Program generator If/Else flow control - origen-sdk

Question

I'm trying to create a flow on 93k with different actions in the if and else branches. The closest thing I've found so far is to test both passed and failed, although this generates two different compare nodes with empty else blocks:
if_passed :some_test_id do
bin 10
end
if_failed :some_test_id do
bin 11
end
Translates to:
if #SOME_TEST_ID_PASSED == 1 then
{
stop_bin "", "fail", , bad, noreprobe, red, 10, over_on;
}
else
{
}
if #SOME_TEST_ID_FAILED == 1 then
{
stop_bin "", "fail", , bad, noreprobe, red, 11, over_on;
}
else
{
}
{
What I'm looking for is something that looks more like this:
if #SOME_TEST_ID_PASSED == 1 then
{
stop_bin "", "fail", , bad, noreprobe, red, 10, over_on;
}
else
{
stop_bin "", "fail", , bad, noreprobe, red, 11, over_on;
}
Is there a way to generate this type of flow structure?

Answer 1

For the kind of example you show, then this is the recommended way to code it:
test :test1, id: :t1
bin 10, if_passed: :t1
bin 11, if_failed: :t1
That's pretty clear and easy to review at the Origen source level. Indeed that will generate with independent T1_PASSED and T1_FAILED branches, which may not be to your taste, but it is logically correct which is the main thing.
The reason we do it that way is because the internal representation which results in that output on the 93K, is easier to target to Teradyne platforms which are row-based and don't really have the concept of if/else logic. So, you get the benefit that anything you can express in Origen is guaranteed to compile for both platforms, but that may be of little consolation if you only need V93K.
Typically though, the more that you use Origen, then the less you care about the generated implementation, similar really to how you don't really care about the generated assembly when you write embedded code in C.
However, it is better if the generated code is easier to follow when you are debugging on the tester, so we are always striving to make the generated output cleaner.
Having said all that, there is a lower-level API which will get close to what you want.
You could manually implement the flag like this:
test :test1, on_fail: { set_flag: :my_flag }, continue: true
if_flag :my_flag do
bin 10
end
unless_flag :my_flag do
bin 11
end
That will generate something like this:
run_and_branch(test1)
then
{
}
else
{
#MY_FLAG = 1;
}
if #MY_FLAG == 1 then
{
stop_bin "", "fail", , bad, noreprobe, red, 10, over_on;
}
else
{
}
if #MY_FLAG == 1 then
{
}
else
{
stop_bin "", "fail", , bad, noreprobe, red, 10, over_on;
}
That's slightly better (from a desired output perspective) since we're down to one flag, but we still have two adjacent conditionals which could/should be collapsed into one.
We actually already have an issue open for that: https://github.com/Origen-SDK/origen_testers/issues/43
Funnily enough I was looking at that earlier tonight, its not implemented yet, but it will be soon.
At that point the output would be:
run_and_branch(test1)
then
{
}
else
{
#MY_FLAG = 1;
}
if #MY_FLAG == 1 then
{
stop_bin "", "fail", , bad, noreprobe, red, 10, over_on;
}
else
{
stop_bin "", "fail", , bad, noreprobe, red, 11, over_on;
}
I would again say though, that consciously structuring the Origen source just to get a particular output is not really the way it should be done, but in this case the source code is not much more complex by being a bit more explicit about it.