-
Sometimes, debugging the kernel code can be extremely painful, only to discover that the issue was caused by a missing `DUP(i)` or an incorrect `SWAP(i)`. These errors often result in evaluating a fun…
-
DaphneIR has a `GroupOp` for relational-style grouping and aggregation, plus a corresponding `group`-kernel. However, at the moment, this operation can only be created through DAPHNE's SQL parser, but…
-
The thought is the following: let the compiler to the bare minimum work of bootstrapping the alpaca primitives and implement as much of the rest in alpaca itself. So everything that is in the compiler…
-
Purpose of this task is to provide some guidance and samples for developers on how control which kernel functions get advertised to the LLM
Some options
1. Use the tool behaviour which allows th…
-
A use-case I've come across a number of times now is wanting to access the results of a fragment from within a kernel.
This often comes up in the context of `AggregateExpFragment` where one aggrega…
-
ir193 updated
3 years ago
-
Following the merging of #3143 the `*spacings` functions return a ` KernelFunctionOperator`.
This is great, but should we be worried that this is inconsistent with the `*nodes` functions which retu…
-
The CUDA kernel doesn't seem to respect the NUMBA_OPT environment variable. This makes switching debugging on and off from shell difficult, especially for kernels that make many device function calls.…
-
**Describe the bug**
Not exactly sure if this is a bug, but at least a gap. I have moved away from using the Planners and rely on "vanilla" function calling. Yet my tools that I defined can be chaine…
-
**Your question**
Ask a clear and concise question about Flux.
I'm puzzled by how flux handles the problem of computation and communication competing for hardware resources when they overlap.
…