-
_Sorry if there exist already an issue, but I was not able to find it._
**Description**
We have seen recently that we have many issues with the Atomix based internal networking/communication l…
-
**Description**
Currently we have a feature to promote or demote a Raft member, which means you can put a member from ACTIVE to PASSIVE or vice versa. This includes an configuration change, since i…
-
**Description**
Currently we have multiple approaches how to run things. Mostly in our components we use our ActorScheduler. In the atomix module the ExecutorServices are widely used. This brings s…
-
Sampling from `LKJ ` or `LKJCholesky` and assigning to a vector of matrices errors, regardless of AD backend.
The code compiles but errors with a `BoundsError`. Sampling from the Prior works does n…
-
```julia
julia> @time_imports using SciMLSensitivity
293.5 ms Preferences
0.5 ms PrecompileTools
0.3 ms SuiteSparse
0.4 ms Requires
2.2 ms ArrayInterface
…
-
**Summary**
> Try to answer the following as best as possible
- How often does the test fail? At least once. It didn't finish
- Does it block your work? Nah, but it did fail a `Zeebe CI` run on…
-
At the minute the GPU/differentiable path is Zygote-compatible and hence uses non-mutating broadcasted operations. This works, but is rather slow and very GPU memory-intensive.
Long term the plan i…
-
**Describe the bug 🐞**
Consider the example at [doc](https://docs.sciml.ai/SciMLSensitivity/stable/getting_started/#Forward-Mode-Automatic-Differentiation-with-ForwardDiff.jl), but one wishes to ta…
-
I tried to create a small benchmark to see if `DaggerGPU.jl` can help me asynchronously move data to and from the GPU.
However, it seems that the memory on the CUDA device is not free'd properly, and …
-
As discussed on slack, I am sharing an MWE to reproduce the issues with some backends on the GPU that I hit. I believe that most of these failures are very representative of the failures that I have s…