Get rid of "parallel" and "oneDNN" parameters in "build_graph"#223
Conversation
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #223 +/- ##
==========================================
+ Coverage 83.84% 83.97% +0.13%
==========================================
Files 44 44
Lines 2655 2665 +10
Branches 1466 1466
==========================================
+ Hits 2226 2238 +12
Misses 219 219
+ Partials 210 208 -2 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
| int main(int argc, char* argv[]) { | ||
| std::string model_name = "alexnet_mnist"; | ||
| bool parallel = false; | ||
| // bool parallel = false; |
There was a problem hiding this comment.
Please, remove all debugging leftovers
| int main(int argc, char* argv[]) { | ||
| std::string model_name = "alexnet_mnist"; | ||
| bool parallel = false; | ||
| // bool parallel = false; |
| // bool parallel = false; | ||
| bool onednn = false; | ||
| for (int i = 1; i < argc; ++i) { | ||
| if (std::string(argv[i]) == "--parallel") { |
There was a problem hiding this comment.
Having --parallel as a runtime parameter still sounds reasonable. Why it was removed?
It we have decided to remove this then it feels like --onednn flag needs to be removed as well
There was a problem hiding this comment.
static analizys complained about an unused parallel variable
There was a problem hiding this comment.
I wonder why onednn is used then. They hold the similar purpose at this point
There was a problem hiding this comment.
onednn we can only set it this way, when Andrey implements his version with parallel, he will most likely return the --parallel processing here, if I understand the question correctly
There was a problem hiding this comment.
Ok, I got it. This explains what I was concerned about
| } | ||
| }; | ||
| } // namespace it_lab_ai | ||
| } // namespace it_lab_ai No newline at end of file |
| Tensor output = make_tensor(vec, sh1); | ||
| InputLayer a1(kNhwc, kNchw, 1, 2); | ||
|
|
||
| // Используем shared_ptr для всех слоев |
3 participants
#218