Repository navigation
[SCHEDULE] New Reduction Mode for Tensorize #714
Description
So far we are implementing reduction as
init(C)
for rx in reduce_range_x:
for ry in reduce_range_y:
update(C, reduce_value)This may not be desirable, especially when there is no reset(init) intrinsic available. So alternatively, we can do
for rx in reduce_range_x:
for ry in reduce_range_y:
if likely(rx == 0) and likely(ry == 0):
compute(C, reduce_value)
else:
update(C, reduce_value)The subsequent loop split process will peel the loop to remove the likely condition when possible.
For now, we can enable this mode, when user did not provide init intrinsic in tensor_intrin, but provided compute and update intrinsic
Activity
tqchen commentedon Dec 19, 2017
cc @derisavi-huawei @merrymercy let me know if it makes sense to you and see if any of you are interested in this
derisavi-huawei commentedon Dec 19, 2017
I'm not crystal clear on what the problem is. It seems to me that an init intrinsic should always be available because a reduction always has an identity element. Is this about an alternative way for TVM to generate IR for
reduce_axis(so that it eventually leads to more efficient code on a particular hardware), or is it about what kind of patterns (init/update vs compute/update) tensorize can match, or both, or something else? That was a lot of ORs :-) Can you please elaborate?Also, I don't clearly see the pros and cons of the design decisions involved here. Could you give us initial ideas so we can brainstorm?
In terms of interest, it's not something that will be on our radar in the next month or so. If we grow an interest in it, I'll definitely let you know.
Sorry! You were expecting answers and you received many questions instead :-)
tqchen commentedon Dec 20, 2017
It is related to #512 on comment in terms of the alternative way of generating reduction, and is useful when init intrinsic is not available due too limitation of hardware
merrymercy commentedon Dec 21, 2017
I can try this but may not too soon.
derisavi-huawei commentedon Dec 21, 2017
Second form makes life easier for us, so we are interested. Unfortunately we can't invest time in this in the short term.
taknevski commentedon Dec 22, 2017
Hi,
Let me know if this is off-topic..
I was wondering if the reduction initialization could be scheduled separately much in the same way as the tvm.scan init and update operations. Currently I want to move the zero initialization for the output tensor in convolution to a separate loop nest(at the outermost level) but I can't seem to find a way to do it easily.
kun-zh commentedon Dec 23, 2017
@tqchen , I have a solution in my local now, as we discussed, firstly set the reset part as "None" in intrinsic definition, then transform the update part in tensorize.cc. I will send to u for review after some tests.
kun-zh commentedon Dec 26, 2017
@tqchen , I created a pull request, can u help to review it?
tqchen commentedon Dec 27, 2017
#727
tqchen commentedon Dec 29, 2017
close this, thanks to @kun-zh