Pytorch: Illegal instruction 4 in 1.3, OSX & GPU
🐛 Bug
Upon import torch, I get Illegal instruction 4.
This issue appears to have arisen sometime after 1.2 was released, but it was hard to nail down because of the issue resolved in #27583. For the same reason, I haven't been able to track it to a particular PR.
Compiling with XCode 9.4.1, CUDA 10, Cudnn 7.6.4, Python 3.6.
To Reproduce
Steps to reproduce the behavior:
- Apply patch per #27583
- Compile per instructions
import torch
Relevant portion of valrgind output:
==39767==
vex amd64->IR: unhandled instruction bytes: 0xC5 0xF8 0x57 0xC0 0xC5 0xF8 0x29 0x45 0xE0 0x48
vex amd64->IR: REX=0 REX.W=0 REX.R=0 REX.X=0 REX.B=0
vex amd64->IR: VEX=0 VEX.L=0 VEX.nVVVV=0x0 ESC=NONE
vex amd64->IR: PFX.66=0 PFX.F2=0 PFX.F3=0
==39767== valgrind: Unrecognised instruction at address 0x1157715a9.
==39767== at 0x1157715A9: _GLOBAL__sub_I_AVX2.cpp (in /Volumes/home500/anaconda/envs/pytorch1.3/lib/python3.6/site-packages/torch/lib/libtorch.dylib)
==39767== by 0x1003FDAC5: ImageLoaderMachO::doModInitFunctions(ImageLoader::LinkContext const&) (in /usr/lib/dyld)
==39767== by 0x1003FDCF5: ImageLoaderMachO::doInitialization(ImageLoader::LinkContext const&) (in /usr/lib/dyld)
==39767== by 0x1003F9217: ImageLoader::recursiveInitialization(ImageLoader::LinkContext const&, unsigned int, char const*, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F91AA: ImageLoader::recursiveInitialization(ImageLoader::LinkContext const&, unsigned int, char const*, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F91AA: ImageLoader::recursiveInitialization(ImageLoader::LinkContext const&, unsigned int, char const*, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F834D: ImageLoader::processInitializers(ImageLoader::LinkContext const&, unsigned int, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F83E1: ImageLoader::runInitializers(ImageLoader::LinkContext const&, ImageLoader::InitializerTimingList&) (in /usr/lib/dyld)
==39767== by 0x1003EC3E4: dyld::runInitializers(ImageLoader*) (in /usr/lib/dyld)
==39767== by 0x1003F50A7: dlopen (in /usr/lib/dyld)
==39767== by 0x100604D85: dlopen (in /usr/lib/system/libdyld.dylib)
==39767== by 0x1001CBBF2: _PyImport_FindSharedFuncptr (in /Volumes/home500/anaconda/envs/pytorch1.3/bin/python)
==39767== Your program just tried to execute an instruction that Valgrind
==39767== did not recognise. There are two possible reasons for this.
==39767== 1. Your program has a bug and erroneously jumped to a non-code
==39767== location. If you are running Memcheck and you just saw a
==39767== warning about a bad jump, it's probably your program's fault.
==39767== 2. The instruction is legitimate but Valgrind doesn't handle it,
==39767== i.e. it's Valgrind's fault. If you think this is the case or
==39767== you are not sure, please let us know and we'll try to fix it.
==39767== Either way, Valgrind will now raise a SIGILL signal which will
==39767== probably kill your program.
==39767==
==39767== Process terminating with default action of signal 4 (SIGILL)
==39767== Illegal opcode at address 0x1157715A9
==39767== at 0x1157715A9: _GLOBAL__sub_I_AVX2.cpp (in /Volumes/home500/anaconda/envs/pytorch1.3/lib/python3.6/site-packages/torch/lib/libtorch.dylib)
==39767== by 0x1003FDAC5: ImageLoaderMachO::doModInitFunctions(ImageLoader::LinkContext const&) (in /usr/lib/dyld)
==39767== by 0x1003FDCF5: ImageLoaderMachO::doInitialization(ImageLoader::LinkContext const&) (in /usr/lib/dyld)
==39767== by 0x1003F9217: ImageLoader::recursiveInitialization(ImageLoader::LinkContext const&, unsigned int, char const*, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F91AA: ImageLoader::recursiveInitialization(ImageLoader::LinkContext const&, unsigned int, char const*, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F91AA: ImageLoader::recursiveInitialization(ImageLoader::LinkContext const&, unsigned int, char const*, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F834D: ImageLoader::processInitializers(ImageLoader::LinkContext const&, unsigned int, ImageLoader::InitializerTimingList&, ImageLoader::UninitedUpwards&) (in /usr/lib/dyld)
==39767== by 0x1003F83E1: ImageLoader::runInitializers(ImageLoader::LinkContext const&, ImageLoader::InitializerTimingList&) (in /usr/lib/dyld)
==39767== by 0x1003EC3E4: dyld::runInitializers(ImageLoader*) (in /usr/lib/dyld)
==39767== by 0x1003F50A7: dlopen (in /usr/lib/dyld)
==39767== by 0x100604D85: dlopen (in /usr/lib/system/libdyld.dylib)
==39767== by 0x1001CBBF2: _PyImport_FindSharedFuncptr (in /Volumes/home500/anaconda/envs/pytorch1.3/bin/python)
==39767==
==39767== HEAP SUMMARY:
==39767== in use at exit: 25,400,951 bytes in 95,859 blocks
==39767== total heap usage: 139,863 allocs, 44,004 frees, 78,454,531 bytes allocated
==39767==
==39767== LEAK SUMMARY:
==39767== definitely lost: 32 bytes in 1 blocks
==39767== indirectly lost: 33 bytes in 2 blocks
==39767== possibly lost: 221,553 bytes in 318 blocks
==39767== still reachable: 25,124,123 bytes in 95,380 blocks
==39767== suppressed: 55,210 bytes in 158 blocks
==39767== Rerun with --leak-check=full to see details of leaked memory
==39767==
==39767== Use --track-origins=yes to see where uninitialised values come from
==39767== For lists of detected and suppressed errors, rerun with: -s
==39767== ERROR SUMMARY: 14653 errors from 274 contexts (suppressed: 4 from 4)
Illegal instruction: 4
Expected behavior
Not terminate because of illegal instruction.
Environment
PyTorch version: N/A
Is debug build: N/A
CUDA used to build PyTorch: N/A
OS: Mac OSX 10.13.6
GCC version: Could not collect
CMake version: version 3.15.4
Python version: 3.6
Is CUDA available: No
CUDA runtime version: 10.0.130
GPU models and configuration: Could not collect
Nvidia driver version: 1.1.0
cuDNN version: Probably one of the following:
/usr/local/cuda/lib/libcudnn.7.dylib
/usr/local/cuda/lib/libcudnn_static.a
Versions of relevant libraries:
[pip] 19.2.3
[conda] mkl 2019.4 233
[conda] mkl-service 2.3.0 py36hfbe908c_0
[conda] mkl_fft 1.0.14 py36h5e564d8_0
[conda] mkl_random 1.1.0 py36ha771720_0
Additional context
cc @ezyang @gchanan @zou3519 @VitalyFedyunin
elbamos
All 61 comments
Seems bad. We should repro and fix this.
zou3519
on 10 Oct 2019
unfortunately a fix for this is not making into v1.3.0.
We'll have to wait for v1.3.1 or nightlies
soumith
on 10 Oct 2019
Any way I can help reproduce or diagnose this?
elbamos
on 10 Oct 2019
Just want to double check: the error also occurs if you don't valgrind, is that right? (We had some issues with valgrind in the past where valgrind didn't understand an instruction but it was fine.)
@elbamos is it possible to do a bisect while cherry-picking the other commit?
ezyang
on 11 Oct 2019
Yes. I’m compiling with debug info and running valgrind to provide the issue with something to go on.
On Oct 11, 2019, at 10:16 AM, Edward Z. Yang notifications@github.com wrote:
Just want to double check: the error also occurs if you don't valgrind, is that right? (We had some issues with valgrind in the past where valgrind didn't understand an instruction but it was fine.)@elbamos is it possible to do a bisect while cherry-picking the other commit?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
elbamos
on 11 Oct 2019
Removing triaged as this wasn't triaged.
gchanan
on 14 Oct 2019
@ezyang I tried to bisect this. Its coming from the named tensor code; that's why it wasn't showing up until named tensors were permanently activated. Trying to go further back and compiling with named tensors enabled, its present in 44bd63c at the end of August.
Going back further than that, and we get to a place in the code where there are layers of other errors that prevent compilation or loading.
What else can I do to help you track this down?
elbamos
on 14 Oct 2019
I find it hard to believe that named tensor would cause this. In any case, at this point, I think one of us has to repro it.
ezyang
on 15 Oct 2019
What can I do to convince you?
On Oct 15, 2019, at 6:43 AM, Edward Z. Yang notifications@github.com wrote:
I find it hard to believe that named tensor would cause this. In any case, at this point, I think one of us has to repro it.—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
elbamos
on 15 Oct 2019
I can't reproduce this locally with the built binary (does that crash for you, @elbamos?). I can't reproduce this while building from source, but I don't have the same configuration (I'm using Apple LLVM version 10.0.1, Python 3.6.7, and no CUDA)
zou3519
on 15 Oct 2019
No, it doesn’t crash unless it’s been compiled with gpu support.
That isn’t surprising - from the valgrind output, the error occurs when the cuda libraries are being loaded. If I included more of the valgrind trace, you can see which of the libraries have loaded successfully. Would that be helpful to you?
On Oct 15, 2019, at 9:10 AM, Richard Zou notifications@github.com wrote:
I can't reproduce this locally with the built binary (does that crash for you, @elbamos?). I can't reproduce this while building from source, but I don't have the same configuration (I'm using Apple LLVM version 10.0.1, Python 3.6.7, and no CUDA)—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub, or unsubscribe.
elbamos
on 15 Oct 2019
Here's the full valgrind output...
errors.txt
elbamos
on 16 Oct 2019
I wish we had an OS X machine with GPU machines locally, this would make it a lot easier XD. (I guess there's no chance of this happening on Linux, is there...)
ezyang
on 16 Oct 2019
That isn’t surprising - from the valgrind output, the error occurs when the cuda libraries are being loaded.
I reviewed the valgrind output, and while I agree that there are CUDA related errors in the output, the actual illegal instruction is, IMO, unrelated. The backtrace for the illegal instruction is clearly related to our vectorization support.
Since we're having trouble reproducing, what I would like to see is if, when you do a build with CUDA disabled, the same error occurs. I think it will! If it doesn't that would be very interesting information.
ezyang
on 16 Oct 2019
You're correct - I do get the same error without cuda. (Valgrind output attached.) I also tried the distribution on conda, and get the same error from that.
errorsnocuda.txt
So, it has to be conflicting with something installed somewhere on my system. If you have any advice for tracking it down, I'd appreciate it.
Regarding an OSX machine with GPU, those are pretty cheap these days, because they're mostly 7+ years old. In fact, I have an old Mac Pro I'd contribute to the cause, but you'd have to put a GPU in it.
elbamos
on 17 Oct 2019
I looked a little further. I note this line from valgrind: ==96361== by 0x1153E93C8: _GLOBAL__sub_I_AVX2.cpp (AVX2.cpp:0) This CPU doesn't support AVX2 instructions. So it seems something in the build is forcing the use of AVX2 without checking for CPU support?
elbamos
on 17 Oct 2019
This CPU doesn't support AVX2 instructions. So it seems something in the build is forcing the use of AVX2 without checking for CPU support?
Bingo.
We do detection of cpu capabilities using the cpuinfo library. https://github.com/pytorch/cpuinfo You might be able to work out what the bug in the detection here is. If you need a cheap and cheerful workaround you could edit aten/src/ATen/native/DispatchStub.h to just never consider AVX2
ezyang
on 17 Oct 2019
which CPU are you using?
gchanan
on 17 Oct 2019
@gchanan Its a Xeon, model W3680.
@ezyang That makes sense. But what's the intersection with the namedtensor code? Because that part of the codebase is definitely where this arose. When I build a commit from when namedtensor was optional, the result will run if I build without namedtensor, it'll produce the error if I turn namedtensor on.
elbamos
on 17 Oct 2019
@elbamos this is unrelated to the named tensors support, perhaps just incidental connection.
most helpful for us would be checking what cpuinfo library tells you when you call cpuinfo_has_x86_avx as your processor clearly doesn't support AVX instructions per Intel specs.
Also could you please try installing torch 1.3 from pip and run simple code:
import torch
torch.randn(10) + torch.randn(10)
And tell us if it works or not.
Also if by any chance you are using any sort of virtualization, what might be reason of the issue you are seeing, as they prone to report AVX support mistakenly.
VitalyFedyunin
on 17 Oct 2019
@VitalyFedyunin cpuinfo (the isa-info executable) reports AVX: no and AVX2: no. So, that's correct. isainfo.txt
pytorch 1.3 installed from pip, like 1.3 installed from conda, crashes on the illegal instruction error after calling import torch.
I'm not running any virtualization (I don't think there's any OS X virtualization that supports CUDA).
Can you point me to the commit that first brought namedtensor support into master? Or the time period so I can hunt for it? (I spent some time looking, but couldn't find it.)
elbamos
on 17 Oct 2019
cc @zou3519 for this question
ezyang
on 18 Oct 2019
https://github.com/pytorch/pytorch/pull/26264 turns on named tensors for master.
https://github.com/pytorch/pytorch/pull/20162 is when I started working on it.
I'm the main contributor of code for named tensors, so you can look through all of the commits in my pull request history: https://github.com/pulls?q=is%3Apr+archived%3Afalse+is%3Aclosed+author%3Azou3519
zou3519
on 18 Oct 2019
The issue doesn't appear with #20162, but I'm not sure how to test if the namedtensor code is even being included in that revision of it. The build reports that named tensors will be enabled, but when I run, is_named etc don't appear to be members of the tensor class.
The issue definitely does appear with #26264, which I tested above.
I tried working through the commits in zou3519's pull history, but again hit the issue that when we get back into June, July, August, there are layers of additional build problems that are other impediments.
What can I do to help track this down?
elbamos
on 19 Oct 2019
Why don't you post your build problems here and we can try to help resolve them.
ezyang
on 21 Oct 2019
@ezyang That's going to be a rather lengthy process. Can you at least point me to commits that are useful starting places here? I don't want to keep spending time, and tying up my machine for hours, just to test a commit like #20162 that doesn't appear to include any potentially relevant code, where we then don't learn anything from testing after compiling. I'm happy to do it with some guidance so the time is used efficiently.
elbamos
on 24 Oct 2019
The issue doesn't appear with #20162, but I'm not sure how to test if the namedtensor code is even being included in that revision of it. The build reports that named tensors will be enabled, but when I run,
is_namedetc don't appear to be members of the tensor class
20162 adds a USE_NAMEDTENSOR build flag, but doesn't actually add any named tensor code. The named tensor code was added incrementally.
zou3519
on 25 Oct 2019
@zou3519 Yeah, that's my point. That's the commit that you had suggested I start with.
That is why I said, if I'm going to be bisecting this through a period when the code had a lot of build problems, and working through those problems with you guys, which is going to consume a lot of my time, then please at least point me to some commits that involve some relevant code.
elbamos
on 25 Oct 2019
@elbamos my apologies, I didn't read through the entire thread before replying to your above comment asking for which commit brought in named tensor code and instead gave you the beginning and end points.
I don't think any of the named tensor code interacts with AVX at all so it is difficult for me to give you code pointers.
However, here are some things that might be of interest:
- https://github.com/pytorch/pytorch/pull/25569: I think this PR is in the middle of my work with named tensors. It would be interesting to see if the problem is here or not so I know to go read commits before or after the midpoint in more detail.
- https://github.com/pytorch/pytorch/pull/23193 : Add name propagation support in TensorIterator. TensorIterator deals with AVX but I don't know what that interaction looks like. This support is only implemented for some TensorIterator functions.
When you're building code before #26264, you should build with BUILD_NAMEDTENSOR=1 (append that to before your python setup.py build command), but I think that is what you were doing already.
zou3519
on 25 Oct 2019
Alternately, if you can give us access to an OS X box with GPU we can try debugging it.
ezyang
on 25 Oct 2019
@ezyang I have an old pre-AVX mac pro sitting around I'd donate to the cause. It doesn't have a GPU in it - you wouldn't need one to debug this, but you could put an old one into it to help debug future OS X GPU build issues (they come up a lot).
@zou3519 Thanks, I'll try those today!
elbamos
on 25 Oct 2019
The problem does appear in #23193 (commit 505fa83, July 29). I also tested the immediately preceding commit, d3fcb4c, however, and the issue was present there as well.
Can you point me to the earliest PR or commit that contained testable namedtensor code? (Meaning that if torch loads, there is some sequence of calls I can execute to confirm the presence of compiled namedtensor code.)
elbamos
on 26 Oct 2019
First build problem to resolve:
With f51de8b (June 26), I get an error linking libtorch:
Undefined symbols for architecture x86_64:
"_cblas_gemm_s8u8s32_compute", referenced from:
mkldnn::impl::cpu::_ref_rnn_common_t<(mkldnn_prop_kind_t)64, (mkldnn_data_type_t)6, (mkldnn_data_type_t)5>::packed_gemm(char, char, int, int, int, float, signed char const*, int, unsigned char const*, int, float, int*, int) const in libmkldnn.a(ref_rnn.cpp.o)
"_cblas_gemm_s8u8s32_pack", referenced from:
mkldnn::impl::cpu::rnn_weights_reorder_t<(mkldnn_data_type_t)1, (mkldnn_data_type_t)5>::execute(mkldnn::impl::event_t*) const in libmkldnn.a(cpu_reorder.cpp.o)
"_cblas_gemm_s8u8s32_pack_get_size", referenced from:
mkldnn::impl::cpu::rnn_utils::init_conf(mkldnn::impl::cpu::rnn_utils::rnn_conf_t&, mkldnn_rnn_desc_t const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&) in libmkldnn.a(rnn_utils.cpp.o)
"_cblas_sgemm_pack_get_size", referenced from:
mkldnn::impl::cpu::rnn_utils::init_conf(mkldnn::impl::cpu::rnn_utils::rnn_conf_t&, mkldnn_rnn_desc_t const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&, mkldnn::impl::memory_desc_wrapper const&) in libmkldnn.a(rnn_utils.cpp.o)
ld: symbol(s) not found for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)
With a7ec889 (June 24), I get a host of compile errors coming from aten_op.h, for missing functions:
./caffe2/contrib/aten/aten_op.h:5908:37: error: no member named '_mkldnn_reshape' in namespace 'at'; did you mean 'mkldnn_reshape'?
auto the_result = at::_mkldnn_reshape(self, shape);
And non-matched function arguments:
./caffe2/contrib/aten/aten_op.h:11894:68: error: no viable conversion from 'const std::__1::vector<long long, std::__1::allocator<long long> >' to 'const at::Tensor'
auto the_result = at::conv_transpose3d(self, weight, kernel_size, bias, stride);
There are a lot of these, but I figure those excerpts should be sufficient to id what's going on.
elbamos
on 27 Oct 2019
For the mkldnn errors, build with USE_MKLDNN=0 (this is under the assumption that MKLDNN isn't the library which is causing the problem, which to be honest, it's pretty high on the suspicious list.) For the aten_op.h error, you need to do a clean before rebuilding (aten_op.h doesn't recompile correctly).
ezyang
on 28 Oct 2019
Binary search tracked it to commit 8ffcbfb, June 9, #21341. (This is with MKLDNN disabled, so its likely not coming from there after all.) I don't see anything in that commit that's suspicious, but that's the first one producing the error.
elbamos
on 29 Oct 2019
Yeah, I don't understand how this commit could have produced the error. Maybe it's the std::all_of call?
ezyang
on 29 Oct 2019
Comparing the binary produced by 8ffcbfb and the one produced by the preceding commit f9c4d0d, 8ffcbfb is 120k larger. This makes me think that what's going on, is that the commit is causing some larger set of code to be compiled that was not being compiled before.
elbamos
on 29 Oct 2019
What more can I do to be helpful?
elbamos
on 31 Oct 2019
Ping :)
elbamos
on 5 Nov 2019
I'm curious if you check out that commit and begin deleting parts of it to see what exactly would cause the error (crash on import torch) to go away. One thing to try is to delete the std::all_of call.
zou3519
on 5 Nov 2019
I tried deleting the call to std::all_of and modifying the function to always return false. The error still occurs.
elbamos
on 6 Nov 2019
From the valgrind output, the error occurs in _GLOBAL__sub_I_AVX2.cpp which initialises the global variables for TH/vector/AVX2.cpp. Since that file doesn't declare any globals itself, I suspect it's caused by a header that is being included due to 8ffcbfb7d45579c4761cd8a8aafcd82218efb4ab.
One candidate could be ATen/Dimname.h which is now included via the following include chain:
TH/vector/AVX2.cpp -> ATen/Context.h -> ATen/core/Tensor.h -> ATen/NamedTensor.h -> ATen/Dimname.h
In there we see one global variable being initialised:
https://github.com/pytorch/pytorch/blob/8ffcbfb7d45579c4761cd8a8aafcd82218efb4ab/aten/src/ATen/Dimname.h#L30
@elbamos could you try running from https://github.com/peterbell10/pytorch/commit/1a2b2de79c22a091c07147ecd6105192d94980de to see if it changes anything?
peterbell10
on 6 Nov 2019
@peterbell10 Yup, that did clear it out. Torch loads properly when I replace Dimname.h and .cpp with the versions in that commit.
elbamos
on 7 Nov 2019
Applying @peterbell10's approach to v1.3.0, I'm able to get a good and working install.
elbamos
on 7 Nov 2019
Very nice catch! Static variable in the header is definitely wrong. @peterbell10 are you going to PR your fix?
ezyang
on 7 Nov 2019
I can submit a PR but I'm not entirely convinced this is the whole story and not just a work-around. A static variable in a header is odd but I don't think it should be causing a SIGILL. My patch means that Symbol::dimname("*") isn't being called during import torch and so any potentially broken code in Symbol isn't being run.
To say for sure, @elbamos could try running the dimname tests to see if it fails on Dimname::wildcard(). The executable should be in build/bin/Dimname_test.
peterbell10
on 7 Nov 2019
@peterbell10 I ran build/bin/Dimname_test (running from v1.3.0+your patch code) and all tests pass.
I agree with you this can't be the whole story. In particular, I don't see why the inclusion of that symbol should cause compilation to use AVX instructions. For one thing, that static variable isn't doing anything that should implicate AVX. For another, the build should be detecting that the cpu doesn't support AVX, as it does for the remainder of the codebase. To me, that's the mystery, why the build is not correctly detecting AVX support for this part of the codebase.
elbamos
on 7 Nov 2019
Okay, I think I see what's going on now. AVX.cpp is intentionally compiled with AVX enabled to generate AVX kernels. Normally the code from this file is only run after runtime detection that your CPU supports AVX. However, the global variable constructor is also being inlined into that TU and is being optimised with AVX enabled.
peterbell10
on 7 Nov 2019
That sounds like the rest of the story! Please let me know if you want me to test another patch.
elbamos
on 7 Nov 2019
@elbamos would you mind testing #29384?
peterbell10
on 7 Nov 2019
@peterbell10 Running from your branch, I get the onnx-protobuf errors reported in #26945. Applying the files in #29384 to v1.3.0, I get a clean compile and a working binary.
elbamos
on 7 Nov 2019
@ezyang Do you still have an interest in the Mac box? My wife wants it out of our kitchen anyway, so all you have to do is send an uber to pick it up.
elbamos
on 11 Nov 2019
haha, I'll defer that to @soumith XD
ezyang
on 11 Nov 2019
haha, we should probably pass, considering how many unused boxes are under the common desk
soumith
on 12 Nov 2019
I can't get it working by simply installing https://github.com/pytorch/pytorch/issues/29967. I don't need GPU support but just a way to install it. I tried with every version, even 0.4.1 but I always get the same error. Do I have to build it?
absentio
on 22 Mar 2020
Dears, I still have the problem with pytorch-1.4.0:
Python 3.8.1 (default, Jan 8 2020, 16:15:59)
[Clang 4.0.1 (tags/RELEASE_401/final)] :: Anaconda, Inc. on darwin
Type "help", "copyright", "credits" or "license" for more information.
import torch
Illegal instruction: 4
Christux
on 31 Mar 2020
Dears, I still have the problem with pytorch-1.4.0:
Python 3.8.1 (default, Jan 8 2020, 16:15:59)
[Clang 4.0.1 (tags/RELEASE_401/final)] :: Anaconda, Inc. on darwin
Type "help", "copyright", "credits" or "license" for more information.import torch
Illegal instruction: 4
It works fine with conda-forge install. Pip or conda versions seem brocken.
Christux
on 1 Apr 2020
@Christux there is no conda-forge package for pytorch 1.4.0, so that doesn't quite make sense. Can you please provide reproducible instructions of how to trigger this, including your exact install commands?
rgommers
on 1 Apr 2020
step 1: classic install
(see next comment for details)
Christux
on 1 Apr 2020
Thanks @Christux, that helps. Email replies don't support Markdown on GitHub, so I'll copy and reformat your reply here:
step 1: classic install
$ conda config --show channels
channels:
- defaults
$ conda create --name test python=3.6 anaconda
Output:
Collecting package metadata (current_repodata.json): done
Solving environment: done
==> WARNING: A newer version of conda exists. <==
current version: 4.8.2
latest version: 4.8.3
Please update conda by running
$ conda update -n base -c defaults conda
## Package Plan ##
environment location: /opt/anaconda3/envs/test
added / updated specs:
- anaconda
- python=3.6
The following packages will be downloaded:
package | build
---------------------------|-----------------
applaunchservices-0.2.1 | py_0 9 KB
asn1crypto-1.3.0 | py36_0 163 KB
attrs-19.3.0 | py_0 39 KB
babel-2.8.0 | py_0 5.3 MB
backports-1.0 | py_2 139 KB
bkcharts-0.2 | py36_0 132 KB
cloudpickle-1.3.0 | py_0 30 KB
colorama-0.4.3 | py_0 20 KB
contextlib2-0.6.0.post1 | py_0 16 KB
dask-2.13.0 | py_0 14 KB
dask-core-2.13.0 | py_0 569 KB
decorator-4.4.2 | py_0 14 KB
defusedxml-0.6.0 | py_0 23 KB
diff-match-patch-20181111 | py_0 39 KB
flask-1.1.1 | py_1 73 KB
heapdict-1.0.1 | py_0 9 KB
idna-2.9 | py_1 49 KB
imageio-2.8.0 | py_0 3.0 MB
imagesize-1.2.0 | py_0 10 KB
intervaltree-3.0.2 | py_0 25 KB
ipywidgets-7.5.1 | py_0 107 KB
jdcal-1.4.1 | py_0 11 KB
jedi-0.15.2 | py36_0 738 KB
jinja2-2.11.1 | py_0 104 KB
joblib-0.14.1 | py_0 201 KB
jupyter_client-6.1.2 | py_0 82 KB
jupyterlab_server-1.1.0 | py_0 27 KB
libcxxabi-4.0.1 | hcfea43d_1 350 KB
more-itertools-8.2.0 | py_0 41 KB
nbconvert-5.6.1 | py36_0 459 KB
nbformat-5.0.4 | py_0 89 KB
notebook-6.0.3 | py36_0 4.0 MB
numpydoc-0.9.2 | py_0 31 KB
openpyxl-3.0.3 | py_0 162 KB
parso-0.5.2 | py_0 69 KB
partd-1.1.0 | py_0 20 KB
path-13.1.0 | py36_0 35 KB
path.py-12.4.0 | 0 7 KB
pathtools-0.1.2 | py_1 10 KB
prometheus_client-0.7.1 | py_0 42 KB
py-1.8.1 | py_0 71 KB
pycparser-2.20 | py_0 92 KB
pygments-2.6.1 | py_0 654 KB
pylint-2.4.4 | py36_0 422 KB
pyparsing-2.4.6 | py_0 64 KB
pytest-astropy-header-0.1.2| py_0 12 KB
pytest-openfiles-0.4.0 | py_0 10 KB
python-dateutil-2.8.1 | py_0 224 KB
python-jsonrpc-server-0.3.4| py_0 13 KB
pytz-2019.3 | py_0 231 KB
qtawesome-0.7.0 | py_0 726 KB
qtpy-1.9.0 | py_0 39 KB
rope-0.16.0 | py_0 126 KB
snowballstemmer-2.0.0 | py_0 58 KB
sphinx-2.4.4 | py_0 1.1 MB
sphinxcontrib-applehelp-1.0.2| py_0 27 KB
sphinxcontrib-devhelp-1.0.2| py_0 22 KB
sphinxcontrib-htmlhelp-1.0.3| py_0 27 KB
sphinxcontrib-jsmath-1.0.1 | py_0 8 KB
sphinxcontrib-qthelp-1.0.3 | py_0 25 KB
sphinxcontrib-serializinghtml-1.1.4| py_0 24 KB
tblib-1.6.0 | py_0 16 KB
testpath-0.4.4 | py_0 88 KB
toolz-0.10.0 | py_0 50 KB
werkzeug-1.0.0 | py_0 240 KB
widgetsnbextension-3.5.1 | py36_0 867 KB
xlsxwriter-1.2.8 | py_0 112 KB
zict-2.0.0 | py_0 13 KB
------------------------------------------------------------
Total: 21.4 MB
The following NEW packages will be INSTALLED:
_anaconda_depends pkgs/main/osx-64::_anaconda_depends-2019.03-py36_0
alabaster pkgs/main/osx-64::alabaster-0.7.12-py36_0
anaconda pkgs/main/osx-64::anaconda-custom-py36_1
anaconda-client pkgs/main/osx-64::anaconda-client-1.7.2-py36_0
anaconda-project pkgs/main/noarch::anaconda-project-0.8.4-py_0
applaunchservices pkgs/main/noarch::applaunchservices-0.2.1-py_0
appnope pkgs/main/osx-64::appnope-0.1.0-py36hf537a9a_0
appscript pkgs/main/osx-64::appscript-1.1.0-py36h1de35cc_0
argh pkgs/main/osx-64::argh-0.26.2-py36_0
asn1crypto pkgs/main/osx-64::asn1crypto-1.3.0-py36_0
astroid pkgs/main/osx-64::astroid-2.3.3-py36_0
astropy pkgs/main/osx-64::astropy-4.0-py36h1de35cc_0
atomicwrites pkgs/main/osx-64::atomicwrites-1.3.0-py36_1
attrs pkgs/main/noarch::attrs-19.3.0-py_0
autopep8 pkgs/main/noarch::autopep8-1.4.4-py_0
babel pkgs/main/noarch::babel-2.8.0-py_0
backcall pkgs/main/osx-64::backcall-0.1.0-py36_0
backports pkgs/main/noarch::backports-1.0-py_2
backports.os pkgs/main/osx-64::backports.os-0.1.1-py36_0
backports.shutil_~
pkgs/main/osx-64::backports.shutil_get_terminal_size-1.0.0-py36_2
beautifulsoup4 pkgs/main/osx-64::beautifulsoup4-4.8.2-py36_0
bitarray pkgs/main/osx-64::bitarray-1.2.1-py36h1de35cc_0
bkcharts pkgs/main/osx-64::bkcharts-0.2-py36_0
blas pkgs/main/osx-64::blas-1.0-mkl
bleach pkgs/main/osx-64::bleach-3.1.0-py36_0
blosc pkgs/main/osx-64::blosc-1.16.3-hd9629dc_0
bokeh pkgs/main/osx-64::bokeh-2.0.1-py36_0
boto pkgs/main/osx-64::boto-2.49.0-py36_0
bottleneck pkgs/main/osx-64::bottleneck-1.3.2-py36h776bbcc_0
bzip2 pkgs/main/osx-64::bzip2-1.0.8-h1de35cc_0
ca-certificates pkgs/main/osx-64::ca-certificates-2020.1.1-0
certifi pkgs/main/osx-64::certifi-2019.11.28-py36_1
cffi pkgs/main/osx-64::cffi-1.14.0-py36hb5b8e2f_0
chardet pkgs/main/osx-64::chardet-3.0.4-py36_1003
click pkgs/main/noarch::click-7.1.1-py_0
cloudpickle pkgs/main/noarch::cloudpickle-1.3.0-py_0
clyent pkgs/main/osx-64::clyent-1.2.2-py36_1
colorama pkgs/main/noarch::colorama-0.4.3-py_0
contextlib2 pkgs/main/noarch::contextlib2-0.6.0.post1-py_0
cryptography pkgs/main/osx-64::cryptography-2.8-py36ha12b0ac_0
curl pkgs/main/osx-64::curl-7.69.1-ha441bb4_0
cycler pkgs/main/osx-64::cycler-0.10.0-py36hfc81398_0
cython pkgs/main/osx-64::cython-0.29.15-py36h0a44026_0
cytoolz pkgs/main/osx-64::cytoolz-0.10.1-py36h1de35cc_0
dask pkgs/main/noarch::dask-2.13.0-py_0
dask-core pkgs/main/noarch::dask-core-2.13.0-py_0
dbus pkgs/main/osx-64::dbus-1.13.12-h90a0687_0
decorator pkgs/main/noarch::decorator-4.4.2-py_0
defusedxml pkgs/main/noarch::defusedxml-0.6.0-py_0
diff-match-patch pkgs/main/noarch::diff-match-patch-20181111-py_0
distributed pkgs/main/osx-64::distributed-2.13.0-py36_0
docutils pkgs/main/osx-64::docutils-0.16-py36_0
entrypoints pkgs/main/osx-64::entrypoints-0.3-py36_0
et_xmlfile pkgs/main/osx-64::et_xmlfile-1.0.1-py36h1315bdc_0
expat pkgs/main/osx-64::expat-2.2.6-h0a44026_0
fastcache pkgs/main/osx-64::fastcache-1.1.0-py36h1de35cc_0
flake8 pkgs/main/osx-64::flake8-3.7.9-py36_0
flask pkgs/main/noarch::flask-1.1.1-py_1
freetype pkgs/main/osx-64::freetype-2.9.1-hb4e5f40_0
fsspec pkgs/main/noarch::fsspec-0.6.3-py_0
future pkgs/main/osx-64::future-0.18.2-py36_0
get_terminal_size pkgs/main/osx-64::get_terminal_size-1.0.0-h7520d66_0
gettext pkgs/main/osx-64::gettext-0.19.8.1-h15daf44_3
gevent pkgs/main/osx-64::gevent-1.4.0-py36h1de35cc_0
glib pkgs/main/osx-64::glib-2.63.1-hd977a24_0
gmp pkgs/main/osx-64::gmp-6.1.2-hb37e062_1
gmpy2 pkgs/main/osx-64::gmpy2-2.0.8-py36h6ef4df4_2
greenlet pkgs/main/osx-64::greenlet-0.4.15-py36h1de35cc_0
h5py pkgs/main/osx-64::h5py-2.10.0-py36h3134771_0
hdf5 pkgs/main/osx-64::hdf5-1.10.4-hfa1e0ec_0
heapdict pkgs/main/noarch::heapdict-1.0.1-py_0
html5lib pkgs/main/osx-64::html5lib-1.0.1-py36_0
hypothesis pkgs/main/noarch::hypothesis-5.5.4-py_0
icu pkgs/main/osx-64::icu-58.2-h4b95b61_1
idna pkgs/main/noarch::idna-2.9-py_1
imageio pkgs/main/noarch::imageio-2.8.0-py_0
imagesize pkgs/main/noarch::imagesize-1.2.0-py_0
importlib_metadata pkgs/main/osx-64::importlib_metadata-1.5.0-py36_0
intel-openmp pkgs/main/osx-64::intel-openmp-2019.4-233
intervaltree pkgs/main/noarch::intervaltree-3.0.2-py_0
ipykernel pkgs/main/osx-64::ipykernel-5.1.4-py36h39e3cac_0
ipython pkgs/main/osx-64::ipython-7.13.0-py36h5ca1d4c_0
ipython_genutils pkgs/main/osx-64::ipython_genutils-0.2.0-py36_0
ipywidgets pkgs/main/noarch::ipywidgets-7.5.1-py_0
isort pkgs/main/osx-64::isort-4.3.21-py36_0
itsdangerous pkgs/main/osx-64::itsdangerous-1.1.0-py36_0
jbig pkgs/main/osx-64::jbig-2.1-h4d881f8_0
jdcal pkgs/main/noarch::jdcal-1.4.1-py_0
jedi pkgs/main/osx-64::jedi-0.15.2-py36_0
jinja2 pkgs/main/noarch::jinja2-2.11.1-py_0
joblib pkgs/main/noarch::joblib-0.14.1-py_0
jpeg pkgs/main/osx-64::jpeg-9b-he5867d9_2
json5 pkgs/main/noarch::json5-0.9.3-py_0
jsonschema pkgs/main/osx-64::jsonschema-3.2.0-py36_0
jupyter pkgs/main/osx-64::jupyter-1.0.0-py36_7
jupyter_client pkgs/main/noarch::jupyter_client-6.1.2-py_0
jupyter_console pkgs/main/noarch::jupyter_console-6.1.0-py_0
jupyter_core pkgs/main/osx-64::jupyter_core-4.6.3-py36_0
jupyterlab pkgs/main/noarch::jupyterlab-1.2.6-pyhf63ae98_0
jupyterlab_server pkgs/main/noarch::jupyterlab_server-1.1.0-py_0
keyring pkgs/main/osx-64::keyring-21.1.0-py36_0
kiwisolver pkgs/main/osx-64::kiwisolver-1.1.0-py36h0a44026_0
krb5 pkgs/main/osx-64::krb5-1.17.1-hddcf347_0
lazy-object-proxy
pkgs/main/osx-64::lazy-object-proxy-1.4.3-py36h1de35cc_0
libcurl pkgs/main/osx-64::libcurl-7.69.1-h051b688_0
libcxx pkgs/main/osx-64::libcxx-4.0.1-hcfea43d_1
libcxxabi pkgs/main/osx-64::libcxxabi-4.0.1-hcfea43d_1
libedit pkgs/main/osx-64::libedit-3.1.20181209-hb402a30_0
libffi pkgs/main/osx-64::libffi-3.2.1-h475c297_4
libgfortran pkgs/main/osx-64::libgfortran-3.0.1-h93005f0_2
libiconv pkgs/main/osx-64::libiconv-1.15-hdd342a3_7
libpng pkgs/main/osx-64::libpng-1.6.37-ha441bb4_0
libsodium pkgs/main/osx-64::libsodium-1.0.16-h3efe00b_0
libspatialindex pkgs/main/osx-64::libspatialindex-1.9.3-h0a44026_0
libssh2 pkgs/main/osx-64::libssh2-1.9.0-ha12b0ac_1
libtiff pkgs/main/osx-64::libtiff-4.1.0-hcb84e12_0
libxml2 pkgs/main/osx-64::libxml2-2.9.9-hf6e021a_1
libxslt pkgs/main/osx-64::libxslt-1.1.33-h33a18ac_0
llvm-openmp pkgs/main/osx-64::llvm-openmp-4.0.1-hcfea43d_1
llvmlite pkgs/main/osx-64::llvmlite-0.31.0-py36h1341992_0
locket pkgs/main/osx-64::locket-0.2.0-py36hca03003_1
lxml pkgs/main/osx-64::lxml-4.5.0-py36hef8c89e_0
lz4-c pkgs/main/osx-64::lz4-c-1.8.1.2-h1de35cc_0
lzo pkgs/main/osx-64::lzo-2.10-h362108e_2
markupsafe pkgs/main/osx-64::markupsafe-1.1.1-py36h1de35cc_0
matplotlib pkgs/main/osx-64::matplotlib-3.1.3-py36_0
matplotlib-base pkgs/main/osx-64::matplotlib-base-3.1.3-py36h9aa3819_0
mccabe pkgs/main/osx-64::mccabe-0.6.1-py36_1
mistune pkgs/main/osx-64::mistune-0.8.4-py36h1de35cc_0
mkl pkgs/main/osx-64::mkl-2019.4-233
mkl-service pkgs/main/osx-64::mkl-service-2.3.0-py36hfbe908c_0
mkl_fft pkgs/main/osx-64::mkl_fft-1.0.15-py36h5e564d8_0
mkl_random pkgs/main/osx-64::mkl_random-1.1.0-py36ha771720_0
mock pkgs/main/noarch::mock-4.0.1-py_0
more-itertools pkgs/main/noarch::more-itertools-8.2.0-py_0
mpc pkgs/main/osx-64::mpc-1.1.0-h6ef4df4_1
mpfr pkgs/main/osx-64::mpfr-4.0.1-h3018a27_3
mpmath pkgs/main/osx-64::mpmath-1.1.0-py36_0
msgpack-python pkgs/main/osx-64::msgpack-python-1.0.0-py36h04f5b5a_1
multipledispatch pkgs/main/osx-64::multipledispatch-0.6.0-py36_0
nbconvert pkgs/main/osx-64::nbconvert-5.6.1-py36_0
nbformat pkgs/main/noarch::nbformat-5.0.4-py_0
ncurses pkgs/main/osx-64::ncurses-6.2-h0a44026_0
networkx pkgs/main/noarch::networkx-2.4-py_0
nltk pkgs/main/osx-64::nltk-3.4.5-py36_0
nose pkgs/main/osx-64::nose-1.3.7-py36_2
notebook pkgs/main/osx-64::notebook-6.0.3-py36_0
numba pkgs/main/osx-64::numba-0.48.0-py36h6c726b0_0
numexpr pkgs/main/osx-64::numexpr-2.7.1-py36hce01a72_0
numpy pkgs/main/osx-64::numpy-1.18.1-py36h7241aed_0
numpy-base pkgs/main/osx-64::numpy-base-1.18.1-py36h6575580_1
numpydoc pkgs/main/noarch::numpydoc-0.9.2-py_0
olefile pkgs/main/osx-64::olefile-0.46-py36_0
openpyxl pkgs/main/noarch::openpyxl-3.0.3-py_0
openssl pkgs/main/osx-64::openssl-1.1.1f-h1de35cc_0
packaging pkgs/main/noarch::packaging-20.3-py_0
pandas pkgs/main/osx-64::pandas-1.0.3-py36h6c726b0_0
pandoc pkgs/main/osx-64::pandoc-2.2.3.2-0
pandocfilters pkgs/main/osx-64::pandocfilters-1.4.2-py36_1
parso pkgs/main/noarch::parso-0.5.2-py_0
partd pkgs/main/noarch::partd-1.1.0-py_0
path pkgs/main/osx-64::path-13.1.0-py36_0
path.py pkgs/main/noarch::path.py-12.4.0-0
pathlib2 pkgs/main/osx-64::pathlib2-2.3.5-py36_0
pathtools pkgs/main/noarch::pathtools-0.1.2-py_1
patsy pkgs/main/osx-64::patsy-0.5.1-py36_0
pcre pkgs/main/osx-64::pcre-8.43-h0a44026_0
pep8 pkgs/main/osx-64::pep8-1.7.1-py36_0
pexpect pkgs/main/osx-64::pexpect-4.8.0-py36_0
pickleshare pkgs/main/osx-64::pickleshare-0.7.5-py36_0
pillow pkgs/main/osx-64::pillow-7.0.0-py36h4655f20_0
pip pkgs/main/osx-64::pip-20.0.2-py36_1
pluggy pkgs/main/osx-64::pluggy-0.13.1-py36_0
ply pkgs/main/osx-64::ply-3.11-py36_0
prometheus_client pkgs/main/noarch::prometheus_client-0.7.1-py_0
prompt-toolkit pkgs/main/noarch::prompt-toolkit-3.0.4-py_0
prompt_toolkit pkgs/main/noarch::prompt_toolkit-3.0.4-0
psutil pkgs/main/osx-64::psutil-5.7.0-py36h1de35cc_0
ptyprocess pkgs/main/osx-64::ptyprocess-0.6.0-py36_0
py pkgs/main/noarch::py-1.8.1-py_0
pycodestyle pkgs/main/osx-64::pycodestyle-2.5.0-py36_0
pycosat pkgs/main/osx-64::pycosat-0.6.3-py36h1de35cc_0
pycparser pkgs/main/noarch::pycparser-2.20-py_0
pycrypto pkgs/main/osx-64::pycrypto-2.6.1-py36h1de35cc_9
pycurl pkgs/main/osx-64::pycurl-7.43.0.5-py36ha12b0ac_0
pydocstyle pkgs/main/noarch::pydocstyle-4.0.1-py_0
pyflakes pkgs/main/osx-64::pyflakes-2.1.1-py36_0
pygments pkgs/main/noarch::pygments-2.6.1-py_0
pylint pkgs/main/osx-64::pylint-2.4.4-py36_0
pyodbc pkgs/main/osx-64::pyodbc-4.0.30-py36h0a44026_0
pyopenssl pkgs/main/osx-64::pyopenssl-19.1.0-py36_0
pyparsing pkgs/main/noarch::pyparsing-2.4.6-py_0
pyqt pkgs/main/osx-64::pyqt-5.9.2-py36h655552a_2
pyrsistent pkgs/main/osx-64::pyrsistent-0.16.0-py36h1de35cc_0
pysocks pkgs/main/osx-64::pysocks-1.7.1-py36_0
pytables pkgs/main/osx-64::pytables-3.6.1-py36h5bccee9_0
pytest pkgs/main/osx-64::pytest-5.4.1-py36_0
pytest-arraydiff pkgs/main/osx-64::pytest-arraydiff-0.3-py36h39e3cac_0
pytest-astropy pkgs/main/noarch::pytest-astropy-0.8.0-py_0
pytest-astropy-he~ pkgs/main/noarch::pytest-astropy-header-0.1.2-py_0
pytest-doctestplus pkgs/main/noarch::pytest-doctestplus-0.5.0-py_0
pytest-openfiles pkgs/main/noarch::pytest-openfiles-0.4.0-py_0
pytest-remotedata pkgs/main/osx-64::pytest-remotedata-0.3.2-py36_0
python pkgs/main/osx-64::python-3.6.10-hc70fcce_1
python-dateutil pkgs/main/noarch::python-dateutil-2.8.1-py_0
python-jsonrpc-se~ pkgs/main/noarch::python-jsonrpc-server-0.3.4-py_0
python-language-s~ pkgs/main/osx-64::python-language-server-0.31.9-py36_0
python.app pkgs/main/osx-64::python.app-2-py36_10
pytz pkgs/main/noarch::pytz-2019.3-py_0
pywavelets pkgs/main/osx-64::pywavelets-1.1.1-py36h1de35cc_0
pyyaml pkgs/main/osx-64::pyyaml-5.3.1-py36h1de35cc_0
pyzmq pkgs/main/osx-64::pyzmq-18.1.1-py36h0a44026_0
qdarkstyle pkgs/main/noarch::qdarkstyle-2.8-py_0
qt pkgs/main/osx-64::qt-5.9.7-h468cd18_1
qtawesome pkgs/main/noarch::qtawesome-0.7.0-py_0
qtconsole pkgs/main/noarch::qtconsole-4.7.2-py_0
qtpy pkgs/main/noarch::qtpy-1.9.0-py_0
readline pkgs/main/osx-64::readline-8.0-h1de35cc_0
requests pkgs/main/osx-64::requests-2.23.0-py36_0
rope pkgs/main/noarch::rope-0.16.0-py_0
rtree pkgs/main/osx-64::rtree-0.9.3-py36_0
ruamel_yaml pkgs/main/osx-64::ruamel_yaml-0.15.87-py36h1de35cc_0
scikit-image pkgs/main/osx-64::scikit-image-0.16.2-py36h6c726b0_0
scikit-learn pkgs/main/osx-64::scikit-learn-0.22.1-py36h27c97d8_0
scipy pkgs/main/osx-64::scipy-1.4.1-py36h9fa6033_0
seaborn pkgs/main/noarch::seaborn-0.10.0-py_0
send2trash pkgs/main/osx-64::send2trash-1.5.0-py36_0
setuptools pkgs/main/osx-64::setuptools-46.1.3-py36_0
simplegeneric pkgs/main/osx-64::simplegeneric-0.8.1-py36_2
singledispatch pkgs/main/osx-64::singledispatch-3.4.0.3-py36hf20db9d_0
sip pkgs/main/osx-64::sip-4.19.8-py36h0a44026_0
six pkgs/main/osx-64::six-1.14.0-py36_0
snappy pkgs/main/osx-64::snappy-1.1.7-he62c110_3
snowballstemmer pkgs/main/noarch::snowballstemmer-2.0.0-py_0
sortedcollections pkgs/main/osx-64::sortedcollections-1.1.2-py36_0
sortedcontainers pkgs/main/osx-64::sortedcontainers-2.1.0-py36_0
soupsieve pkgs/main/noarch::soupsieve-2.0-py_0
sphinx pkgs/main/noarch::sphinx-2.4.4-py_0
sphinxcontrib pkgs/main/osx-64::sphinxcontrib-1.0-py36_1
sphinxcontrib-app~ pkgs/main/noarch::sphinxcontrib-applehelp-1.0.2-py_0
sphinxcontrib-dev~ pkgs/main/noarch::sphinxcontrib-devhelp-1.0.2-py_0
sphinxcontrib-htm~ pkgs/main/noarch::sphinxcontrib-htmlhelp-1.0.3-py_0
sphinxcontrib-jsm~ pkgs/main/noarch::sphinxcontrib-jsmath-1.0.1-py_0
sphinxcontrib-qth~ pkgs/main/noarch::sphinxcontrib-qthelp-1.0.3-py_0
sphinxcontrib-ser~
pkgs/main/noarch::sphinxcontrib-serializinghtml-1.1.4-py_0
sphinxcontrib-web~ pkgs/main/noarch::sphinxcontrib-websupport-1.2.1-py_0
spyder pkgs/main/osx-64::spyder-4.1.1-py36_1
spyder-kernels pkgs/main/osx-64::spyder-kernels-1.9.0-py36_0
sqlalchemy pkgs/main/osx-64::sqlalchemy-1.3.15-py36h1de35cc_1
sqlite pkgs/main/osx-64::sqlite-3.31.1-ha441bb4_0
statsmodels pkgs/main/osx-64::statsmodels-0.11.0-py36h1de35cc_0
sympy pkgs/main/osx-64::sympy-1.5.1-py36_0
tbb pkgs/main/osx-64::tbb-2020.0-h04f5b5a_0
tblib pkgs/main/noarch::tblib-1.6.0-py_0
terminado pkgs/main/osx-64::terminado-0.8.3-py36_0
testpath pkgs/main/noarch::testpath-0.4.4-py_0
tk pkgs/main/osx-64::tk-8.6.8-ha441bb4_0
toolz pkgs/main/noarch::toolz-0.10.0-py_0
tornado pkgs/main/osx-64::tornado-6.0.4-py36h1de35cc_1
traitlets pkgs/main/osx-64::traitlets-4.3.3-py36_0
typed-ast pkgs/main/osx-64::typed-ast-1.4.1-py36h1de35cc_0
typing_extensions pkgs/main/osx-64::typing_extensions-3.7.4.1-py36_0
ujson pkgs/main/osx-64::ujson-1.35-py36h1de35cc_0
unicodecsv pkgs/main/osx-64::unicodecsv-0.14.1-py36he531d66_0
unixodbc pkgs/main/osx-64::unixodbc-2.3.7-h1de35cc_0
urllib3 pkgs/main/osx-64::urllib3-1.25.8-py36_0
watchdog pkgs/main/osx-64::watchdog-0.10.2-py36h1de35cc_0
wcwidth pkgs/main/noarch::wcwidth-0.1.9-py_0
webencodings pkgs/main/osx-64::webencodings-0.5.1-py36_1
werkzeug pkgs/main/noarch::werkzeug-1.0.0-py_0
wheel pkgs/main/osx-64::wheel-0.34.2-py36_0
widgetsnbextension pkgs/main/osx-64::widgetsnbextension-3.5.1-py36_0
wrapt pkgs/main/osx-64::wrapt-1.12.1-py36h1de35cc_1
wurlitzer pkgs/main/osx-64::wurlitzer-2.0.0-py36_0
xlrd pkgs/main/osx-64::xlrd-1.2.0-py36_0
xlsxwriter pkgs/main/noarch::xlsxwriter-1.2.8-py_0
xlwings pkgs/main/osx-64::xlwings-0.18.0-py36_0
xlwt pkgs/main/osx-64::xlwt-1.2.0-py36h5ad1178_0
xz pkgs/main/osx-64::xz-5.2.4-h1de35cc_4
yaml pkgs/main/osx-64::yaml-0.1.7-hc338f04_2
yapf pkgs/main/noarch::yapf-0.28.0-py_0
zeromq pkgs/main/osx-64::zeromq-4.3.1-h0a44026_3
zict pkgs/main/noarch::zict-2.0.0-py_0
zipp pkgs/main/noarch::zipp-2.2.0-py_0
zlib pkgs/main/osx-64::zlib-1.2.11-h1de35cc_3
zstd pkgs/main/osx-64::zstd-1.3.7-h5bba6e5_0
Proceed ([y]/n)?
Downloading and Extracting Packages
sphinxcontrib-serial | 24 KB |
################################################################################################################################################################
| 100%
snowballstemmer-2.0. | 58 KB |
################################################################################################################################################################
| 100%
imagesize-1.2.0 | 10 KB |
################################################################################################################################################################
| 100%
nbformat-5.0.4 | 89 KB |
################################################################################################################################################################
| 100%
prometheus_client-0. | 42 KB |
################################################################################################################################################################
| 100%
openpyxl-3.0.3 | 162 KB |
################################################################################################################################################################
| 100%
contextlib2-0.6.0.po | 16 KB |
################################################################################################################################################################
| 100%
asn1crypto-1.3.0 | 163 KB |
################################################################################################################################################################
| 100%
dask-core-2.13.0 | 569 KB |
################################################################################################################################################################
| 100%
joblib-0.14.1 | 201 KB |
################################################################################################################################################################
| 100%
pycparser-2.20 | 92 KB |
################################################################################################################################################################
| 100%
qtawesome-0.7.0 | 726 KB |
################################################################################################################################################################
| 100%
pygments-2.6.1 | 654 KB |
################################################################################################################################################################
| 100%
numpydoc-0.9.2 | 31 KB |
################################################################################################################################################################
| 100%
diff-match-patch-201 | 39 KB |
################################################################################################################################################################
| 100%
pytest-astropy-heade | 12 KB |
################################################################################################################################################################
| 100%
ipywidgets-7.5.1 | 107 KB |
################################################################################################################################################################
| 100%
bkcharts-0.2 | 132 KB |
################################################################################################################################################################
| 100%
python-dateutil-2.8. | 224 KB |
################################################################################################################################################################
| 100%
pytz-2019.3 | 231 KB |
################################################################################################################################################################
| 100%
intervaltree-3.0.2 | 25 KB |
################################################################################################################################################################
| 100%
nbconvert-5.6.1 | 459 KB |
################################################################################################################################################################
| 100%
babel-2.8.0 | 5.3 MB |
################################################################################################################################################################
| 100%
notebook-6.0.3 | 4.0 MB |
################################################################################################################################################################
| 100%
toolz-0.10.0 | 50 KB |
################################################################################################################################################################
| 100%
zict-2.0.0 | 13 KB |
################################################################################################################################################################
| 100%
sphinx-2.4.4 | 1.1 MB |
################################################################################################################################################################
| 100%
qtpy-1.9.0 | 39 KB |
################################################################################################################################################################
| 100%
jedi-0.15.2 | 738 KB |
################################################################################################################################################################
| 100%
heapdict-1.0.1 | 9 KB |
################################################################################################################################################################
| 100%
xlsxwriter-1.2.8 | 112 KB |
################################################################################################################################################################
| 100%
attrs-19.3.0 | 39 KB |
################################################################################################################################################################
| 100%
defusedxml-0.6.0 | 23 KB |
################################################################################################################################################################
| 100%
libcxxabi-4.0.1 | 350 KB |
################################################################################################################################################################
| 100%
widgetsnbextension-3 | 867 KB |
################################################################################################################################################################
| 100%
cloudpickle-1.3.0 | 30 KB |
################################################################################################################################################################
| 100%
jinja2-2.11.1 | 104 KB |
################################################################################################################################################################
| 100%
sphinxcontrib-appleh | 27 KB |
################################################################################################################################################################
| 100%
werkzeug-1.0.0 | 240 KB |
################################################################################################################################################################
| 100%
dask-2.13.0 | 14 KB |
################################################################################################################################################################
| 100%
path.py-12.4.0 | 7 KB |
################################################################################################################################################################
| 100%
sphinxcontrib-jsmath | 8 KB |
################################################################################################################################################################
| 100%
colorama-0.4.3 | 20 KB |
################################################################################################################################################################
| 100%
sphinxcontrib-devhel | 22 KB |
################################################################################################################################################################
| 100%
sphinxcontrib-htmlhe | 27 KB |
################################################################################################################################################################
| 100%
backports-1.0 | 139 KB |
################################################################################################################################################################
| 100%
python-jsonrpc-serve | 13 KB |
################################################################################################################################################################
| 100%
sphinxcontrib-qthelp | 25 KB |
################################################################################################################################################################
| 100%
flask-1.1.1 | 73 KB |
################################################################################################################################################################
| 100%
rope-0.16.0 | 126 KB |
################################################################################################################################################################
| 100%
jdcal-1.4.1 | 11 KB |
################################################################################################################################################################
| 100%
decorator-4.4.2 | 14 KB |
################################################################################################################################################################
| 100%
pathtools-0.1.2 | 10 KB |
################################################################################################################################################################
| 100%
more-itertools-8.2.0 | 41 KB |
################################################################################################################################################################
| 100%
testpath-0.4.4 | 88 KB |
################################################################################################################################################################
| 100%
imageio-2.8.0 | 3.0 MB |
################################################################################################################################################################
| 100%
pytest-openfiles-0.4 | 10 KB |
################################################################################################################################################################
| 100%
path-13.1.0 | 35 KB |
################################################################################################################################################################
| 100%
tblib-1.6.0 | 16 KB |
################################################################################################################################################################
| 100%
parso-0.5.2 | 69 KB |
################################################################################################################################################################
| 100%
py-1.8.1 | 71 KB |
################################################################################################################################################################
| 100%
applaunchservices-0. | 9 KB |
################################################################################################################################################################
| 100%
idna-2.9 | 49 KB |
################################################################################################################################################################
| 100%
jupyterlab_server-1. | 27 KB |
################################################################################################################################################################
| 100%
pylint-2.4.4 | 422 KB |
################################################################################################################################################################
| 100%
jupyter_client-6.1.2 | 82 KB |
################################################################################################################################################################
| 100%
partd-1.1.0 | 20 KB |
################################################################################################################################################################
| 100%
pyparsing-2.4.6 | 64 KB |
################################################################################################################################################################
| 100%
Preparing transaction: done
$ conda activate test
$ conda install pytorch torchvision -c pytorch
$ python
Python 3.6.10 |Anaconda, Inc.| (default, Mar 25 2020, 18:53:43)
[GCC 4.2.1 Compatible Clang 4.0.1 (tags/RELEASE_401/final)] on darwin
Type "help", "copyright", "credits" or "license" for more information.
>>> import torch
Illegal instruction: 4
step 2: Migration to conda forge
$ conda config --add channels conda-forge
$ conda config --set channel_priority strict
$ conda update --all // Perhaps this is the point, it might update a
brocken dependency
Output:
Collecting package metadata (current_repodata.json): done
Solving environment: failed with repodata from current_repodata.json,
will retry with next repodata source.
Solving environment: failed with repodata from current_repodata.json,
will retry with next repodata source.
Collecting package metadata (repodata.json): done
Solving environment: done
==> WARNING: A newer version of conda exists. <==
current version: 4.8.2
latest version: 4.8.3
Please update conda by running
$ conda update -n base -c defaults conda
## Package Plan ##
environment location: /opt/anaconda3/envs/test
The following packages will be downloaded:
package | build
---------------------------|-----------------
_pytorch_select-0.1 | cpu_0 169 KB
applaunchservices-0.2.1 | py_0 8 KB
conda-forge
asn1crypto-1.3.0 | py36_0 159 KB
conda-forge
attrs-19.3.0 | py_0 35 KB
conda-forge
babel-2.8.0 | py_0 6.0 MB
conda-forge
backports-1.0 | py_2 4 KB
conda-forge
bkcharts-0.2 | py36_0 126 KB
conda-forge
blas-1.0 | mkl 1 KB
conda-forge
cloudpickle-1.3.0 | py_0 24 KB
conda-forge
colorama-0.4.3 | py_0 17 KB
conda-forge
contextlib2-0.6.0.post1 | py_0 12 KB
conda-forge
dask-2.13.0 | py_0 4 KB
conda-forge
dask-core-2.13.0 | py_0 596 KB
conda-forge
decorator-4.4.2 | py_0 11 KB
conda-forge
defusedxml-0.6.0 | py_0 22 KB
conda-forge
diff-match-patch-20181111 | py_0 37 KB
conda-forge
flask-1.1.1 | py_1 69 KB
conda-forge
heapdict-1.0.1 | py_0 7 KB
conda-forge
idna-2.9 | py_1 52 KB
conda-forge
imageio-2.8.0 | py_0 3.1 MB
conda-forge
imagesize-1.2.0 | py_0 8 KB
conda-forge
intervaltree-3.0.2 | py_0 23 KB
conda-forge
ipywidgets-7.5.1 | py_0 101 KB
conda-forge
jdcal-1.4.1 | py_0 9 KB
conda-forge
jedi-0.15.2 | py36_0 757 KB
conda-forge
jinja2-2.11.1 | py_0 94 KB
conda-forge
joblib-0.14.1 | py_0 198 KB
conda-forge
jupyter_client-6.1.2 | py_0 74 KB
conda-forge
jupyterlab_server-1.1.0 | py_0 24 KB
conda-forge
libblas-3.8.0 | 14_mkl 10 KB
conda-forge
libcblas-3.8.0 | 14_mkl 10 KB
conda-forge
libcxxabi-4.0.1 | hcfea43d_1 458 KB
conda-forge
liblapack-3.8.0 | 14_mkl 10 KB
conda-forge
more-itertools-8.2.0 | py_0 35 KB
conda-forge
nbconvert-5.6.1 | py36_0 466 KB
conda-forge
nbformat-5.0.4 | py_0 98 KB
conda-forge
notebook-6.0.3 | py36_0 6.2 MB
conda-forge
numpydoc-0.9.2 | py_0 29 KB
conda-forge
openpyxl-3.0.3 | py_0 152 KB
conda-forge
parso-0.5.2 | py_0 66 KB
conda-forge
partd-1.1.0 | py_0 17 KB
conda-forge
path-13.1.0 | py36_0 34 KB
conda-forge
path.py-12.4.0 | 0 4 KB
conda-forge
pathtools-0.1.2 | py_1 8 KB
conda-forge
prometheus_client-0.7.1 | py_0 38 KB
conda-forge
py-1.8.1 | py_0 66 KB
conda-forge
pycparser-2.20 | py_0 89 KB
conda-forge
pygments-2.6.1 | py_0 683 KB
conda-forge
pylint-2.4.4 | py36_0 425 KB
conda-forge
pyparsing-2.4.6 | py_0 59 KB
conda-forge
python-dateutil-2.8.1 | py_0 220 KB
conda-forge
python-jsonrpc-server-0.3.4| py_0 11 KB
conda-forge
pytorch-1.4.0 |cpu_py36hf9bb1df_0 26.3 MB
pytz-2019.3 | py_0 237 KB
conda-forge
qtawesome-0.7.0 | py_0 786 KB
conda-forge
qtpy-1.9.0 | py_0 34 KB
conda-forge
rope-0.16.0 | py_0 117 KB
conda-forge
snowballstemmer-2.0.0 | py_0 55 KB
conda-forge
sphinx-2.4.4 | py_0 1.4 MB
conda-forge
sphinxcontrib-applehelp-1.0.2| py_0 28 KB
conda-forge
sphinxcontrib-devhelp-1.0.2| py_0 22 KB
conda-forge
sphinxcontrib-htmlhelp-1.0.3| py_0 27 KB
conda-forge
sphinxcontrib-jsmath-1.0.1 | py_0 7 KB
conda-forge
sphinxcontrib-qthelp-1.0.3 | py_0 25 KB
conda-forge
sphinxcontrib-serializinghtml-1.1.4| py_0 24 KB
conda-forge
tblib-1.6.0 | py_0 14 KB
conda-forge
testpath-0.4.4 | py_0 85 KB
conda-forge
toolz-0.10.0 | py_0 46 KB
conda-forge
widgetsnbextension-3.5.1 | py36_0 1.8 MB
conda-forge
xlsxwriter-1.2.8 | py_0 103 KB
conda-forge
zict-2.0.0 | py_0 10 KB
conda-forge
------------------------------------------------------------
Total: 51.9 MB
The following NEW packages will be INSTALLED:
_pytorch_select pkgs/main/osx-64::_pytorch_select-0.1-cpu_0
blaze conda-forge/osx-64::blaze-0.11.3-py36_0
datashape conda-forge/noarch::datashape-0.5.4-py_1
filelock conda-forge/noarch::filelock-3.0.10-py_0
flask-cors conda-forge/noarch::flask-cors-3.0.8-py_0
glob2 conda-forge/noarch::glob2-0.7-py_0
importlib-metadata
conda-forge/osx-64::importlib-metadata-1.6.0-py36h9f0ad1d_0
jupyterlab_launch~ conda-forge/noarch::jupyterlab_launcher-0.13.1-py_2
libblas conda-forge/osx-64::libblas-3.8.0-14_mkl
libcblas conda-forge/osx-64::libcblas-3.8.0-14_mkl
libclang conda-forge/osx-64::libclang-9.0.1-default_hf57f61e_0
liblapack conda-forge/osx-64::liblapack-3.8.0-14_mkl
libllvm8 conda-forge/osx-64::libllvm8-8.0.1-h770b8ee_0
libllvm9 conda-forge/osx-64::libllvm9-9.0.1-ha1b3eb9_0
libpq conda-forge/osx-64::libpq-12.2-h554dc5a_0
libwebp-base conda-forge/osx-64::libwebp-base-1.1.0-h0b31af3_3
nspr conda-forge/osx-64::nspr-4.20-h0a44026_1000
nss conda-forge/osx-64::nss-3.47-hc0980d9_0
odo conda-forge/noarch::odo-0.5.1-py_1
pkginfo conda-forge/noarch::pkginfo-1.5.0.1-py_0
python_abi conda-forge/osx-64::python_abi-3.6-1_cp36m
typing conda-forge/osx-64::typing-3.6.4-py36_0
The following packages will be REMOVED:
backports.os-0.1.1-py36_0
hypothesis-5.5.4-py_0
mkl_fft-1.0.15-py36h5e564d8_0
mkl_random-1.1.0-py36ha771720_0
numpy-base-1.18.1-py36h6575580_1
pytest-arraydiff-0.3-py36h39e3cac_0
pytest-astropy-0.8.0-py_0
pytest-astropy-header-0.1.2-py_0
pytest-doctestplus-0.5.0-py_0
pytest-openfiles-0.4.0-py_0
pytest-remotedata-0.3.2-py36_0
snappy-1.1.7-he62c110_3
tbb-2020.0-h04f5b5a_0
typing_extensions-3.7.4.1-py36_0
The following packages will be UPDATED:
appnope pkgs/main::appnope-0.1.0-py36hf537a9a~ -->
conda-forge::appnope-0.1.0-py36h9f0ad1d_1001
appscript pkgs/main::appscript-1.1.0-py36h1de35~ -->
conda-forge::appscript-1.1.0-py36h37b9a7d_1
argh pkgs/main::argh-0.26.2-py36_0 -->
conda-forge::argh-0.26.2-py36_1001
astroid pkgs/main::astroid-2.3.3-py36_0 -->
conda-forge::astroid-2.3.3-py36_1
astropy pkgs/main::astropy-4.0-py36h1de35cc_0 -->
conda-forge::astropy-4.0-py36h37b9a7d_2
autopep8 pkgs/main::autopep8-1.4.4-py_0 -->
conda-forge::autopep8-1.5-py_0
backports.shutil_~ pkgs/main/osx-64::backports.shutil_ge~ -->
conda-forge/noarch::backports.shutil_get_terminal_size-1.0.0-py_3
beautifulsoup4 pkgs/main::beautifulsoup4-4.8.2-py36_0 -->
conda-forge::beautifulsoup4-4.8.2-py36h9f0ad1d_1
bleach pkgs/main/osx-64::bleach-3.1.0-py36_0 -->
conda-forge/noarch::bleach-3.1.4-pyh9f0ad1d_0
blosc pkgs/main::blosc-1.16.3-hd9629dc_0 -->
conda-forge::blosc-1.17.1-h4a8c4bd_0
bottleneck pkgs/main::bottleneck-1.3.2-py36h776b~ -->
conda-forge::bottleneck-1.3.2-py36h255dfe6_1
bzip2 pkgs/main::bzip2-1.0.8-h1de35cc_0 -->
conda-forge::bzip2-1.0.8-h0b31af3_2
chardet pkgs/main::chardet-3.0.4-py36_1003 -->
conda-forge::chardet-3.0.4-py36h9f0ad1d_1006
cryptography pkgs/main::cryptography-2.8-py36ha12b~ -->
conda-forge::cryptography-2.8-py36hc9d8292_2
cycler pkgs/main/osx-64::cycler-0.10.0-py36h~ -->
conda-forge/noarch::cycler-0.10.0-py_2
cython pkgs/main::cython-0.29.15-py36h0a4402~ -->
conda-forge::cython-0.29.16-py36h0130604_0
docutils pkgs/main::docutils-0.16-py36_0 -->
conda-forge::docutils-0.16-py36h9f0ad1d_1
entrypoints pkgs/main::entrypoints-0.3-py36_0 -->
conda-forge::entrypoints-0.3-py36h9f0ad1d_1001
et_xmlfile pkgs/main/osx-64::et_xmlfile-1.0.1-py~ -->
conda-forge/noarch::et_xmlfile-1.0.1-py_1001
expat pkgs/main::expat-2.2.6-h0a44026_0 -->
conda-forge::expat-2.2.9-h4a8c4bd_2
fastcache pkgs/main::fastcache-1.1.0-py36h1de35~ -->
conda-forge::fastcache-1.1.0-py36h37b9a7d_1
flake8 pkgs/main::flake8-3.7.9-py36_0 -->
conda-forge::flake8-3.7.9-py36h9f0ad1d_1
freetype pkgs/main::freetype-2.9.1-hb4e5f40_0 -->
conda-forge::freetype-2.10.1-h8da9a1a_0
fsspec pkgs/main::fsspec-0.6.3-py_0 -->
conda-forge::fsspec-0.7.1-py_0
future pkgs/main::future-0.18.2-py36_0 -->
conda-forge::future-0.18.2-py36h9f0ad1d_1
gettext pkgs/main::gettext-0.19.8.1-h15daf44_3 -->
conda-forge::gettext-0.19.8.1-h46ab8bc_1002
gmp pkgs/main::gmp-6.1.2-hb37e062_1 -->
conda-forge::gmp-6.2.0-h4a8c4bd_2
gmpy2 pkgs/main::gmpy2-2.0.8-py36h6ef4df4_2 -->
conda-forge::gmpy2-2.1.0b1-py36h4160ff4_0
greenlet pkgs/main::greenlet-0.4.15-py36h1de35~ -->
conda-forge::greenlet-0.4.15-py36h37b9a7d_1
h5py pkgs/main::h5py-2.10.0-py36h3134771_0 -->
conda-forge::h5py-2.10.0-nompi_py36h106b333_102
hdf5 pkgs/main::hdf5-1.10.4-hfa1e0ec_0 -->
conda-forge::hdf5-1.10.5-nompi_h3e39495_1104
icu pkgs/main::icu-58.2-h4b95b61_1 -->
conda-forge::icu-64.2-h6de7cb9_1
importlib_metadata pkgs/main/osx-64::importlib_metadata-~ -->
conda-forge/noarch::importlib_metadata-1.6.0-0
ipykernel pkgs/main::ipykernel-5.1.4-py36h39e3c~ -->
conda-forge::ipykernel-5.2.0-py36h95af2a2_1
ipython pkgs/main::ipython-7.13.0-py36h5ca1d4~ -->
conda-forge::ipython-7.13.0-py36h9f0ad1d_2
ipython_genutils pkgs/main/osx-64::ipython_genutils-0.~ -->
conda-forge/noarch::ipython_genutils-0.2.0-py_1
isort pkgs/main::isort-4.3.21-py36_0 -->
conda-forge::isort-4.3.21-py36h9f0ad1d_1
jbig pkgs/main::jbig-2.1-h4d881f8_0 -->
conda-forge::jbig-2.1-h1de35cc_2001
jpeg pkgs/main::jpeg-9b-he5867d9_2 -->
conda-forge::jpeg-9c-h1de35cc_1001
jsonschema pkgs/main::jsonschema-3.2.0-py36_0 -->
conda-forge::jsonschema-3.2.0-py36h9f0ad1d_1
jupyter_console pkgs/main::jupyter_console-6.1.0-py_0 -->
conda-forge::jupyter_console-6.1.0-py_1
jupyter_core pkgs/main::jupyter_core-4.6.3-py36_0 -->
conda-forge::jupyter_core-4.6.3-py36h9f0ad1d_1
jupyterlab pkgs/main::jupyterlab-1.2.6-pyhf63ae9~ -->
conda-forge::jupyterlab-2.0.1-py_0
keyring pkgs/main::keyring-21.1.0-py36_0 -->
conda-forge::keyring-21.1.1-py36h9f0ad1d_2
kiwisolver pkgs/main::kiwisolver-1.1.0-py36h0a44~ -->
conda-forge::kiwisolver-1.1.0-py36h863e41a_1
lazy-object-proxy pkgs/main::lazy-object-proxy-1.4.3-py~ -->
conda-forge::lazy-object-proxy-1.4.3-py36h37b9a7d_2
libcxx pkgs/main::libcxx-4.0.1-hcfea43d_1 -->
conda-forge::libcxx-9.0.1-1
libffi pkgs/main::libffi-3.2.1-h475c297_4 -->
conda-forge::libffi-3.2.1-h4a8c4bd_1007
libgfortran pkgs/main::libgfortran-3.0.1-h93005f0~ -->
conda-forge::libgfortran-4.0.0-2
libiconv pkgs/main::libiconv-1.15-hdd342a3_7 -->
conda-forge::libiconv-1.15-h0b31af3_1006
libpng pkgs/main::libpng-1.6.37-ha441bb4_0 -->
conda-forge::libpng-1.6.37-hbbe82c9_1
libsodium pkgs/main::libsodium-1.0.16-h3efe00b_0 -->
conda-forge::libsodium-1.0.17-h01d97ff_0
libspatialindex pkgs/main::libspatialindex-1.9.3-h0a4~ -->
conda-forge::libspatialindex-1.9.3-h4a8c4bd_3
libtiff pkgs/main::libtiff-4.1.0-hcb84e12_0 -->
conda-forge::libtiff-4.1.0-h2ae36a8_6
libxml2 pkgs/main::libxml2-2.9.9-hf6e021a_1 -->
conda-forge::libxml2-2.9.10-h53d96d6_0
llvm-openmp pkgs/main::llvm-openmp-4.0.1-hcfea43d~ -->
conda-forge::llvm-openmp-9.0.1-h28b9765_2
llvmlite pkgs/main::llvmlite-0.31.0-py36h13419~ -->
conda-forge::llvmlite-0.31.0-py36hde82470_1
locket pkgs/main/osx-64::locket-0.2.0-py36hc~ -->
conda-forge/noarch::locket-0.2.0-py_2
lxml pkgs/main::lxml-4.5.0-py36hef8c89e_0 -->
conda-forge::lxml-4.5.0-py36h2ab0afd_1
lz4-c pkgs/main::lz4-c-1.8.1.2-h1de35cc_0 -->
conda-forge::lz4-c-1.8.3-h6de7cb9_1001
lzo pkgs/main::lzo-2.10-h362108e_2 -->
conda-forge::lzo-2.10-h1de35cc_1000
markupsafe pkgs/main::markupsafe-1.1.1-py36h1de3~ -->
conda-forge::markupsafe-1.1.1-py36h37b9a7d_1
matplotlib pkgs/main::matplotlib-3.1.3-py36_0 -->
conda-forge::matplotlib-3.2.1-0
matplotlib-base pkgs/main::matplotlib-base-3.1.3-py36~ -->
conda-forge::matplotlib-base-3.2.1-py36h83d3ec1_0
mistune pkgs/main::mistune-0.8.4-py36h1de35cc~ -->
conda-forge::mistune-0.8.4-py36h0b31af3_1000
mkl pkgs/main::mkl-2019.4-233 -->
conda-forge::mkl-2019.5-281
mpc pkgs/main::mpc-1.1.0-h6ef4df4_1 -->
conda-forge::mpc-1.1.0-h4160ff4_1006
mpfr pkgs/main::mpfr-4.0.1-h3018a27_3 -->
conda-forge::mpfr-4.0.2-h44b798e_0
networkx pkgs/main::networkx-2.4-py_0 -->
conda-forge::networkx-2.4-py_1
ninja pkgs/main::ninja-1.9.0-py36h04f5b5a_0 -->
conda-forge::ninja-1.10.0-ha1b3eb9_0
nose pkgs/main::nose-1.3.7-py36_2 -->
conda-forge::nose-1.3.7-py36h9f0ad1d_1004
numexpr pkgs/main::numexpr-2.7.1-py36hce01a72~ -->
conda-forge::numexpr-2.7.1-py36hcc1bba6_1
numpy pkgs/main::numpy-1.18.1-py36h7241aed_0 -->
conda-forge::numpy-1.18.1-py36hdc5ca10_1
pandoc pkgs/main::pandoc-2.2.3.2-0 -->
conda-forge::pandoc-2.9.2-0
pathlib2 pkgs/main::pathlib2-2.3.5-py36_0 -->
conda-forge::pathlib2-2.3.5-py36h9f0ad1d_1
pcre pkgs/main::pcre-8.43-h0a44026_0 -->
conda-forge::pcre-8.44-h4a8c4bd_0
pexpect pkgs/main::pexpect-4.8.0-py36_0 -->
conda-forge::pexpect-4.8.0-py36h9f0ad1d_1
pickleshare pkgs/main::pickleshare-0.7.5-py36_0 -->
conda-forge::pickleshare-0.7.5-py36h9f0ad1d_1001
pillow pkgs/main::pillow-7.0.0-py36h4655f20_0 -->
conda-forge::pillow-7.0.0-py36h2ae5dfa_1
pip pkgs/main/osx-64::pip-20.0.2-py36_1 -->
conda-forge/noarch::pip-20.0.2-py_2
ply pkgs/main/osx-64::ply-3.11-py36_0 -->
conda-forge/noarch::ply-3.11-py_1
prompt-toolkit pkgs/main::prompt-toolkit-3.0.4-py_0 -->
conda-forge::prompt-toolkit-3.0.5-py_0
prompt_toolkit pkgs/main::prompt_toolkit-3.0.4-0 -->
conda-forge::prompt_toolkit-3.0.5-0
psutil pkgs/main::psutil-5.7.0-py36h1de35cc_0 -->
conda-forge::psutil-5.7.0-py36h37b9a7d_1
ptyprocess pkgs/main/osx-64::ptyprocess-0.6.0-py~ -->
conda-forge/noarch::ptyprocess-0.6.0-py_1001
pycosat pkgs/main::pycosat-0.6.3-py36h1de35cc~ -->
conda-forge::pycosat-0.6.3-py36h37b9a7d_1004
pycrypto pkgs/main::pycrypto-2.6.1-py36h1de35c~ -->
conda-forge::pycrypto-2.6.1-py36h37b9a7d_1004
pydocstyle pkgs/main::pydocstyle-4.0.1-py_0 -->
conda-forge::pydocstyle-5.0.2-py_0
pyopenssl pkgs/main/osx-64::pyopenssl-19.1.0-py~ -->
conda-forge/noarch::pyopenssl-19.1.0-py_1
pyqt pkgs/main::pyqt-5.9.2-py36h655552a_2 -->
conda-forge::pyqt-5.12.3-py36he22c54c_1
pysocks pkgs/main::pysocks-1.7.1-py36_0 -->
conda-forge::pysocks-1.7.1-py36h9f0ad1d_1
pytables pkgs/main::pytables-3.6.1-py36h5bccee~ -->
conda-forge::pytables-3.6.1-py36h6f8395a_1
python pkgs/main::python-3.6.10-hc70fcce_1 -->
conda-forge::python-3.6.10-hce46be0_1009_cpython
pyzmq pkgs/main::pyzmq-18.1.1-py36h0a44026_0 -->
conda-forge::pyzmq-19.0.0-py36h820b253_1
qdarkstyle pkgs/main::qdarkstyle-2.8-py_0 -->
conda-forge::qdarkstyle-2.8.1-pyh9f0ad1d_0
qt pkgs/main::qt-5.9.7-h468cd18_1 -->
conda-forge::qt-5.12.5-h514805e_3
requests pkgs/main/osx-64::requests-2.23.0-py3~ -->
conda-forge/noarch::requests-2.23.0-pyh8c360ce_2
rtree pkgs/main::rtree-0.9.3-py36_0 -->
conda-forge::rtree-0.9.4-py36he053a7a_1
scikit-learn pkgs/main::scikit-learn-0.22.1-py36h2~ -->
conda-forge::scikit-learn-0.22.2.post1-py36h3dc85bc_0
scipy pkgs/main::scipy-1.4.1-py36h9fa6033_0 -->
conda-forge::scipy-1.4.1-py36h1dac7e4_2
seaborn pkgs/main::seaborn-0.10.0-py_0 -->
conda-forge::seaborn-0.10.0-py_1
singledispatch pkgs/main::singledispatch-3.4.0.3-py3~ -->
conda-forge::singledispatch-3.4.0.3-py36_1000
sip pkgs/main::sip-4.19.8-py36h0a44026_0 -->
conda-forge::sip-4.19.20-py36h4a8c4bd_0
six pkgs/main/osx-64::six-1.14.0-py36_0 -->
conda-forge/noarch::six-1.14.0-py_1
statsmodels pkgs/main::statsmodels-0.11.0-py36h1d~ -->
conda-forge::statsmodels-0.11.1-py36h37b9a7d_1
sympy pkgs/main::sympy-1.5.1-py36_0 -->
conda-forge::sympy-1.5.1-py36h9f0ad1d_3
terminado pkgs/main::terminado-0.8.3-py36_0 -->
conda-forge::terminado-0.8.3-py36h9f0ad1d_1
tk pkgs/main::tk-8.6.8-ha441bb4_0 -->
conda-forge::tk-8.6.10-hbbe82c9_0
traitlets pkgs/main::traitlets-4.3.3-py36_0 -->
conda-forge::traitlets-4.3.3-py36h9f0ad1d_1
ujson pkgs/main::ujson-1.35-py36h1de35cc_0 -->
conda-forge::ujson-1.35-py36h0130604_1002
unicodecsv pkgs/main/osx-64::unicodecsv-0.14.1-p~ -->
conda-forge/noarch::unicodecsv-0.14.1-py_1
unixodbc pkgs/main::unixodbc-2.3.7-h1de35cc_0 -->
conda-forge::unixodbc-2.3.7-hea208f4_1000
werkzeug pkgs/main::werkzeug-1.0.0-py_0 -->
conda-forge::werkzeug-1.0.1-pyh9f0ad1d_0
wheel pkgs/main/osx-64::wheel-0.34.2-py36_0 -->
conda-forge/noarch::wheel-0.34.2-py_1
wurlitzer pkgs/main::wurlitzer-2.0.0-py36_0 -->
conda-forge::wurlitzer-2.0.0-py36h9f0ad1d_1
xlwings pkgs/main::xlwings-0.18.0-py36_0 -->
conda-forge::xlwings-0.18.0-py36h9f0ad1d_1
xlwt pkgs/main/osx-64::xlwt-1.2.0-py36h5ad~ -->
conda-forge/noarch::xlwt-1.3.0-py_1
xz pkgs/main::xz-5.2.4-h1de35cc_4 -->
conda-forge::xz-5.2.4-h0b31af3_1002
yaml pkgs/main::yaml-0.1.7-hc338f04_2 -->
conda-forge::yaml-0.2.2-h0b31af3_1
yapf pkgs/main::yapf-0.28.0-py_0 -->
conda-forge::yapf-0.29.0-py_0
zeromq pkgs/main::zeromq-4.3.1-h0a44026_3 -->
conda-forge::zeromq-4.3.2-h6de7cb9_2
zipp pkgs/main::zipp-2.2.0-py_0 -->
conda-forge::zipp-3.1.0-py_0
zlib pkgs/main::zlib-1.2.11-h1de35cc_3 -->
conda-forge::zlib-1.2.11-h0b31af3_1006
zstd pkgs/main::zstd-1.3.7-h5bba6e5_0 -->
conda-forge::zstd-1.4.4-hed8d7c8_2
The following packages will be SUPERSEDED by a higher-priority channel:
alabaster pkgs/main/osx-64::alabaster-0.7.12-py~ -->
conda-forge/noarch::alabaster-0.7.12-py_0
anaconda-client pkgs/main/osx-64::anaconda-client-1.7~ -->
conda-forge/noarch::anaconda-client-1.7.2-py_0
anaconda-project pkgs/main::anaconda-project-0.8.4-py_0 -->
conda-forge::anaconda-project-0.8.3-py_0
applaunchservices pkgs/main --> conda-forge
asn1crypto pkgs/main --> conda-forge
atomicwrites pkgs/main/osx-64::atomicwrites-1.3.0-~ -->
conda-forge/noarch::atomicwrites-1.3.0-py_0
attrs pkgs/main --> conda-forge
babel pkgs/main --> conda-forge
backcall pkgs/main/osx-64::backcall-0.1.0-py36~ -->
conda-forge/noarch::backcall-0.1.0-py_0
backports pkgs/main --> conda-forge
bitarray pkgs/main::bitarray-1.2.1-py36h1de35c~ -->
conda-forge::bitarray-1.2.1-py36h0b31af3_0
bkcharts pkgs/main --> conda-forge
blas pkgs/main --> conda-forge
bokeh pkgs/main::bokeh-2.0.1-py36_0 -->
conda-forge::bokeh-1.4.0-py36h9f0ad1d_1
boto pkgs/main/osx-64::boto-2.49.0-py36_0 -->
conda-forge/noarch::boto-2.49.0-py_0
ca-certificates pkgs/main::ca-certificates-2020.1.1-0 -->
conda-forge::ca-certificates-2019.11.28-hecc5488_0
certifi pkgs/main::certifi-2019.11.28-py36_1 -->
conda-forge::certifi-2019.11.28-py36h9f0ad1d_1
cffi pkgs/main::cffi-1.14.0-py36hb5b8e2f_0 -->
conda-forge::cffi-1.14.0-py36h356ff06_0
click pkgs/main::click-7.1.1-py_0 -->
conda-forge::click-7.1.1-pyh8c360ce_0
cloudpickle pkgs/main --> conda-forge
clyent pkgs/main/osx-64::clyent-1.2.2-py36_1 -->
conda-forge/noarch::clyent-1.2.2-py_1
colorama pkgs/main --> conda-forge
contextlib2 pkgs/main --> conda-forge
curl pkgs/main::curl-7.69.1-ha441bb4_0 -->
conda-forge::curl-7.68.0-h8754def_0
cytoolz pkgs/main::cytoolz-0.10.1-py36h1de35c~ -->
conda-forge::cytoolz-0.10.1-py36h0b31af3_0
dask pkgs/main --> conda-forge
dask-core pkgs/main --> conda-forge
dbus pkgs/main::dbus-1.13.12-h90a0687_0 -->
conda-forge::dbus-1.13.6-h2f22bb5_0
decorator pkgs/main --> conda-forge
defusedxml pkgs/main --> conda-forge
diff-match-patch pkgs/main --> conda-forge
distributed pkgs/main::distributed-2.13.0-py36_0 -->
conda-forge::distributed-2.13.0-py36h9f0ad1d_0
flask pkgs/main --> conda-forge
gevent pkgs/main::gevent-1.4.0-py36h1de35cc_0 -->
conda-forge::gevent-1.4.0-py36h0b31af3_0
glib pkgs/main::glib-2.63.1-hd977a24_0 -->
conda-forge::glib-2.58.3-py36hb0ce7ff_1003
heapdict pkgs/main --> conda-forge
html5lib pkgs/main/osx-64::html5lib-1.0.1-py36~ -->
conda-forge/noarch::html5lib-1.0.1-py_0
idna pkgs/main --> conda-forge
imageio pkgs/main --> conda-forge
imagesize pkgs/main --> conda-forge
intervaltree pkgs/main --> conda-forge
ipywidgets pkgs/main --> conda-forge
itsdangerous pkgs/main/osx-64::itsdangerous-1.1.0-~ -->
conda-forge/noarch::itsdangerous-1.1.0-py_0
jdcal pkgs/main --> conda-forge
jedi pkgs/main --> conda-forge
jinja2 pkgs/main --> conda-forge
joblib pkgs/main --> conda-forge
json5 pkgs/main::json5-0.9.3-py_0 -->
conda-forge::json5-0.9.0-py_0
jupyter pkgs/main/osx-64::jupyter-1.0.0-py36_7 -->
conda-forge/noarch::jupyter-1.0.0-py_2
jupyter_client pkgs/main --> conda-forge
jupyterlab_server pkgs/main --> conda-forge
krb5 pkgs/main::krb5-1.17.1-hddcf347_0 -->
conda-forge::krb5-1.16.4-h1752a42_0
libcurl pkgs/main::libcurl-7.69.1-h051b688_0 -->
conda-forge::libcurl-7.68.0-h709d2b2_0
libcxxabi pkgs/main --> conda-forge
libedit pkgs/main::libedit-3.1.20181209-hb402~ -->
conda-forge::libedit-3.1.20170329-hcfe32e1_1001
libssh2 pkgs/main::libssh2-1.9.0-ha12b0ac_1 -->
conda-forge::libssh2-1.8.2-hcdc9a53_2
libxslt pkgs/main::libxslt-1.1.33-h33a18ac_0 -->
conda-forge::libxslt-1.1.33-h320ff13_0
mccabe pkgs/main/osx-64::mccabe-0.6.1-py36_1 -->
conda-forge/noarch::mccabe-0.6.1-py_1
mkl-service pkgs/main::mkl-service-2.3.0-py36hfbe~ -->
conda-forge::mkl-service-2.3.0-py36h0b31af3_0
mock pkgs/main/noarch::mock-4.0.1-py_0 -->
conda-forge/osx-64::mock-3.0.5-py36h9f0ad1d_1
more-itertools pkgs/main --
Okay that clears up one issue:
pytorch-1.4.0 |cpu_py36hf9bb1df_0 26.3 MB
looks like a bug in the logging output of conda. That package is the one from the pytorch channel that's already installed. Looking further down in the output at what will be INSTALLED, REMOVED and UPDATED will confirm that the pytorch package doesn't change. So there's no pytorch from conda-forge.
Then, you don't give the output from conda install pytorch torchvision -c pytorch, right now that gives:
The following packages will be downloaded:
package | build
---------------------------|-----------------
ninja-1.9.0 | py36h04f5b5a_0 90 KB
pytorch-1.4.0 | py3.6_0 34.5 MB pytorch
torchvision-0.5.0 | py36_cpu 5.8 MB pytorch
------------------------------------------------------------
Total: 40.4 MB
The following NEW packages will be INSTALLED:
ninja pkgs/main/osx-64::ninja-1.9.0-py36h04f5b5a_0
pytorch pytorch/osx-64::pytorch-1.4.0-py3.6_0
torchvision pytorch/osx-64::torchvision-0.5.0-py36_cpu
for me. osx-64/pytorch-1.4.0-py3.6_0.tar.bz2 is the most recent version on https://anaconda.org/pytorch/pytorch/files, and it doesn't crash on import torch for me.
Side note: that changing from defaults to conda-forge channel order is something you never want to do, it's a recipe for trouble. Always create a new env instead.
rgommers
on 1 Apr 2020
Related issues
dzhulgakov
·
68Comments
hameerabbasi
·
74Comments
ChocolateDave
·
75Comments
GinSoda
·
61Comments
EMarquer
·
91Comments
Most helpful comment
From the valgrind output, the error occurs in
_GLOBAL__sub_I_AVX2.cppwhich initialises the global variables forTH/vector/AVX2.cpp. Since that file doesn't declare any globals itself, I suspect it's caused by a header that is being included due to 8ffcbfb7d45579c4761cd8a8aafcd82218efb4ab.One candidate could be
ATen/Dimname.hwhich is now included via the following include chain:TH/vector/AVX2.cpp->ATen/Context.h->ATen/core/Tensor.h->ATen/NamedTensor.h->ATen/Dimname.hIn there we see one global variable being initialised:
https://github.com/pytorch/pytorch/blob/8ffcbfb7d45579c4761cd8a8aafcd82218efb4ab/aten/src/ATen/Dimname.h#L30
@elbamos could you try running from https://github.com/peterbell10/pytorch/commit/1a2b2de79c22a091c07147ecd6105192d94980de to see if it changes anything?