runtime/opencl/opencl_common.h:
##
@@ -459,20 +461,10 @@ class OpenCLTimerNode : public TimerNode {
virtual int64_t SyncAndGetElapsedNanos() { return
NicolaLancellotti commented on code in PR #11226:
URL: https://github.com/apache/tvm/pull/11226#discussion_r866684707
##
CONTRIBUTORS.md:
##
@@ -117,6 +117,7 @@ We do encourage everyone to work anything they are
interested in.
- [Marisa
leandron commented on code in PR #10915:
URL: https://github.com/apache/tvm/pull/10915#discussion_r866673240
##
src/relay/qnn/op/convolution.cc:
##
@@ -50,12 +50,15 @@ bool QnnConv2DRel(const Array& types, int num_inputs,
const Attrs& attrs,
if (data == nullptr || weight ==
leandron commented on code in PR #10915:
URL: https://github.com/apache/tvm/pull/10915#discussion_r866671261
##
tests/python/frontend/tflite/test_forward.py:
##
@@ -955,6 +964,7 @@ def _test_convolution(
is_depthwise=False,
quantized=False,
fp16_quantized=False,
NicolaLancellotti commented on code in PR #10959:
URL: https://github.com/apache/tvm/pull/10959#discussion_r866646140
##
src/tir/contrib/ethosu/passes.cc:
##
@@ -27,7 +27,17 @@
#include
#include
+#include
+
namespace tvm {
+
+/*!
+ * \brief The maximum number of
ibsidorenko opened a new pull request, #11228:
URL: https://github.com/apache/tvm/pull/11228
This PR is an attempt to revive
[PR#9164](https://github.com/apache/tvm/pull/9164) . It enables folding of
constants for QNN operations. Motivation to have this feature is BYOC use
cases. For some
ashutosh-arm commented on PR #11221:
URL: https://github.com/apache/tvm/pull/11221#issuecomment-1119412360
> > @grant-arm Thanks for the PR. Just for my understanding, why do we need
to align .rodata to 16 bytes? I have seen this for constants in the code
generated C files as well. Also,
leandron opened a new pull request, #11227:
URL: https://github.com/apache/tvm/pull/11227
This reverts commit aa3bcd9d3374878c5e958b842f51bfd82f0ebd9e, because it
fails on Windows CI as reported in issue #11220. PR #11223 tries to address it
but is is failing in the regular CI with testing
Mousius commented on PR #11221:
URL: https://github.com/apache/tvm/pull/11221#issuecomment-1119408205
> @grant-arm Thanks for the PR. Just for my understanding, why do we need to
align .rodata to 16 bytes? I have seen this for constants in the code generated
C files as well. Also, does it
ashutosh-arm commented on PR #11221:
URL: https://github.com/apache/tvm/pull/11221#issuecomment-1119399310
@grant-arm Thanks for the PR. Just for my understanding, why do we need to
align .rodata to 16 bytes? I have seen this for constants in the code generated
C files as well. Also, does
manupa-arm commented on code in PR #10959:
URL: https://github.com/apache/tvm/pull/10959#discussion_r866582888
##
src/tir/contrib/ethosu/passes.cc:
##
@@ -27,7 +27,17 @@
#include
#include
+#include
+
namespace tvm {
+
+/*!
+ * \brief The maximum number of movements
81 matches
Mail list logo