| e86fc00d097cb8cc118435ebd464497825bcad1e |
|
18-Jan-2018 |
Yuefeng Zhou <yuefengz@google.com> |
Combine host and device memory proto fields.
PiperOrigin-RevId: 182284426
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 4241b86dc8da0f8ba23cb832c090469635bf09a9 |
|
12-Oct-2017 |
Benoit Steiner <bsteiner@google.com> |
Updated the virtual scheduler to use legal names when inserting Send/Recv nodes in the graph.
PiperOrigin-RevId: 171986401
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| d08cb107e6eeedd74c44f0d3654753b141cfa645 |
|
10-Oct-2017 |
A. Unique TensorFlower <gardener@tensorflow.org> |
Scheduler exports tensor size info to RunMetadata. In addition, tensor size histogram
is printed out optionally (use vmodule=analytical_cost_estimator=1 or 2).
PiperOrigin-RevId: 171619454
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 24c5f7784889a5ecf1b3fe42ebb08709590abafd |
|
25-Aug-2017 |
Yuefeng Zhou <yuefengz@google.com> |
Fixed the ExtractExtraProperties function: 1) skip control input 2) log when input properties don't match the graph.
PiperOrigin-RevId: 166414766
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 513def0bb27e4a7c29f6ff533d8ca150b2ab78b4 |
|
18-Aug-2017 |
Benoit Steiner <bsteiner@google.com> |
Fixed BuildOpInfoWithoutDevice
PiperOrigin-RevId: 165653933
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| e85d3df92deb9d717befdf173966a2913ac2aea0 |
|
29-Jun-2017 |
Geoffrey Irving <geoffreyi@google.com> |
Prepare to remove a bunch of proto.h includes from tensorflow/core headers
The goal is to make kernels mostly independent of proto headers, which will let
us lock down our .so imports. This CL does not remove any actual headers, but
changes a bunch of files so that header removal is possible in a followup CL.
It also marks the headers that will be removed with
// TODO(b/62899350): Remove
RELNOTES: n/a
PiperOrigin-RevId: 160552878
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 0cc851c08f9ff4c084c98e28e72d3f99a78d161f |
|
06-Jun-2017 |
Yuefeng Zhou <yuefengz@google.com> |
Call maxcut algorithm in the model_based_cost_estimator.
PiperOrigin-RevId: 158078511
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| e5088cb823964216adfba3155965e0f6f2c7bf7c |
|
01-Jun-2017 |
Yao Zhang <yaozhang@google.com> |
Fix discrepancy between measured and analytical cost graph. Use tf_cuda_library for utils.
PiperOrigin-RevId: 157660745
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 2251633a509d6807d309bff60013c86650bb891d |
|
26-May-2017 |
Benoit Steiner <bsteiner@google.com> |
Open sourced the cost analyzer
PiperOrigin-RevId: 157178951
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 4a09e96797fdcf55b308fade4fd719ef77497d0d |
|
25-May-2017 |
A. Unique TensorFlower <gardener@tensorflow.org> |
Print out detailed node counts, shapes and costs. This is useful for a couple of reasons:
1) Shapes are useful for debugging whether the grappler operations match the expected operations. For example, checking that the batch sizes propagate to the individual matrix multiply.
2) The counts are useful for matching the number of individual grappler ops against the actual op counts from a timeline profile. This will help in debugging the cost that grappler outputs.
3) The individual per-op timings give a more detailed breakdown of how grappler is estimating the run time.
PiperOrigin-RevId: 157062681
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 337748d9800717c50195f05e5e1027426b965408 |
|
20-May-2017 |
Benoit Steiner <bsteiner@google.com> |
Properly handle nodes that have control dependencies as input: in this case,
the number of node inputs is greater than the number of inputs on the
corresponding OpDef.
Avoid asserting on unexpected inputs.
PiperOrigin-RevId: 156613037
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 13b552e332698ed6282aad3b702c52cc619a8c37 |
|
17-May-2017 |
Yuefeng Zhou <yuefengz@google.com> |
Add filesize as a node attribute if filename exists as its input.
PiperOrigin-RevId: 156269550
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 1937df84a17f8f520ab2d45d083d65188afd9c62 |
|
09-May-2017 |
Benoit Steiner <bsteiner@google.com> |
Propagate the cluster to the grappler optimizers: this will enable the development of smarter optimizers that can take the available hardware into account when optimizing a graph.
PiperOrigin-RevId: 155520653
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 87ba9f5370c0f7068760f9536979d9183f6dfe9c |
|
06-May-2017 |
Benoit Steiner <bsteiner@google.com> |
Return more complete device information from the GetDevices() method of grappler clusters.
Change: 155264843
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| fe16da297ceb2cd71e1c7128116e0339ec6789ed |
|
06-May-2017 |
Benoit Steiner <bsteiner@google.com> |
Use the proper units to encode frequencies
Change: 155247835
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 9b336b4a33158061535fd6ba4973605248055b69 |
|
04-May-2017 |
Benoit Steiner <bsteiner@google.com> |
Open sourced op level cost prediction
Change: 155123817
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| a88053349afdd85f9d0efca7917fdb8eedbcc647 |
|
14-Apr-2017 |
Yuefeng Zhou <yuefengz@google.com> |
Fold the content a const op into its consumer node.
Change: 153204827
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 406199e2835dd1ed6c1f8406de081accb0c12b75 |
|
01-Apr-2017 |
Yuefeng Zhou <yuefengz@google.com> |
Set cpu family, model and frequency in the op info.
Change: 151923693
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|
| 96cb8f886ad84202e363c5a9da56cdbce4eaf408 |
|
08-Mar-2017 |
Benoit Steiner <bsteiner@google.com> |
Started to open source Grappler. First application is the GPU layout optimizer.
Change: 149558284
/external/tensorflow/tensorflow/core/grappler/costs/utils.cc
|