leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-643385571
Single process build of MXNet master with gcc7 gives the following results:
```
Child user: 5288.372 s
Child sys : 188.645 s
Child wall: 5481.062 s
Child h
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-643016517
Thanks @wkcn. I'll report the same with gcc7. You are using gcc10 right?
This is an automated message from t
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-643014864
Doing a single-threaded build (`ninja -j1`) just costs around 2 GB memory at
maximum.
Child user: 4167.479 s
Child sys : 159.497 s
Child wall: 4327.964 s
Chi
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-642980960
Based on my preliminary test, the memory issue is present in 1.7.x @ciyongch
I measure the overall memory consumption during compilation using linux
control group feature
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-642785535
I disagree. Official MXNet releases are source releases. At this point in
time, there exist 0 compliant binary releases.
It's very important that we don't introduce regres
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-641669164
There seem to be some more issues. In certain build configuration with llvm
7, many of the numpy object files blow up
```
875M
build/CMakeFiles/mxnet.dir/src/op
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-641655747
There are more problems with the numpy ops:
```
875M
build/CMakeFiles/mxnet.dir/src/operator/tensor/broadcast_reduce_norm_value.cc.o
918M
build/CMakeFiles
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-640730423
I think we should consider this a release-critical bug. @woreom @wkcn did
you try if this affects the 1.7 / 1.x branches as well?
cc: @ciyongch
---
leezu commented on issue #18501:
URL:
https://github.com/apache/incubator-mxnet/issues/18501#issuecomment-640286646
Fixing this would be a welcome improvement. Did you investigate if the high
memory consumption is consistent among gcc and clang, as well as still present
on gcc 9 (or 10) /
