-
Currently, we use FIFO to pack blocks/chunks. This is nowhere near an optimal strategy to pack blocks/chunks (especially with #717).
We should devise some algorithm that is better at packing (maybe…
-
This is a pretty neat library, but I have one minor qualm about it. It would be nice if, in the case where the pack function could not find a solution, it threw a specific type of exception instead of…
-
In order to support sub-byte dtypes for quantization, I (and many others) believe that it is better to pack these smaller dtypes into existing pytorch dtypes in order to reduce memory bandwidth conten…
-
Issue https://github.com/NVIDIA/cccl/issues/220 describes a tiles state that uses bit-packing to combine a tile's message type with its status flag. The idea here is to make use of this idea for bit-p…
-
![circlepack2](https://cloud.githubusercontent.com/assets/737638/14762936/664641da-09b9-11e6-86c8-4b136f21961a.gif)
dhowe updated
2 months ago
-
The algorithm (bottom up quad tree packing) is only good if all input textures are power-of-2 squares. Use a general bin packing algorithm, e.g. http://www.csc.liv.ac.uk/~epa/surveyhtml.html
crobi updated
12 years ago
-
Investigate opportunities and costs of lossy packing and lossless compression of data.
* Packing: reducing the precision of the data to quantity-specific levels of precision
* lossy - by defin…
-
The current line packing algorithm lets you lay out a series of velements with arbitrarily complex layout problems. Simplify this by having three categories of boxes: rigid, elastic, and fill. The fol…
-
We do not want to tightly pack as we currently are, since it doesn't make sense.
Instead, boxes (container) must be diagonally stacked.
-
Although the spec doesn't necessarily need to prescribe a sector-packing algorithm, it is important that we know a viable one exists before we commit to requiring a maximum (even zero) allowable paddi…