This repository was archived by the owner on Nov 17, 2023. It is now read-only.

CudnnFind() usage improvements (v1.3.x) - #13123

Merged

szha merged 14 commits into

apache:v1.3.xfrom

lebeg:v1.3.x_cudnnfind_usage_improvements

Nov 7, 2018

Commits on Nov 5, 2018

Add mx.context.gpu_memory_info() to python api for flexible tests.

DickJC123
authored and
lebeg
committed
003aeda
Add test_gluon_gpu.py:test_large_models to show cudnnFind headroom issue.

DickJC123
authored and
lebeg
committed
6962199
Output model sizes tried by test_gluon_gpu.py:test_large_models.

DickJC123
authored and
lebeg
committed
8af7d0e
Fix perl interface to MXGetGPUMemoryInformation.

DickJC123
authored and
lebeg
committed
f08d585
Increase difficulty of test_gluon_gpu.py:test_large_models.

DickJC123
authored and
lebeg
committed
d561e68
Forgot a file in fix for perl.

DickJC123
authored and
lebeg
committed
b1b25a2
Modify test to pass on no-cudnn CI runner.

DickJC123
authored and
lebeg
committed
e0892d2
Mutex algo reg updates, serialize cudnnFind calls.

DickJC123
authored and
lebeg
committed
0fcc3b2
Fix for cudnnFind memory headroom issue.

DickJC123
authored and
lebeg
committed
c3becef
Fix cpplint.

DickJC123
authored and
lebeg
committed
2dd37b3
Respond to reviewers comments.

DickJC123
authored and
lebeg
committed
56c20ed
Guard against improper MXNET_GPU_MEM_LARGE_ALLOC_ROUND_SIZE values.

DickJC123
authored and
lebeg
committed
e226477
Fix potentially unassigned var.

DickJC123
authored and
lebeg
committed
a1719c7

Commits on Nov 7, 2018

Merge branch 'v1.3.x' into v1.3.x_cudnnfind_usage_improvements
szha
authored
434e453