[GitHub] [incubator-tvm] jcf94 commented on a change in pull request #4234: Auto TensorCore CodeGen

jcf94 commented on a change in pull request #4234: Auto TensorCore CodeGen URL: https://github.com/apache/incubator-tvm/pull/4234#discussion_r343509337 ## File path: src/pass/tensor_core.cc ## @@ -0,0 +1,1247 @@ +/* + * Licensed to the Apache Software Foundation (ASF) under

[GitHub] [incubator-tvm] tqchen commented on issue #4268: [Codegen][cuda-fp16] fallback to fp32 simulation when cuda arch < sm53

tqchen commented on issue #4268: [Codegen][cuda-fp16] fallback to fp32 simulation when cuda arch < sm53 URL: https://github.com/apache/incubator-tvm/pull/4268#issuecomment-550732338 I am not sure if we really want to enable fp16 on devices that does not have native support. never-the-less.

[GitHub] [incubator-tvm] comaniac commented on issue #4188: [RFC][AutoTVM] Selective Tuning

comaniac commented on issue #4188: [RFC][AutoTVM] Selective Tuning URL: https://github.com/apache/incubator-tvm/issues/4188#issuecomment-550660673 The tuning time posted here is the total time of tuning each task. In AutoTVM, one task means one op. Since we don't have a tunable template for

[GitHub] [incubator-tvm] Edwardmark commented on issue #4188: [RFC][AutoTVM] Selective Tuning

Edwardmark commented on issue #4188: [RFC][AutoTVM] Selective Tuning URL: https://github.com/apache/incubator-tvm/issues/4188#issuecomment-550654917 @comaniac Hi, the time cost for detection networks, did you include the nms post-processing steps? And how you implemented nms?Using other fra

[GitHub] [incubator-tvm] adityaatluri commented on a change in pull request #4234: Auto TensorCore CodeGen

adityaatluri commented on a change in pull request #4234: Auto TensorCore CodeGen URL: https://github.com/apache/incubator-tvm/pull/4234#discussion_r343462536 ## File path: src/pass/tensor_core.cc ## @@ -0,0 +1,1247 @@ +/* + * Licensed to the Apache Software Foundation (AS

[GitHub] [incubator-tvm] anirudh2290 closed pull request #4269: Add helper function ElemwiseAttrHelper

anirudh2290 closed pull request #4269: Add helper function ElemwiseAttrHelper URL: https://github.com/apache/incubator-tvm/pull/4269 This is an automated message from the Apache Git Service. To respond to the message, please

[GitHub] [incubator-tvm] anirudh2290 commented on issue #4269: Add helper function ElemwiseAttrHelper

anirudh2290 commented on issue #4269: Add helper function ElemwiseAttrHelper URL: https://github.com/apache/incubator-tvm/pull/4269#issuecomment-550577933 Closing, didnt realize these functions are also in MXNet. This is an au

[GitHub] [incubator-tvm] anirudh2290 opened a new pull request #4269: Add helper function ElemwiseAttrHelper

anirudh2290 opened a new pull request #4269: Add helper function ElemwiseAttrHelper URL: https://github.com/apache/incubator-tvm/pull/4269 Sometimes, we need to call ElemwiseAttr without NodeAttrs for example when calling from legacy op infer type in MXNet. Adding a helper function for the

[GitHub] [incubator-tvm] yzhliu opened a new pull request #4268: [Codegen][cuda-fp16] fallback to fp32 simulation when cuda arch < sm53

yzhliu opened a new pull request #4268: [Codegen][cuda-fp16] fallback to fp32 simulation when cuda arch < sm53 URL: https://github.com/apache/incubator-tvm/pull/4268 This ensures fp16 can still be computed on old devices. The code is borrowed from mshadow. An immediate usage will

[GitHub] [incubator-tvm] vinx13 edited a comment on issue #4260: [TOPI] Fix bug in Winograd on CUDA

vinx13 edited a comment on issue #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#issuecomment-550574722 @tqchen I still don't have permission. I'm already in the Apache github org, is there anything I'm missing? -

[GitHub] [incubator-tvm] vinx13 commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA

vinx13 commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#issuecomment-550574722 @tqchen I'm already in the Apache github org, is there anything I'm missing? Thi

[GitHub] [incubator-tvm] soiferj opened a new pull request #4267: [Relay][Frontend][ONNX] Add support for broadcasting to Where and MatMul

soiferj opened a new pull request #4267: [Relay][Frontend][ONNX] Add support for broadcasting to Where and MatMul URL: https://github.com/apache/incubator-tvm/pull/4267 The x and y inputs of `Where` can be broadcast, as can the `b` matrix of `MatMul`. @jwfromm would you mind taking

[incubator-tvm] branch master updated (7211c27 -> 14a5a35)

This is an automated email from the ASF dual-hosted git repository. haichen pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from 7211c27 [TOPI] Fix bug in Winograd on CUDA (#4260) add 14a5a35 [AutoTVM] Add batch_matmul to tunable o

[GitHub] [incubator-tvm] icemelon9 commented on issue #4242: [AutoTVM] Add batch_matmul to tunable operations

icemelon9 commented on issue #4242: [AutoTVM] Add batch_matmul to tunable operations URL: https://github.com/apache/incubator-tvm/pull/4242#issuecomment-550558551 Thanks @jwfromm This is an automated message from the Apache

[GitHub] [incubator-tvm] icemelon9 merged pull request #4242: [AutoTVM] Add batch_matmul to tunable operations

icemelon9 merged pull request #4242: [AutoTVM] Add batch_matmul to tunable operations URL: https://github.com/apache/incubator-tvm/pull/4242 This is an automated message from the Apache Git Service. To respond to the message

[GitHub] [incubator-tvm] icemelon9 commented on a change in pull request #4242: [AutoTVM] Add batch_matmul to tunable operations

icemelon9 commented on a change in pull request #4242: [AutoTVM] Add batch_matmul to tunable operations URL: https://github.com/apache/incubator-tvm/pull/4242#discussion_r343392582 ## File path: topi/python/topi/x86/batch_matmul.py ## @@ -18,43 +18,70 @@ """x86 batch_matm

[GitHub] [incubator-tvm] tweej commented on issue #1419: C++ Inference Failed

tweej commented on issue #1419: C++ Inference Failed URL: https://github.com/apache/incubator-tvm/issues/1419#issuecomment-550544973 What was the resolution to this problem? Moving to discuss.tvm.ai screws everyone who follows on similar issues.

[GitHub] [incubator-tvm] tqchen commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA

tqchen commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#issuecomment-550543187 @vinx13 you should have permission now, if now, please check if you have linked your github account. I will send you instructions --

[incubator-tvm] branch master updated (ddaa953 -> 7211c27)

This is an automated email from the ASF dual-hosted git repository. laurawly pushed a change to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git. from ddaa953 [Contrib] Fix error message at callback_get_section_size() (#4221) add 7211c27 [TOPI] Fix bu

[GitHub] [incubator-tvm] Laurawly merged pull request #4260: [TOPI] Fix bug in Winograd on CUDA

Laurawly merged pull request #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260 This is an automated message from the Apache Git Service. To respond to the message, please log on

[GitHub] [incubator-tvm] Laurawly commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA

Laurawly commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#issuecomment-550541004 Thanks @comaniac @vinx13, this is now merged. This is an automated message fro

[GitHub] [incubator-tvm] vinx13 commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA

vinx13 commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#issuecomment-550531324 @tqchen seems I don't have write permission This is an automated message from th

[GitHub] [incubator-tvm] comaniac commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA

comaniac commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#issuecomment-550530653 @vinx13 could you help merge it? Thanks. This is an automated message from the

[GitHub] [incubator-tvm] tqchen commented on issue #4221: [Contrib] Fix error message at callback_get_section_size()

tqchen commented on issue #4221: [Contrib] Fix error message at callback_get_section_size() URL: https://github.com/apache/incubator-tvm/pull/4221#issuecomment-550444628 Thanks @cchung100m This is an automated message from t

[GitHub] [incubator-tvm] tqchen edited a comment on issue #4221: [Contrib] Fix error message at callback_get_section_size()

tqchen edited a comment on issue #4221: [Contrib] Fix error message at callback_get_section_size() URL: https://github.com/apache/incubator-tvm/pull/4221#issuecomment-550444628 Thanks @cchung100m @weberlo This is an automate

[incubator-tvm] branch master updated (1eca1ad -> ddaa953)

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 1eca1ad [VTA] Hotfix for padded load test in Chisel VTA (#4264) add ddaa953 [Contrib] Fix error message

[GitHub] [incubator-tvm] weberlo commented on issue #4221: [Contrib] Fix error message at callback_get_section_size()

weberlo commented on issue #4221: [Contrib] Fix error message at callback_get_section_size() URL: https://github.com/apache/incubator-tvm/pull/4221#issuecomment-550444250 LGTM. Thanks for the fix! This is an automated message

[GitHub] [incubator-tvm] tqchen merged pull request #4221: [Contrib] Fix error message at callback_get_section_size()

tqchen merged pull request #4221: [Contrib] Fix error message at callback_get_section_size() URL: https://github.com/apache/incubator-tvm/pull/4221 This is an automated message from the Apache Git Service. To respond to the

[GitHub] [incubator-tvm] comaniac commented on a change in pull request #4260: [TOPI] Fix bug in Winograd on CUDA

comaniac commented on a change in pull request #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#discussion_r343260513 ## File path: topi/python/topi/cuda/conv2d_winograd.py ## @@ -42,25 +45,24 @@ def winograd_cuda(cfg, data,

[GitHub] [incubator-tvm] zhiics commented on issue #4258: [WIP][TVM] Bring Your Own Codegen to TVM

zhiics commented on issue #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#issuecomment-550433149 Sure, let's give it a try to send the runtime part first. This is an autom

[GitHub] [incubator-tvm] vinx13 commented on a change in pull request #4260: [TOPI] Fix bug in Winograd on CUDA

vinx13 commented on a change in pull request #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#discussion_r343239743 ## File path: topi/python/topi/cuda/conv2d_winograd.py ## @@ -42,25 +45,24 @@ def winograd_cuda(cfg, data, ke

[GitHub] [incubator-tvm] tqchen commented on issue #4258: [WIP][TVM] Bring Your Own Codegen to TVM

tqchen commented on issue #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#issuecomment-550418475 OK, please try to send another PR with a mini customized runtime that loads in something like a graph or steps of dnnl calls, and implement

[GitHub] [incubator-tvm] yzhliu commented on issue #4259: [DEV][DRAFT] TVM v0.6 Release candidate

yzhliu commented on issue #4259: [DEV][DRAFT] TVM v0.6 Release candidate URL: https://github.com/apache/incubator-tvm/issues/4259#issuecomment-550417287 We will mainly refer to [Apache MXNet release note](https://github.com/apache/incubator-mxnet/releases). I'll create a draft today and to

[incubator-tvm] branch master updated: [VTA] Hotfix for padded load test in Chisel VTA (#4264)

This is an automated email from the ASF dual-hosted git repository. moreau pushed a commit to branch master in repository https://gitbox.apache.org/repos/asf/incubator-tvm.git The following commit(s) were added to refs/heads/master by this push: new 1eca1ad [VTA] Hotfix for padded load tes

[GitHub] [incubator-tvm] tmoreau89 merged pull request #4264: [VTA] Hotfix for padded load test in Chisel VTA

tmoreau89 merged pull request #4264: [VTA] Hotfix for padded load test in Chisel VTA URL: https://github.com/apache/incubator-tvm/pull/4264 This is an automated message from the Apache Git Service. To respond to the message,

[GitHub] [incubator-tvm] yinghai commented on a change in pull request #4243: Fix broken loop partitioning due to recent changes.

yinghai commented on a change in pull request #4243: Fix broken loop partitioning due to recent changes. URL: https://github.com/apache/incubator-tvm/pull/4243#discussion_r343217657 ## File path: src/pass/loop_partition.cc ## @@ -513,17 +513,19 @@ Stmt LoopPartitioner::Try

[GitHub] [incubator-tvm] soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM

soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#discussion_r343206515 ## File path: src/runtime/contrib/extern_common.h ## @@ -0,0 +1,179 @@ +/* + * Licensed to the Apache

[GitHub] [incubator-tvm] soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM

soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#discussion_r343180538 ## File path: python/tvm/relay/op/op.py ## @@ -256,6 +257,25 @@ def register_shape_func(op_name, data_

[GitHub] [incubator-tvm] soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM

soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#discussion_r343179357 ## File path: python/tvm/relay/op/contrib/dnnl/extern_op.py ## @@ -0,0 +1,50 @@ +# Licensed to the Apa

[GitHub] [incubator-tvm] soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM

soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#discussion_r343196979 ## File path: src/relay/pass/extern_op.cc ## @@ -0,0 +1,102 @@ +/* + * Licensed to the Apache Software

[GitHub] [incubator-tvm] soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM

soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#discussion_r343173533 ## File path: include/tvm/relay/attrs/annotation.h ## @@ -57,6 +57,19 @@ struct CastHintAttrs : public

[GitHub] [incubator-tvm] soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM

soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#discussion_r343172649 ## File path: include/tvm/relay/attrs/annotation.h ## @@ -57,6 +57,19 @@ struct CastHintAttrs : public

[GitHub] [incubator-tvm] soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM

soiferj commented on a change in pull request #4258: [WIP][TVM] Bring Your Own Codegen to TVM URL: https://github.com/apache/incubator-tvm/pull/4258#discussion_r343169996 ## File path: cmake/config.cmake ## @@ -143,6 +143,12 @@ set(USE_ROCBLAS OFF) # Whether use contrib s

[GitHub] [incubator-tvm] OriAlpha closed issue #4266: An internal invariant was violated while typechecking your program

OriAlpha closed issue #4266: An internal invariant was violated while typechecking your program URL: https://github.com/apache/incubator-tvm/issues/4266 This is an automated message from the Apache Git Service. To respond t

[GitHub] [incubator-tvm] OriAlpha commented on issue #4248: About operators are not implemented

OriAlpha commented on issue #4248: About operators are not implemented URL: https://github.com/apache/incubator-tvm/issues/4248#issuecomment-550309081 Thanks for Replaying This is an automated message from the Apache Git Serv

[GitHub] [incubator-tvm] OriAlpha opened a new issue #4266: An internal invariant was violated while typechecking your program

OriAlpha opened a new issue #4266: An internal invariant was violated while typechecking your program URL: https://github.com/apache/incubator-tvm/issues/4266 I am trying to port a small lstm network to TVM version when I try to convert i see following errors. In main: v0.0.4

[GitHub] [incubator-tvm] cchung100m closed pull request #4235: [CMake] Imporve parameter 'ANTLR4' to avoid cmake will ignore the 'space'

cchung100m closed pull request #4235: [CMake] Imporve parameter 'ANTLR4' to avoid cmake will ignore the 'space' URL: https://github.com/apache/incubator-tvm/pull/4235 This is an automated message from the Apache Git Service.

[GitHub] [incubator-tvm] cbalint13 commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA

cbalint13 commented on issue #4260: [TOPI] Fix bug in Winograd on CUDA URL: https://github.com/apache/incubator-tvm/pull/4260#issuecomment-550260065 @comaniac , Looks good to me (looked at winograd part). This is an aut

[GitHub] [incubator-tvm] FinnWeng opened a new issue #4265: [RELAY][Bug] output type assignment not work for tf.range() in TVM

FinnWeng opened a new issue #4265: [RELAY][Bug] output type assignment not work for tf.range() in TVM URL: https://github.com/apache/incubator-tvm/issues/4265 This issue happens when converting code with `tf.range()`. My environment is: develop: python3.6, tensorflow1.14 conver

[GitHub] [incubator-tvm] liangfu opened a new pull request #4264: [VTA] Hotfix for padded load test in Chisel VTA

liangfu opened a new pull request #4264: [VTA] Hotfix for padded load test in Chisel VTA URL: https://github.com/apache/incubator-tvm/pull/4264 This PR brings a hotfix for evaluating `test_padded_load` in `test_vta_insn.py` with `tsim` backend. In previous implementation, evaluating `test

[GitHub] [incubator-tvm] FinnWeng opened a new issue #4263: Feature Request of conv2d_transpose op of Tensorflow

FinnWeng opened a new issue #4263: Feature Request of conv2d_transpose op of Tensorflow URL: https://github.com/apache/incubator-tvm/issues/4263 Hi, The operation of `tf.nn.conv2d_transpose` in tensorflow, which is related to `Conv2DBackpropinput`, is not in list of `_convert_map` i