-
Notifications
You must be signed in to change notification settings - Fork 82
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[Wait For #2854][FSU] Modify SimpleFC Application for FSU-CPU Test @open sesame 02/10 14:56 #2906
Conversation
ec4e82c
to
d9d1cf6
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
d9d1cf6
to
5074334
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. Nice work!
3f8c249
to
b5324b3
Compare
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Minor comments
Applications/SimpleFC/jni/main.cpp
Outdated
|
||
for (unsigned int j = 0; j < feature_size; ++j) | ||
input[j] = j; | ||
input[j] = (j / feature_size); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This is just the same as
input[j] = 0;
because this is the integer division and all j
s are smaller than feature_size
.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Oh.. i see i will update, Thank you for Review
if (!((model_graph.getNumLoadedWeightPoolTensors() + 1) / 2 < | ||
lookahead + 1)) { | ||
model_graph.checkUnloadComplete(f - 1); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Just for checking:
This layer waits for the previous layer only if
!((model_graph.getNumLoadedWeightPoolTensors() + 1) / 2 < lookahead + 1)
It doesn't need to wait if
(model_graph.getNumLoadedWeightPoolTensors() + 1) / 2 < lookahead + 1)
Am I correct?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
yes. this will wait unload
tensor for check mem dealloc
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
🫶
ddbf5a0
to
22aa3ea
Compare
Can find out the number of currently loaded tensors through the getNumLoadedTensors function. When LoadTensors is executed once, getNumLoadedTensors increases twice. Added the leave_lookahead argument to LoadTensors. Calculate leave_lookahead through getNumLoadedTensors and execute loadTensors. Signed-off-by: SeoHyungjun <[email protected]>
Update FSU SimpleFC Application for real case **Self evaluation:** 1. Build test: [X]Passed [ ]Failed [ ]Skipped 2. Run test: [X]Passed [ ]Failed [ ]Skipped Signed-off-by: Donghak PARK <[email protected]>
at case of fsu, layer's load not needed - add swap parm - when swap enabled : load not work - update SimpleFC Application **Self evaluation:** 1. Build test: [X]Passed [ ]Failed [ ]Skipped 2. Run test: [X]Passed [ ]Failed [ ]Skipped Signed-off-by: Donghak PARK <[email protected]>
Modify Application for CPU side Test **Self evaluation:** 1. Build test: [X]Passed [ ]Failed [ ]Skipped 2. Run test: [X]Passed [ ]Failed [ ]Skipped Signed-off-by: Donghak PARK <[email protected]>
1. fix uint -> unsigned int 2. remove some debug cout for accurate performance record **Self evaluation:** 1. Build test: [X]Passed [ ]Failed [ ]Skipped 2. Run test: [X]Passed [ ]Failed [ ]Skipped Signed-off-by: Donghak PARK <[email protected]>
Add doxygen comment on getNumLoadedTensors **Self evaluation:** 1. Build test: [X]Passed [ ]Failed [ ]Skipped 2. Run test: [X]Passed [ ]Failed [ ]Skipped Signed-off-by: Donghak PARK <[email protected]>
22aa3ea
to
97a7dcd
Compare
Modiy Simple FC Application for FSU-CPU Side TEST
commit summary
commit 1 [Application] Update FSU SimpleFC Application : update SimpleFC Application for more easy test & actual environment
commit 2 [FSU] update layer weight load logic at fsu : update layernode not to load weight at load()
commit 3 [FSU] Modify Application for CPU side Test : update Application
Self evaluation:
Build test: [X]Passed [ ]Failed [ ]Skipped
Run test: [X]Passed [ ]Failed [ ]Skipped
Signed-off-by: Donghak PARK [email protected]