[GitHub] [incubator-tvm] kevinthesun commented on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach.
kevinthesun commented on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866#issuecomment-585575763 Thanks @alexgl-github @anijain2305 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] kevinthesun merged pull request #4866: Optimize x86 conv3d_ndhwc using data packing approach.
kevinthesun merged pull request #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-tvm] branch master updated (70c6382 -> 8d94587)
This is an automated email from the ASF dual-hosted git repository. kevinthesun pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from 70c6382 [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess (#4543) add 8d94587 Optimize x86 conv3d_ndhwc using data packing approach. (#4866) No new revisions were added by this update. Summary of changes: python/tvm/autotvm/task/relay_integration.py | 1 + python/tvm/autotvm/task/topi_integration.py | 12 + topi/python/topi/nn/util.py | 36 +++ topi/python/topi/x86/conv3d.py | 369 --- topi/tests/python/test_topi_conv3d_ndhwc.py | 2 +- 5 files changed, 381 insertions(+), 39 deletions(-)
[GitHub] [incubator-tvm] FrozenGene commented on a change in pull request #4790: Fast exponent
FrozenGene commented on a change in pull request #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790#discussion_r378637381 ## File path: topi/python/topi/math.py ## @@ -449,3 +449,19 @@ def reinterpret(x, dtype): The result. """ return cpp.reinterpret(x, dtype) + + +def fast_exp(x): +"""Take exponential of input x using fastexp implementation Review comment: fastexp -> fast_exp 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] FrozenGene commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
FrozenGene commented on issue #4847: Return empty CSourceModule when no
lowered_funcs exists in Relay mod
URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585527191
> I did try just commenting out the assert and it seems to work. However, I
then ran into a new problem... Because what you produce is ostensibly a 'c
source' file (even if it's empty), all the external modules are also exported
as c files. This results in having to do a proper compile when you try and
merge all the modules together to produce a shared library. If one of your
external modules happens to be a binary artifact that's 150MB (my case), this
compile uses an incredible amount of RAM and takes a long time to complete. I
didn't get this problem with your first solution as it used an LLVMModule (I
think?) which results in the external modules becoming object files so all you
need to do is link, not perform a full compilation.
This is because our new `ModulePackImportsToLLVM` simplifies complication
and generate object file directly, it is designed for large artifact like your
case.
> The reason that LLVMModule simplifies compilation is because it remembers
the correct target triple. We can try to enhance the code by introducing an
Empty LLVM module(with the correct target) when it is available(so that Q2 can
be resolved) and fallback to the CSourceModule.
cc @FrozenGene who created the mechanism for the binary exporting
I agree we could create empty llvm module if we have llvm target support,
then we could fallback to CSourceModule. We could refer logic here:
```cpp
runtime::Module build(const Map>& inputs,
const Target& target_host,
const BuildConfig& config)
Target target_host_val = target_host;
if (!target_host.defined()) {
for (const auto& it : inputs) {
if (it.first->device_type == kDLCPU) {
target_host_val = it.first;
break;
}
}
}
if (!target_host_val.defined()) {
target_host_val = DefaultTargetHost(target_host_val);
}
```
Here, we have logic to detect and extract the correct target host. According
to this, we could know whether we have LLVM (with correct target).
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] FrozenGene edited a comment on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
FrozenGene edited a comment on issue #4847: Return empty CSourceModule when no
lowered_funcs exists in Relay mod
URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585527191
> I did try just commenting out the assert and it seems to work. However, I
then ran into a new problem... Because what you produce is ostensibly a 'c
source' file (even if it's empty), all the external modules are also exported
as c files. This results in having to do a proper compile when you try and
merge all the modules together to produce a shared library. If one of your
external modules happens to be a binary artifact that's 150MB (my case), this
compile uses an incredible amount of RAM and takes a long time to complete. I
didn't get this problem with your first solution as it used an LLVMModule (I
think?) which results in the external modules becoming object files so all you
need to do is link, not perform a full compilation.
This is because our new `ModulePackImportsToLLVM` simplifies compilation and
generate object file directly, it is designed for large artifact like your case.
> The reason that LLVMModule simplifies compilation is because it remembers
the correct target triple. We can try to enhance the code by introducing an
Empty LLVM module(with the correct target) when it is available(so that Q2 can
be resolved) and fallback to the CSourceModule.
cc @FrozenGene who created the mechanism for the binary exporting
I agree we could create empty llvm module if we have llvm target support,
then we could fallback to CSourceModule. We could refer logic here:
```cpp
runtime::Module build(const Map>& inputs,
const Target& target_host,
const BuildConfig& config)
Target target_host_val = target_host;
if (!target_host.defined()) {
for (const auto& it : inputs) {
if (it.first->device_type == kDLCPU) {
target_host_val = it.first;
break;
}
}
}
if (!target_host_val.defined()) {
target_host_val = DefaultTargetHost(target_host_val);
}
```
Here, we have logic to detect and extract the correct target host. According
to this, we could know whether we have LLVM (with correct target).
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] FrozenGene commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess
FrozenGene commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess URL: https://github.com/apache/incubator-tvm/pull/4543#issuecomment-585513267 Thanks everyone , merged now 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] FrozenGene merged pull request #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess
FrozenGene merged pull request #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess URL: https://github.com/apache/incubator-tvm/pull/4543 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-tvm] branch master updated (51a265a -> 70c6382)
This is an automated email from the ASF dual-hosted git repository. zhaowu pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from 51a265a [REFACTOR][PY][API-CHANGE] Establish tvm.target add 70c6382 [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess (#4543) No new revisions were added by this update. Summary of changes: python/tvm/relay/frontend/tflite.py | 197 +++ tests/python/frontend/tflite/test_forward.py | 48 +++ 2 files changed, 245 insertions(+)
[GitHub] [incubator-tvm] alexgl-github commented on issue #4790: Fast exponent
alexgl-github commented on issue #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790#issuecomment-585506768 > Right. I think `fast_exp` fits better with current naming style. @anijain2305 I've changed fastexp to fast_exp 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] masahi commented on issue #4497: [Relay] Add a PyTorch to Relay Parser
masahi commented on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585505734 I don't know what is the plan for updating topi, but that is out of scope for this PR anyway. I think it's fine to disable the mobilenet v2 test along with alexnet and vgg for now and move forward. @zhiics @kazum @FrozenGene please take another look and give approval if it is good to go. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexgl-github edited a comment on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach.
alexgl-github edited a comment on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866#issuecomment-585502820 > Thank you for this work! It would be great if you can provide benchmarking data comparing tvm conv3d performance VS existing solution(tensorflow + mkldnn?) to see where current implementation stands. @kevinthesun Below is benchmark results for certain data/kernel combinations, run on 2 core Intel "ivybridge" vs TF 1.15 + mkldnn. X: value means speedup (or slowdown) of TVM model vs same TF model TVM: 0.007 sec; X: 1.645; TF: 0.011 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 3, 3, 1, 8) TVM: 0.019 sec; X: 1.218; TF: 0.023 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 7, 7, 1, 8) TVM: 0.054 sec; X: 1.490; TF: 0.080 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 3, 3, 8, 16) TVM: 0.262 sec; X: 0.869; TF: 0.228 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 7, 7, 8, 16) TVM: 0.013 sec; X: 1.290; TF: 0.016 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(3, 3, 3, 1, 8) TVM: 0.114 sec; X: 1.148; TF: 0.131 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(7, 7, 7, 1, 8) TVM: 0.146 sec; X: 1.058; TF: 0.154 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(3, 3, 3, 8, 16) TVM: 2.432 sec; X: 0.591; TF: 1.436 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(7, 7, 7, 8, 16) 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexgl-github edited a comment on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach.
alexgl-github edited a comment on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866#issuecomment-585502820 > Thank you for this work! It would be great if you can provide benchmarking data comparing tvm conv3d performance VS existing solution(tensorflow + mkldnn?) to see where current implementation stands. @kevinthesun Below is benchmark results for certain data/kernel combinations, run on 2 core Intel "ivybridge" vs TF 1.15 + mkldnn. X: means speedup (or slowdown) of TVM model vs same TF model TVM: 0.007 sec; X: 1.645; TF: 0.011 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 3, 3, 1, 8) TVM: 0.019 sec; X: 1.218; TF: 0.023 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 7, 7, 1, 8) TVM: 0.054 sec; X: 1.490; TF: 0.080 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 3, 3, 8, 16) TVM: 0.262 sec; X: 0.869; TF: 0.228 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 7, 7, 8, 16) TVM: 0.013 sec; X: 1.290; TF: 0.016 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(3, 3, 3, 1, 8) TVM: 0.114 sec; X: 1.148; TF: 0.131 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(7, 7, 7, 1, 8) TVM: 0.146 sec; X: 1.058; TF: 0.154 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(3, 3, 3, 8, 16) TVM: 2.432 sec; X: 0.591; TF: 1.436 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(7, 7, 7, 8, 16) 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexgl-github edited a comment on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach.
alexgl-github edited a comment on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866#issuecomment-585502820 > Thank you for this work! It would be great if you can provide benchmarking data comparing tvm conv3d performance VS existing solution(tensorflow + mkldnn?) to see where current implementation stands. @kevinthesun Below is benchmark results for certain data/kernel combinations, run on 2 core Intel "ivybridge" vs TV 1.15 + mkldnn. X: means speedup (or slowdown) of TVM model vs same TF model TVM: 0.007 sec; X: 1.645; TF: 0.011 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 3, 3, 1, 8) TVM: 0.019 sec; X: 1.218; TF: 0.023 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 7, 7, 1, 8) TVM: 0.054 sec; X: 1.490; TF: 0.080 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 3, 3, 8, 16) TVM: 0.262 sec; X: 0.869; TF: 0.228 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 7, 7, 8, 16) TVM: 0.013 sec; X: 1.290; TF: 0.016 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(3, 3, 3, 1, 8) TVM: 0.114 sec; X: 1.148; TF: 0.131 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(7, 7, 7, 1, 8) TVM: 0.146 sec; X: 1.058; TF: 0.154 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(3, 3, 3, 8, 16) TVM: 2.432 sec; X: 0.591; TF: 1.436 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(7, 7, 7, 8, 16) 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexgl-github commented on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach.
alexgl-github commented on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866#issuecomment-585502820 > Thank you for this work! It would be great if you can provide benchmarking data comparing tvm conv3d performance VS existing solution(tensorflow + mkldnn?) to see where current implementation stands. @kevinthesun Below is benchmark results for certain data/kernel combinations, run on 2 cpy Intel "ivybridge" vs TV 1.15 + mkldnn. X: means speedup (or slowdown) of TVM model vs same TF model TVM: 0.007 sec; X: 1.645; TF: 0.011 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 3, 3, 1, 8) TVM: 0.019 sec; X: 1.218; TF: 0.023 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(1, 7, 7, 1, 8) TVM: 0.054 sec; X: 1.490; TF: 0.080 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 3, 3, 8, 16) TVM: 0.262 sec; X: 0.869; TF: 0.228 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(1, 7, 7, 8, 16) TVM: 0.013 sec; X: 1.290; TF: 0.016 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(3, 3, 3, 1, 8) TVM: 0.114 sec; X: 1.148; TF: 0.131 sec; input_shape=(1, 16, 256, 256, 1) ; kernel_shape=(7, 7, 7, 1, 8) TVM: 0.146 sec; X: 1.058; TF: 0.154 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(3, 3, 3, 8, 16) TVM: 2.432 sec; X: 0.591; TF: 1.436 sec; input_shape=(1, 16, 256, 256, 8) ; kernel_shape=(7, 7, 7, 8, 16) 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] wpan11nv commented on issue #4867: [TOPI][CUDA] Enable vectorization on fp16 type
wpan11nv commented on issue #4867: [TOPI][CUDA] Enable vectorization on fp16 type URL: https://github.com/apache/incubator-tvm/pull/4867#issuecomment-585497119 Thanks all for the suggestions! Tests updated. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh
zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay
pass to use fast exp/tanh
URL: https://github.com/apache/incubator-tvm/pull/4873#discussion_r378594559
##
File path: src/relay/backend/build_module.cc
##
@@ -307,6 +307,10 @@ class RelayBuildModule : public runtime::ModuleNode {
}
pass_seqs.push_back(transform::FoldConstant());
+// Fast math optimizations.
+pass_seqs.push_back(transform::FastMath());
+pass_seqs.push_back(transform::FoldConstant());
Review comment:
Yeah, then we probably want to move it after alteroplayout.
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[incubator-tvm] branch master updated (79cfab0 -> 51a265a)
This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from 79cfab0 [JVM] Update the runtime PackedFunc for module
add 51a265a [REFACTOR][PY][API-CHANGE] Establish tvm.target
No new revisions were added by this update.
Summary of changes:
docs/api/python/index.rst | 2 +-
docs/api/python/target.rst | 1 +
python/tvm/__init__.py | 2 -
python/tvm/_ffi/runtime_ctypes.py | 35 +-
python/tvm/autotvm/task/dispatcher.py | 4 +-
python/tvm/build_module.py | 5 +-
python/tvm/contrib/clang.py| 9 +-
python/tvm/contrib/rocm.py | 10 +-
python/tvm/hybrid/calls.py | 4 +-
python/tvm/hybrid/runtime.py | 2 +-
python/tvm/intrin.py | 4 +-
python/tvm/relay/backend/vm.py | 2 +-
python/tvm/relay/build_module.py | 2 +-
python/tvm/relay/qnn/op/legalizations.py | 4 +-
python/tvm/relay/quantize/_calibrate.py| 4 +-
python/tvm/relay/quantize/_partition.py| 6 +-
python/tvm/target.py | 559 -
python/tvm/target/__init__.py | 62 +++
python/tvm/{ir => target}/_ffi_api.py | 4 +-
python/tvm/{ => target}/codegen.py | 46 +-
python/tvm/{ => target}/datatype.py| 24 +-
python/tvm/target/generic_func.py | 271 ++
python/tvm/target/target.py| 272 ++
src/runtime/c_runtime_api.cc | 14 +-
src/target/codegen.cc | 2 +-
src/target/datatype/registry.cc| 9 +-
src/target/generic_func.cc | 10 +-
src/target/llvm/llvm_module.cc | 13 +-
src/target/target.cc | 10 +-
src/tir/pass/lower_custom_datatypes.cc | 24 +-
tests/cpp/build_module_test.cc | 6 +-
tests/python/integration/test_dot.py | 2 +-
tests/python/relay/test_op_level2.py | 4 +-
tests/python/unittest/test_autotvm_common.py | 2 +-
tests/python/unittest/test_codegen_c_host.py | 2 +-
tests/python/unittest/test_codegen_device.py | 8 +-
tests/python/unittest/test_codegen_llvm.py | 8 +-
tests/python/unittest/test_codegen_static_init.py | 4 +-
tests/python/unittest/test_codegen_vm_basic.py | 2 +-
tests/python/unittest/test_codegen_x86.py | 4 +-
.../unittest/test_custom_datatypes_mybfloat16.py | 18 +-
tests/python/unittest/test_lang_target.py | 2 +-
tests/python/unittest/test_runtime_extension.py| 2 +-
tests/python/unittest/test_runtime_module_load.py | 2 +-
topi/python/topi/arm_cpu/conv2d.py | 6 +-
topi/python/topi/bifrost/conv2d.py | 2 +-
topi/python/topi/bifrost/transforms.py | 2 +-
topi/python/topi/cuda/batch_matmul.py | 4 +-
topi/python/topi/cuda/conv1d.py| 4 +-
topi/python/topi/cuda/conv1d_transpose_ncw.py | 2 +-
topi/python/topi/cuda/conv2d.py| 6 +-
topi/python/topi/cuda/conv2d_direct.py | 2 +-
topi/python/topi/cuda/conv2d_transpose_nchw.py | 2 +-
topi/python/topi/cuda/conv2d_winograd.py | 6 +-
topi/python/topi/cuda/conv3d.py| 6 +-
topi/python/topi/cuda/conv3d_direct.py | 2 +-
topi/python/topi/cuda/deformable_conv2d.py | 2 +-
topi/python/topi/cuda/dense.py | 8 +-
topi/python/topi/cuda/depthwise_conv2d.py | 4 +-
topi/python/topi/cuda/group_conv2d_nchw.py | 2 +-
topi/python/topi/cuda/injective.py | 2 +-
topi/python/topi/cuda/nms.py | 18 +-
topi/python/topi/cuda/nn.py| 2 +-
topi/python/topi/cuda/pooling.py | 4 +-
topi/python/topi/cuda/rcnn/proposal.py | 6 +-
topi/python/topi/cuda/reduction.py | 4 +-
topi/python/topi/cuda/sort.py | 4 +-
topi/python/topi/cuda/ssd/multibox.py | 6 +-
topi/python/topi/cuda/vision.py| 2 +-
topi/python/topi/generic/extern.py | 2 +-
topi/python/topi/generic/injective.py | 2 +-
topi/python/topi/generic/nn.py | 6 +-
topi/python/topi/generic/vision.py | 4 +-
topi/python/topi/intel_graphics/conv2d.py | 2 +-
.../python/topi/intel_graphics/depthwise_conv2d.py | 4 +-
topi/python/
[GitHub] [incubator-tvm] tqchen merged pull request #4872: [REFACTOR][PY][API-CHANGE] Establish tvm.target
tqchen merged pull request #4872: [REFACTOR][PY][API-CHANGE] Establish tvm.target URL: https://github.com/apache/incubator-tvm/pull/4872 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] anijain2305 commented on issue #4790: Fast exponent
anijain2305 commented on issue #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790#issuecomment-585489041 Right. I think `fast_exp` fits better with current naming style. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] anijain2305 commented on a change in pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh
anijain2305 commented on a change in pull request #4873: [Relay][FastMath]
Relay pass to use fast exp/tanh
URL: https://github.com/apache/incubator-tvm/pull/4873#discussion_r378589877
##
File path: src/relay/backend/build_module.cc
##
@@ -307,6 +307,10 @@ class RelayBuildModule : public runtime::ModuleNode {
}
pass_seqs.push_back(transform::FoldConstant());
+// Fast math optimizations.
+pass_seqs.push_back(transform::FastMath());
+pass_seqs.push_back(transform::FoldConstant());
Review comment:
OR I can just move the FastMath before the previous FoldConstant.
I don't think FastMath brings any new const in the graph. I just put it
there because FastMath becomes the last pass before FuseOps, and we want the
graph to be fully optimized.
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #4790: Fast exponent
tqchen commented on issue #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790#issuecomment-585487279 Overall looks OK, it would be great if we can decide a consistent naming convention. In this case, we can have `fastexp` vs `fast_exp` 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] anijain2305 commented on issue #4873: [Relay][FastMath] Relay pass to use fast exp/tanh
anijain2305 commented on issue #4873: [Relay][FastMath] Relay pass to use fast exp/tanh URL: https://github.com/apache/incubator-tvm/pull/4873#issuecomment-585487055 Thanks @zhiics for quick review. I will incorporate your comments once the parent PR gets merged in. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh
zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh URL: https://github.com/apache/incubator-tvm/pull/4873#discussion_r378587946 ## File path: src/relay/pass/fast_math.cc ## @@ -0,0 +1,79 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +/*! + * \file fast_math.cc + * \brief Replaces non linear activation functions with their fast but appproximate counterparts. Review comment: s/appproximate/approximate 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh
zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh URL: https://github.com/apache/incubator-tvm/pull/4873#discussion_r378588075 ## File path: src/relay/pass/fast_math.cc ## @@ -0,0 +1,79 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under one + * or more contributor license agreements. See the NOTICE file + * distributed with this work for additional information + * regarding copyright ownership. The ASF licenses this file + * to you under the Apache License, Version 2.0 (the + * "License"); you may not use this file except in compliance + * with the License. You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, + * software distributed under the License is distributed on an + * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY + * KIND, either express or implied. See the License for the + * specific language governing permissions and limitations + * under the License. + */ + +/*! + * \file fast_math.cc + * \brief Replaces non linear activation functions with their fast but appproximate counterparts. + */ +#include +#include +#include +#include +#include +#include "./pattern_util.h" Review comment: "pattern_util.h" 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh
zhiics commented on a change in pull request #4873: [Relay][FastMath] Relay
pass to use fast exp/tanh
URL: https://github.com/apache/incubator-tvm/pull/4873#discussion_r378587789
##
File path: src/relay/backend/build_module.cc
##
@@ -307,6 +307,10 @@ class RelayBuildModule : public runtime::ModuleNode {
}
pass_seqs.push_back(transform::FoldConstant());
+// Fast math optimizations.
+pass_seqs.push_back(transform::FastMath());
+pass_seqs.push_back(transform::FoldConstant());
Review comment:
We should create a SequentialPass instead of doing this back-to-back. For
example, you can return `Sequential({FastMathPass, FoldConstant()},
"FastMath")` in fast_math.cc, where `FastMathPass = CreateFunctionPass()`
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] anijain2305 commented on issue #4790: Fast exponent
anijain2305 commented on issue #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790#issuecomment-585485172 @tqchen @FrozenGene Can you please check if the changes you requested are addressed? 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] anijain2305 opened a new pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh
anijain2305 opened a new pull request #4873: [Relay][FastMath] Relay pass to use fast exp/tanh URL: https://github.com/apache/incubator-tvm/pull/4873 As Title, dependent on the following PR - https://github.com/apache/incubator-tvm/pull/4790 @alexgl-github @zhiics @masahi 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser
alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585414848 > Do you think you can fix the padding issue? I haven't looked into what is going on, we may need to update cuda topi schedule. > > It is ok if you leave that to me, I can take a look in the follow up PR. I can try taking a look today but I don't have too much experience here and wouldn't want to block this going out if everything else seems okay. https://github.com/zxy844288792/tvm/blob/00fd0c45ac6a6c880c1f2c21eb79ca86b4778ba2/python/tvm/relay/op/nn/util.py#L43, not sure if obvious but the error is from converting 2d padding to 4d padding here (ie: just returning padding here fixes the cuda error). I spoke with @zxy844288792 and he agrees it is probably a schedule issue as #4787 PR changed the workloads and the tophub file (guessing where default tuned schedules are) are not updated yet (correct me if any of this seems off). 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics edited a comment on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
zhiics edited a comment on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585464161 For Q2, a simpler way would be feeding the first argument of CSourceModuleCreate with ";" instead of empty string. But LLVM module should be faster for compilation. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser
alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585414848 > Do you think you can fix the padding issue? I haven't looked into what is going on, we may need to update cuda topi schedule. > > It is ok if you leave that to me, I can take a look in the follow up PR. I can try taking a look today but I don't have too much experience here and wouldn't want to block this going out if everything else seems okay. https://github.com/zxy844288792/tvm/blob/00fd0c45ac6a6c880c1f2c21eb79ca86b4778ba2/python/tvm/relay/op/nn/util.py#L43, not sure if obvious but the error is from converting 2d padding to 4d padding here (ie: just returning padding here fixes the cuda error). I spoke with @zxy844288792 and he agrees it is probably a schedule issue as #4787 PR changed the workloads and the tophub file (guessing where default tuned schedules are) are not updated yet. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics edited a comment on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
zhiics edited a comment on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585464161 For Q2, a simpler way would be feeding it with ";" instead of empty string. But LLVM module should be faster for compilation. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
zhiics commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585464161 For Q2, a simpler would be feeding it with ";" instead of empty string. But LLVM module should be faster for compilation. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen edited a comment on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
tqchen edited a comment on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585462552 The reason that LLVMModule simplifies compilation is because it remembers the correct target triple. We can try to enhance the code by introducing an Empty LLVM module(with the correct target) when it is available(so that Q2 can be resolved) and fallback to the CSourceModule. cc @FrozenGene who created the mechanism for the binary exporting 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
tqchen commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585462552 The reason that LLVMModule simplifies compilation is because it remembers the correct target triple. We can try to enhance the code by introducing an Empty LLVM module(with the correct target) when it is available(so that Q2 can be resolved) and fallback to the CSourceModule. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] kumasento commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
kumasento commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585454809 @mbarrett97 Thank you for your detailed explanation! I kind of understand what's going here. For Q1, I have updated the logic involving external module importing. It now always import external modules into `ret_.mod` since it is always there. For Q2, I'm afraid that I don't have a better idea for that. Maybe @tqchen @zhiics @FrozenGene can have an input on this? Thanks! 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser
alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585414848 > Do you think you can fix the padding issue? I haven't looked into what is going on, we may need to update cuda topi schedule. > > It is ok if you leave that to me, I can take a look in the follow up PR. I can try taking a look today but I don't have too much experience here and wouldn't want to block this going out if everything else seems okay. https://github.com/zxy844288792/tvm/blob/00fd0c45ac6a6c880c1f2c21eb79ca86b4778ba2/python/tvm/relay/op/nn/util.py#L43, not sure if obvious but the error is from converting 2d padding to 4d padding here (ie: just returning padding here fixes the cuda error). 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser
alexwong edited a comment on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585414848 > Do you think you can fix the padding issue? I haven't looked into what is going on, we may need to update cuda topi schedule. > > It is ok if you leave that to me, I can take a look in the follow up PR. I can try taking a look today but I don't have too much experience here and wouldn't want to block this going out if everything else seems okay. https://github.com/zxy844288792/tvm/blob/00fd0c45ac6a6c880c1f2c21eb79ca86b4778ba2/python/tvm/relay/op/nn/util.py#L43, not sure if obvious but the error is from converting 2d padding to 4d padding here. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] mbarrett97 commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
mbarrett97 commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585449674 I should probably explain my 'test case' :) Apologies if you know this all already. I'm using the external codegen feature which is relatively new. Essentially it allows you to take parts of the Relay graph and run them through a non-TVM compiler. In some cases, we take the entire Relay graph and put it through one of these external compilers resulting in the case where there are no lowered TVM functions. What does get produced though is 'external' runtime modules which correspond to the outputs from these external compilers. You can see this happening in the BuildRelay function. We call 'GetExternalModules' to get those modules that have been produced by external compilers. But to be able to compile these modules into a shared library, we need a 'host' TVM module. This host module then 'imports' in all the external modules, as you can see here: https://github.com/apache/incubator-tvm/blob/a5661611472c8e92b20bbe4d074333b8183f2878/src/relay/backend/build_module.cc#L458 However, if there is no host TVM module (which will happen when there are no lowered funcs), then you can't import the external modules into a host module and accordingly we fail to compile the result to a shared library. So this fix is very helpful by making sure that host module always gets created, even if it's empty, so we can always compile to a shared library. Now addressing your questions. For 1, yes I think we just remove this branch entirely and have it always import the external modules. For 2, I hit the assert when trying to call export_library on the 'lib' output of the build function. I believe SaveToFile is called when doing this export, in particular this line: https://github.com/apache/incubator-tvm/blob/79cfab00d1a5ae55b3324dae803d5e1389e12bbe/python/tvm/runtime/module.py#L302 I did try just commenting out the assert and it seems to work. However, I then ran into a new problem... Because what you produce is ostensibly a 'c source' file (even if it's empty), all the external modules are also exported as c files. This results in having to do a proper compile when you try and merge all the modules together to produce a shared library. If one of your external modules happens to be a binary artifact that's 150MB (my case), this compile uses an incredible amount of RAM and takes a long time to complete. I didn't get this problem with your first solution as it used an LLVMModule (I think?) which results in the external modules becoming object files so all you need to do is link, not perform a full compilation. Sorry for the wall of text, if you have any further questions I'll happily answer them. I realise I may not have perfectly explained everything here :) 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] anijain2305 commented on issue #4790: Fast exponent
anijain2305 commented on issue #4790: Fast exponent URL: https://github.com/apache/incubator-tvm/pull/4790#issuecomment-585443948 Can this get in? I will work on Relay changes. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] anijain2305 commented on a change in pull request #4866: Optimize x86 conv3d_ndhwc using data packing approach.
anijain2305 commented on a change in pull request #4866: Optimize x86
conv3d_ndhwc using data packing approach.
URL: https://github.com/apache/incubator-tvm/pull/4866#discussion_r378538482
##
File path: topi/python/topi/x86/conv3d.py
##
@@ -17,66 +17,359 @@
# pylint: disable=invalid-name, unused-variable, too-many-locals
# pylint: disable=unused-argument, redefined-builtin, no-else-return
"""Conv3D operators"""
+from collections import namedtuple
import tvm
-from .. import generic, tag
+from tvm import autotvm
+from tvm.autotvm.task.space import SplitEntity, OtherOptionEntity
+from .. import generic
from ..util import traverse_inline
+from ..nn.conv3d import conv3d, conv3d_ncdhw
+from ..nn.util import get_pad_tuple3d, infer_pad3d
+from ..nn.pad import pad
+from ..util import get_const_tuple, simplify, get_const_int
+from .util import get_fp32_len
-@generic.schedule_conv3d_ndhwc.register("cpu")
-def schedule_conv3d_ndhwc(outs):
-"""TOPI schedule callback for conv3d
+Workload3D = namedtuple('Workload',
+['in_dtype', 'out_dtype', 'depth', 'height', 'width',
+ 'in_filter', 'groups', 'out_filter', 'dkernel',
+ 'hkernel', 'wkernel', 'dpad', 'hpad', 'wpad',
+ 'dstride', 'hstride', 'wstride'])
+
+@autotvm.register_topi_compute(conv3d, 'cpu', ['direct'])
+def _declaration_conv3d(cfg, data, kernel, strides, padding, dilation,
+layout, out_dtype):
+"""3D convolution nchw forward operator.
Review comment:
nchw --> ncdhw, ndhwc
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] anijain2305 commented on a change in pull request #4866: Optimize x86 conv3d_ndhwc using data packing approach.
anijain2305 commented on a change in pull request #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866#discussion_r378488343 ## File path: topi/python/topi/nn/util.py ## @@ -47,6 +47,36 @@ def infer_pad(data, data_pad): wpad = (TW - IW) // 2 return get_const_int(hpad), get_const_int(wpad) +def infer_pad3d(data, data_pad): Review comment: This needs extra argument specifying input data layout. Currently it assumes NDHWC 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] kumasento commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
kumasento commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585438528 Hi @mbarrett97 Sorry I have got some questions for the case you've mentioned. 1. As you suggested, the assertion for `lowered_funcs.size()` in the branch assuming that external modules exist should be raised in the current implementation, but what is the expected behaviour when we actually use `CSourceModule` with an empty string? Should we simply remove this assertion since `ret_.mod` exists anyway and won't affect external module importing? 2. This line of code: https://github.com/apache/incubator-tvm/blob/a5661611472c8e92b20bbe4d074333b8183f2878/src/target/source/source_module.cc#L101 is located in a `SaveToFile` function, which is not called frequently, and the single use case I've noticed: https://github.com/apache/incubator-tvm/blob/fc1a1d83691f7229e25a37befdcd3ad9c3d430b3/tests/cpp/utvm_runtime_standalone_test.cc#L100 only exists in test and doesn't trigger that assertsion since `fmt` is `"o"`. Even if `fmt` is `"cc"`, assuming no such assertion and the string is empty, there should not be any exception caused (at the code level, I'm not sure whether functionally it is expected). So do you think we could simply remove that empty string checker? Thanks! 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #4872: [REFACTOR][PY][API-CHANGE] Establish tvm.target
tqchen commented on issue #4872: [REFACTOR][PY][API-CHANGE] Establish tvm.target URL: https://github.com/apache/incubator-tvm/pull/4872#issuecomment-585433502 cc @ZihengJiang @icemelon9 @yzhliu @merrymercy @gussmith23 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen opened a new pull request #4872: [REFACTOR][PY][API-CHANGE] Establish tvm.target
tqchen opened a new pull request #4872: [REFACTOR][PY][API-CHANGE] Establish tvm.target URL: https://github.com/apache/incubator-tvm/pull/4872 Move the related target modules into tvm.target. API change: - tvm.target.current_target -> tvm.target.Target.current - tvm.datatype -> tvm.target.datatype 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen merged pull request #4871: [JVM] Update the runtime PackedFunc for module
tqchen merged pull request #4871: [JVM] Update the runtime PackedFunc for module URL: https://github.com/apache/incubator-tvm/pull/4871 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-tvm] branch master updated (aaf62e4 -> 79cfab0)
This is an automated email from the ASF dual-hosted git repository. tqchen pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from aaf62e4 Fix optimize add 79cfab0 [JVM] Update the runtime PackedFunc for module No new revisions were added by this update. Summary of changes: jvm/core/src/main/java/org/apache/tvm/Module.java | 8 1 file changed, 4 insertions(+), 4 deletions(-)
[GitHub] [incubator-tvm] Laurawly commented on a change in pull request #4867: [TOPI][CUDA] Enable vectorization on fp16 type
Laurawly commented on a change in pull request #4867: [TOPI][CUDA] Enable
vectorization on fp16 type
URL: https://github.com/apache/incubator-tvm/pull/4867#discussion_r378516859
##
File path: topi/tests/python/test_topi_tensor.py
##
@@ -84,18 +84,41 @@ def check_device(device):
for device in ["llvm"]:
check_device(device)
+def verify_vectorization(n, m, dtype):
+def check_device(device):
+if not tvm.runtime.enabled(device):
+print("Skip because %s is not enabled" % device)
+return
+with tvm.target.create(device):
+ctx = tvm.context(device, 0)
+A = tvm.placeholder((n, m), name='A', dtype=dtype)
+B = tvm.compute((n, m), lambda i, j:
+ A[i, j] + tvm.const(1, A.dtype), name='B')
+S = topi.generic.schedule_elemwise(B)
+
+fun = tvm.build(S, [A, B], device)
+np_A = tvm.nd.empty((n, m), A.dtype, ctx).copyfrom(
+np.random.uniform(size=(n, m)))
+np_B = tvm.nd.empty((n, m), B.dtype, ctx)
+fun(np_A, np_B)
+tvm.testing.assert_allclose(np_B.asnumpy(), np_A.asnumpy() + 1,
rtol=1e-5)
+
+for device in ["cuda"]:
+check_device(device)
+
+def test_vectorization():
+verify_vectorization(128, 128, "float16")
Review comment:
Same here.
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] Laurawly commented on a change in pull request #4867: [TOPI][CUDA] Enable vectorization on fp16 type
Laurawly commented on a change in pull request #4867: [TOPI][CUDA] Enable vectorization on fp16 type URL: https://github.com/apache/incubator-tvm/pull/4867#discussion_r378516074 ## File path: topi/tests/python/test_topi_relu.py ## @@ -87,12 +87,12 @@ def _prelu_numpy(x, W): tvm.testing.assert_allclose(b.asnumpy(), out_np, rtol=1e-5) def test_relu(): -verify_relu(10, 128) +verify_relu(128, 128, "float32") Review comment: Can you keep a test case as before where `m` and `n` have different values? 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong commented on issue #4497: [Relay] Add a PyTorch to Relay Parser
alexwong commented on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585414848 > Do you think you can fix the padding issue? I haven't looked into what is going on, we may need to update cuda topi schedule. > > It is ok if you leave that to me, I can take a look in the follow up PR. I can try taking a look today but I don't have too much experience here and wouldn't want to block this going out if everything else seems okay. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] masahi commented on issue #4497: [Relay] Add a PyTorch to Relay Parser
masahi commented on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585410718 Do you think you can fix the padding issue? I haven't looked into what is going on, we may need to update cuda topi schedule. It is ok if you leave that to me, I can take a look in the follow up PR. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] vinx13 commented on a change in pull request #4867: [TOPI][CUDA] Enable vectorization on fp16 type
vinx13 commented on a change in pull request #4867: [TOPI][CUDA] Enable
vectorization on fp16 type
URL: https://github.com/apache/incubator-tvm/pull/4867#discussion_r378503237
##
File path: topi/tests/python/test_topi_tensor.py
##
@@ -84,18 +84,41 @@ def check_device(device):
for device in ["llvm"]:
check_device(device)
+def verify_vectorization(n, m, dtype):
+def check_device(device):
+if not tvm.runtime.enabled(device):
+print("Skip because %s is not enabled" % device)
+return
+with tvm.target.create(device):
Review comment:
add a check whether fp16 is supported here and also `verify_relu`
https://github.com/apache/incubator-tvm/blob/aaf62e47e64d592be770e915a7aa59d41eddb729/topi/tests/python/test_topi_transform.py#L57-L59
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #4867: [TOPI][CUDA] Enable vectorization on fp16 type
tqchen commented on issue #4867: [TOPI][CUDA] Enable vectorization on fp16 type URL: https://github.com/apache/incubator-tvm/pull/4867#issuecomment-585395086 @vinx13 @Laurawly please help to review the PR 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong commented on issue #4497: [Relay] Add a PyTorch to Relay Parser
alexwong commented on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585393595 > @alexwong Reverting the commit #4787 fixed the mobilenet issue for me. Yes, that does seem to be the issue. Commenting out the call to get_pad_tuple2d (https://github.com/apache/incubator-tvm/blob/master/python/tvm/relay/op/nn/nn.py#L206) and passing padding directly fixes it. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] paddyhoran commented on issue #2885: [SGX] Use Fortanix EDP
paddyhoran commented on issue #2885: [SGX] Use Fortanix EDP URL: https://github.com/apache/incubator-tvm/pull/2885#issuecomment-585389205 I just thought about this yesterday. I'm not sure I have time to contribute much but development on Rust can't continue without this getting merged as it's such a large PR. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen opened a new pull request #4871: [JVM] Update the runtime PackedFunc for module
tqchen opened a new pull request #4871: [JVM] Update the runtime PackedFunc for module URL: https://github.com/apache/incubator-tvm/pull/4871 for changes in https://github.com/apache/incubator-tvm/pull/4837/ cc @yzhliu @kparzysz-quic 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #4837: [REFACTOR][PY][API-Change] Polish tvm.runtime, tvm.runtime.module API update
tqchen commented on issue #4837: [REFACTOR][PY][API-Change] Polish tvm.runtime, tvm.runtime.module API update URL: https://github.com/apache/incubator-tvm/pull/4837#issuecomment-585373355 @kparzysz-quic Thanks for the catch, here is a patch https://github.com/apache/incubator-tvm/pull/4871 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] comaniac commented on issue #4870: [AutoTVM] Support range in index based tuners
comaniac commented on issue #4870: [AutoTVM] Support range in index based tuners URL: https://github.com/apache/incubator-tvm/pull/4870#issuecomment-585357287 @merrymercy could you help to review this PR? Thanks! 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong commented on issue #4497: [Relay] Add a PyTorch to Relay Parser
alexwong commented on issue #4497: [Relay] Add a PyTorch to Relay Parser URL: https://github.com/apache/incubator-tvm/pull/4497#issuecomment-585355953 > hmm it's weird. After I reboot my machine, alexnet and vgg test both passed on cuda. Do you have accuracy issues with alexnet and vgg locally? @alexwong > > UPDATE: It seems results of alexnet and vgg test in test_forward.py are randomly passing or failing. Using the same TVM install, my test script `torchvision_test.py` always passes. I compare the two IRs but the diff is only in input name. On my local machine which doesn't have CUDA, it passes but when using CUDA (through the CI container), it has accuracy problems on some inputs (haven't spent too much time looking at the issue). 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] wpan11nv commented on issue #4867: [TOPI][CUDA] Enable vectorization on fp16 type
wpan11nv commented on issue #4867: [TOPI][CUDA] Enable vectorization on fp16 type URL: https://github.com/apache/incubator-tvm/pull/4867#issuecomment-585355592 Kindly ping. Thanks! 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] alexwong commented on a change in pull request #4497: [Relay] Add a PyTorch to Relay Parser
alexwong commented on a change in pull request #4497: [Relay] Add a PyTorch to
Relay Parser
URL: https://github.com/apache/incubator-tvm/pull/4497#discussion_r378438728
##
File path: tests/python/frontend/pytorch/test_forward.py
##
@@ -0,0 +1,766 @@
+# Licensed to the Apache Software Foundation (ASF) under one
+# or more contributor license agreements. See the NOTICE file
+# distributed with this work for additional information
+# regarding copyright ownership. The ASF licenses this file
+# to you under the Apache License, Version 2.0 (the
+# "License"); you may not use this file except in compliance
+# with the License. You may obtain a copy of the License at
+#
+# http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing,
+# software distributed under the License is distributed on an
+# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+# KIND, either express or implied. See the License for the
+# specific language governing permissions and limitations
+# under the License.
+# pylint: disable=import-self, invalid-name, unused-argument
+"""Unit tests for various models and operators"""
+from time import time
+import os
+import sys
+from tempfile import TemporaryDirectory
+from scipy.stats import t as tdistr
+import numpy as np
+import torch
+from torch.nn import Module
+import tvm
+import torchvision
+
+from tvm import relay
+from tvm.contrib import graph_runtime
+from tvm.relay.testing.config import ctx_list
+
+sys.setrecursionlimit(1)
+
+def _vectorize(ten):
+return ten.reshape(-1)
+
+def atol(tru, est):
+def _atol_elt(tru, est):
+return abs(tru - est)
+tru = _vectorize(tru)
+est = _vectorize(est)
+return max([_atol_elt(x, y) for x, y in zip(tru, est)])
+
+def rtol(tru, est):
+def _rtol_elt(tru, est):
+return abs(tru - est) / min(abs(tru), abs(est))
+tru = _vectorize(tru)
+est = _vectorize(est)
+return max([_rtol_elt(x, y) for x, y in zip(tru, est)])
+
+def assert_shapes_match(tru, est):
+if tru.shape != est.shape:
+msg = "Output shapes {} and {} don't match"
+raise AssertionError(msg.format(tru.shape, est.shape))
+
+def load_torchvision(model_name):
+"""Given a model name, returns a Torchvision model in eval mode as well
+as an example input."""
+with torch.no_grad():
+if model_name.startswith("inception"):
+height = width = 299
+mean = [0.5, 0.5, 0.5]
+std = [0.5, 0.5, 0.5]
+else:
+height = width = 224
+mean = [0.485, 0.456, 0.406]
+std = [0.229, 0.224, 0.225]
+input_shape = [1, 3, height, width]
+input_data = torch.randn(input_shape).float()
+for channel in range(3):
+input_data[:, channel] -= mean[channel]
+input_data[:, channel] /= std[channel]
+model = getattr(torchvision.models, model_name)(pretrained=True)
+model = model.float().eval()
+return model, input_data
+
+def load_pretrainedmodels(model_name):
+"""Given a model name, returns a pretrainedmodels.pytorch model in eval
+mode as well as an example input."""
+import pretrainedmodels #
https://github.com/Cadene/pretrained-models.pytorch
+model = getattr(pretrainedmodels, model_name)().float().eval()
+input_shape = [1, *model.input_size]
+input_data = torch.rand(input_shape).float() * 256
+for channel in range(3):
+input_data[:, channel] -= model.mean[channel]
+input_data[:, channel] /= model.std[channel]
+return model, input_data
+
+def load_model(model_name):
+"""Given a model name, returns a model as well as an example input."""
+if hasattr(torchvision.models, model_name):
+return load_torchvision(model_name)
+try:
+if hasattr(pretrainedmodels, model_name):
+return load_pretrainedmodels(model_name)
+except ModuleNotFoundError:
+raise ModuleNotFoundError("Please install pretrainedmodels.pytorch")
+raise RuntimeError("Model not supported")
+
+
+def confidence_interval(mean, stdev, count, alpha=.01):
+"""Returns the lower and upper bounds of the confidence interval of a
random
+variable. Confidence is 1 - alpha (default confidence is 99%)."""
+stdval = tdistr.ppf(1 - alpha / 2, count - 1)
+lower, upper = mean + np.array([-1, 1]) * stdval * stdev / np.sqrt(count)
+return lower, upper
+
+def measure_latency(model, input_shapes, output_shapes, thresh, dryruns=40):
+"""Compute the latency of the given model"""
+latencies = []
+count = 0
+while True:
+if isinstance(model, torch.nn.Module):
+input_data = [torch.rand(shape).float() for shape in input_shapes]
+if torch.cuda.is_available():
+input_data = list(map(lambda x: x.cuda(), input_data))
+model = model.cuda()
+t_start = time()
+with torch.no_
[GitHub] [incubator-tvm] comaniac opened a new pull request #4870: [AutoTVM] Support range in index based tuners
comaniac opened a new pull request #4870: [AutoTVM] Support range in index based tuners URL: https://github.com/apache/incubator-tvm/pull/4870 This PR includes the following changes in order to let grid search and random tuner accept an index range. This is beneficial for distributed tuning since we can easily separate a tuning space to chunks and distribute them to different machines. **Note**: This change will not affect any existing use cases of grid search tuner and random tuner. - Support `range_idx` optional argument when constructing `GridSearchTuner` and `RandomTuner` to let them only tune a part of the tuning space. - Refactor `GridSearchTuner` and `RandomTuner` to have a common base class called `IndexBaseTuner`. `IndexBaseTuner` processs `range_idx` and unifies common fields such as `counter`. - Improve `RandomTuner` to generate non-repetitive random indices in constant time without increasing memory usage. - Unit tests. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] kparzysz-quic commented on issue #4837: [REFACTOR][PY][API-Change] Polish tvm.runtime, tvm.runtime.module API update
kparzysz-quic commented on issue #4837: [REFACTOR][PY][API-Change] Polish tvm.runtime, tvm.runtime.module API update URL: https://github.com/apache/incubator-tvm/pull/4837#issuecomment-585348595 The file `jvm/core/src/main/java/org/apache/tvm/Module.java` still uses old names. This broke the Java RPC app. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] kevinthesun commented on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach.
kevinthesun commented on issue #4866: Optimize x86 conv3d_ndhwc using data packing approach. URL: https://github.com/apache/incubator-tvm/pull/4866#issuecomment-585332792 Thank you for this work! It would be great if you can provide benchmarking data comparing tvm conv3d performance VS existing solution(tensorflow + mkldnn?) to see where current implementation stands. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #2885: [SGX] Use Fortanix EDP
tqchen commented on issue #2885: [SGX] Use Fortanix EDP URL: https://github.com/apache/incubator-tvm/pull/2885#issuecomment-585305001 ping @nhynes 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #4845: [DEV] TVM v0.7 Roadmap
tqchen commented on issue #4845: [DEV] TVM v0.7 Roadmap URL: https://github.com/apache/incubator-tvm/issues/4845#issuecomment-585290701 @yangjunpro Perhaps it worth to start a new thread in the discuss forum to discuss MLIR related topics. We certainly would love some proposals about interpolation between MLIR and TVM. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] yangjunpro commented on issue #4845: [DEV] TVM v0.7 Roadmap
yangjunpro commented on issue #4845: [DEV] TVM v0.7 Roadmap URL: https://github.com/apache/incubator-tvm/issues/4845#issuecomment-585288898 Is there any plan to integrate TVM as a dialect into MLIR? So other components based on MLIR can leverage the capability of TVM, such as high performance codegen, and fusion, etc. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] tqchen commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess
tqchen commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess URL: https://github.com/apache/incubator-tvm/pull/4543#issuecomment-585287746 @FrozenGene please https://docs.tvm.ai/contribute/code_review.html#approve-and-request-changes-explicitly 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] zhiics commented on a change in pull request #4868: [doc][VM] Update the vm doc
zhiics commented on a change in pull request #4868: [doc][VM] Update the vm doc URL: https://github.com/apache/incubator-tvm/pull/4868#discussion_r378356522 ## File path: docs/dev/virtual_machine.rst ## @@ -284,40 +297,76 @@ Dispatch Loop ~ A critical piece of a VM is the dispatch loop. The dispatch loop usually dominates the execution time of a virtual machine, but we have experimentally found this not to be the case for Relay. We have just implemented -a simple `switch`/`goto` dispatch loop which dispatches based on instruction op code. +a simple ``switch``/``goto`` dispatch loop which dispatches based on instruction op code. -This loop is implemented by `VirtualMachine::Run()`. +This loop is implemented by ``VirtualMachine::Run()``. VM Compiler ~~~ An important part of this infrastructure is a compiler from Relay's full IR into a sequence of bytecode. -The VM compiler transforms a `tvm::relay::Module` into a `tvm::relay::vm::VirtualMachine`. The virtual -machine contains a set of compiled functions, the compiled functions are contained in `tvm::relay::vm::Function`. The functions contain metadata about the the function as well as its compiled bytecode. For full definitions of the data structures see `vm.h`. +The VM compiler transforms a ``tvm::relay::Module`` into a ``tvm::relay::vm::Executable``. The executable +contains a set of compiled functions, the compiled functions are contained in ``tvm::relay::vm::Function``. The functions contain metadata about the the function as well as its compiled bytecode. The emitted executable object then can be loaded and run by a ``tvm::relay::vm::VirtualMachine`` object. For full definitions of the data structures, please see `include/tvm/runtime/vm.h`_. Optimizations ~ -There are quite a few optimizations required by the VM compiler. - -We have implemented them in the old pass style, but plan to port them to -the new pass manager (#2546) before merging. +There are quite a few optimizations required by the VM compiler. Each of them +is implemented as a pass which is managed by the Relay pass manager. Optimizations marked with `TODO` are not implemented yet. - A-Normal Form -- Lambda Lift (see `src/relay/vm/lambda_lift.cc`) -- Inline Primitives (see `src/relay/vm/inline_primitives.cc`) -- Inliner (see `src/relay/pass/inliner.cc`) -- Constant Pool Layout (see `src/relay/backend/vm/compiler.cc`) -- ADT Tag Allocation (see `src/relay/backend/vm/compiler.cc`) +- Lambda Lift (see `src/relay/vm/lambda_lift.cc`_) +- Inline Primitives (see `src/relay/vm/inline_primitives.cc`_) +- Constant Pool Layout (see `src/relay/backend/vm/compiler.cc`_) +- ADT Tag Allocation (see `src/relay/backend/vm/compiler.cc`_) Review comment: Yeah, it was from the old version. We need to remove it. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[incubator-tvm] branch master updated (a566161 -> aaf62e4)
This is an automated email from the ASF dual-hosted git repository.
tqchen pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git.
from a566161 [REFACTOR][PY][API-CHANGE] establish tvm.ir, migrate
corresponding files (#4862)
add 176ffe5 [DOCS][PY] Sphinx docs about tvm.ir
add aaf62e4 Fix optimize
No new revisions were added by this update.
Summary of changes:
docs/api/python/index.rst | 9 +++--
docs/api/python/{container.rst => ir.rst} | 9 +
docs/api/python/runtime.rst | 2 +-
docs/conf.py | 3 +++
python/tvm/ir/__init__.py | 4 ++--
python/tvm/ir/module.py | 6 +++---
python/tvm/relay/build_module.py | 2 +-
7 files changed, 18 insertions(+), 17 deletions(-)
rename docs/api/python/{container.rst => ir.rst} (91%)
[GitHub] [incubator-tvm] tqchen merged pull request #4869: [DOCS][PY] Add docs for tvm.ir
tqchen merged pull request #4869: [DOCS][PY] Add docs for tvm.ir URL: https://github.com/apache/incubator-tvm/pull/4869 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] kumasento commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod
kumasento commented on issue #4847: Return empty CSourceModule when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585243629 @mbarrett97 Thanks, I just noticed that the base of my PR is not the latest commit. I will update it soon. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] u99127 commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess
u99127 commented on issue #4543: [FRONTEND][TFLITE] Add support for TFLite_Detection_PostProcess URL: https://github.com/apache/incubator-tvm/pull/4543#issuecomment-585199370 All changes have been done ? Anything left to merge this in ? 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] kumasento commented on issue #4847: Use dummy func when no lowered_funcs exists in Relay mod
kumasento commented on issue #4847: Use dummy func when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585197212 > I don't think the empty CSourceModule method works. There's a check in source_module.cc that fails when you try and create one with an empty string. Hi @mbarrett97 thanks for your comment. Currently I haven't met such an issue while testing. Would you mind letting me know which assertion you were referring to? 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] mbarrett97 commented on issue #4847: Use dummy func when no lowered_funcs exists in Relay mod
mbarrett97 commented on issue #4847: Use dummy func when no lowered_funcs exists in Relay mod URL: https://github.com/apache/incubator-tvm/pull/4847#issuecomment-585180105 I don't think the empty CSourceModule method works. There's a check in source_module.cc that fails when you try and create one with an empty string. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] masahi edited a comment on issue #4816: [TFLite] Using real image for QNN testing.
masahi edited a comment on issue #4816: [TFLite] Using real image for QNN testing. URL: https://github.com/apache/incubator-tvm/pull/4816#issuecomment-585153972 What happen if the zero point is a vector, as in per channel quantization? What should the pad value be? 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-tvm] masahi commented on issue #4816: [TFLite] Using real image for QNN testing.
masahi commented on issue #4816: [TFLite] Using real image for QNN testing. URL: https://github.com/apache/incubator-tvm/pull/4816#issuecomment-585153972 What happen if the zero point is a vector, as in per channel quantuzation? What should the pad value be? 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
