requesting bindings for `TextExtents`
teresi opened this issue · comments
Could the bindings for cudaFont.TextExtents be added?
I see that cudaFont has GetSize and OverlayText exposed, but not TextExtents
https://github.com/dusty-nv/jetson-utils/blob/master/cuda/cudaFont.h
class cudaFont(builtins.object)
| Bitmap font overlay rendering with CUDA
|
| Methods defined here:
|
| GetSize(...)
| Return the size of the font (height in pixels)
|
| OverlayText(...)
| Render the font overlay for a given text string
|
| __init__(self, /, *args, **kwargs)
| Initialize self. See help(type(self)) for accurate signature.
|
I'm trying to debug an issue where I get
[cuda] cudaDeviceSynchronize()
[cuda] an illegal memory access was encountered (error 700) (hex 0x2BC)
[cuda] /jetson-inference/utils/codec/gstEncoder.cpp:717
Exception ignored in: 'cupy.cuda.function.Module.__dealloc__'
I think this happens when I use jetson_utils.cudaFont.OverlayText and yolov8
and I'm looking into creating a minimum working example for that.
return torch.ops.torchvision.nms(boxes, scores, iou_threshold)
File "/usr/local/lib/python3.10/dist-packages/torch/_ops.py", line 854, in __call__
return self_._op(*args, **(kwargs or {}))
RuntimeError: CUDA error: an illegal instruction was encountered
CUDA kernel errors might be asynchronously reported at some other API call, so the stacktrace below might be incorrect.
File "cupy_backends/cuda/api/runtime.pyx", line 144, in cupy_backends.cuda.api.runtime.check_status
cupy_backends.cuda.api.runtime.CUDARuntimeError: cudaErrorIllegalInstruction: an illegal instruction was encountered
I'm wondering if I can prevent the cudaDeviceSynchronize if I don't draw when it get's too close to an edge.
I can rail the top left position, and if I could find the size of the overlay I could skip the overlay if it would go out of bounds on the bottom right position.
I understand that this may not be the fix but I'm looking to experiment.
I'm on jetson_utils 1.0.0
>>> import jetson_utils
>>> jetson_utils.VERSION
'1.0.0'
much appreciated, thank you