Pradeep Garigipati
Pradeep Garigipati
@umar456 Has raised a bug report on nvidia wesbite, we haven't heard from them yet. @umar456 tested on a the following devices where tests failed to preserve input values. -...
I think it is a bad idea to pollute current class.
This is not a bug @jansende , axes aren't supposed to move as per current design. If that is a more desired way to have these plots, it can be...
I see your point. The design stayed from an old pipeline I think. Frankly, I don't see use case for panning plots at all. Perhaps panning should be disabled completely...
Auto correction of labels is something that can be definitely be added, although I am not sure how much of an dev effort that would be. Although, I think some...
@lfdmn Can you please rerun the test with `export AF_PRINT_ERRORS=1`, it doesn't look it is value errors - something else is going on - once canny failed, all the other...
@umar456 Could it be launch configuration issue ? I thought those issues where sent into 3.5.1 or they aren't yet.
@umar456 Any ideas on how he can get more info? Do we have TX2? On Fri 1 Dec, 2017, 8:01 PM Umar Arshad, wrote: > I don't think these tests...
@HoneyPatouceul I see those errors on normal CUDA device too. I will look into them, thank you for reporting them.
Interestingly, I am not getting these errors with CUDA 7.5. I wonder what is causing this in CUDA 9. The relevant kernel in the error doesn't seem to have any...